^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1) // SPDX-License-Identifier: GPL-2.0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3) * Generic ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5) * Copyright (C) 2008 Steven Rostedt <srostedt@redhat.com>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) #include <linux/trace_events.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) #include <linux/ring_buffer.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9) #include <linux/trace_clock.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) #include <linux/sched/clock.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) #include <linux/trace_seq.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) #include <linux/spinlock.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) #include <linux/irq_work.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) #include <linux/security.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) #include <linux/uaccess.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) #include <linux/hardirq.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) #include <linux/kthread.h> /* for self test */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) #include <linux/module.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) #include <linux/percpu.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) #include <linux/mutex.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) #include <linux/delay.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) #include <linux/slab.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) #include <linux/init.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) #include <linux/hash.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) #include <linux/list.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) #include <linux/cpu.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) #include <linux/oom.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) #include <asm/local.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) static void update_pages_handler(struct work_struct *work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) * The ring buffer header is special. We must manually up keep it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) int ring_buffer_print_entry_header(struct trace_seq *s)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38) trace_seq_puts(s, "# compressed entry header\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) trace_seq_puts(s, "\ttype_len : 5 bits\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) trace_seq_puts(s, "\ttime_delta : 27 bits\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) trace_seq_puts(s, "\tarray : 32 bits\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) trace_seq_putc(s, '\n');
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) trace_seq_printf(s, "\tpadding : type == %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) RINGBUF_TYPE_PADDING);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) trace_seq_printf(s, "\ttime_extend : type == %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46) RINGBUF_TYPE_TIME_EXTEND);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) trace_seq_printf(s, "\ttime_stamp : type == %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) RINGBUF_TYPE_TIME_STAMP);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) trace_seq_printf(s, "\tdata max type_len == %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) RINGBUF_TYPE_DATA_TYPE_LEN_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) return !trace_seq_has_overflowed(s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) * The ring buffer is made up of a list of pages. A separate list of pages is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) * allocated for each CPU. A writer may only write to a buffer that is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) * associated with the CPU it is currently executing on. A reader may read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) * from any per cpu buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) * The reader is special. For each per cpu buffer, the reader has its own
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) * reader page. When a reader has read the entire reader page, this reader
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) * page is swapped with another page in the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) * Now, as long as the writer is off the reader page, the reader can do what
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) * ever it wants with that page. The writer will never write to that page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) * again (as long as it is out of the ring buffer).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) * Here's some silly ASCII art.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) * +------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) * |reader| RING BUFFER
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) * |page |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) * +------+ +---+ +---+ +---+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) * | |-->| |-->| |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) * +---+ +---+ +---+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) * ^ |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) * | |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79) * +---------------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82) * +------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) * |reader| RING BUFFER
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84) * |page |------------------v
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) * +------+ +---+ +---+ +---+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) * | |-->| |-->| |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87) * +---+ +---+ +---+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88) * ^ |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) * | |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) * +---------------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93) * +------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) * |reader| RING BUFFER
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) * |page |------------------v
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96) * +------+ +---+ +---+ +---+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) * ^ | |-->| |-->| |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) * | +---+ +---+ +---+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) * | |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) * | |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) * +------------------------------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) * +------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) * |buffer| RING BUFFER
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) * |page |------------------v
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) * +------+ +---+ +---+ +---+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) * ^ | | | |-->| |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) * | New +---+ +---+ +---+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) * | Reader------^ |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) * | page |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) * +------------------------------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) * After we make this swap, the reader can hand this page off to the splice
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) * code and be done with it. It can even allocate a new page if it needs to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) * and swap that into the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) * We will be using cmpxchg soon to make all this lockless.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) /* Used for individual buffers (after the counter) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) #define RB_BUFFER_OFF (1 << 20)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) #define BUF_PAGE_HDR_SIZE offsetof(struct buffer_data_page, data)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) #define RB_EVNT_HDR_SIZE (offsetof(struct ring_buffer_event, array))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) #define RB_ALIGNMENT 4U
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) #define RB_MAX_SMALL_DATA (RB_ALIGNMENT * RINGBUF_TYPE_DATA_TYPE_LEN_MAX)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) #define RB_EVNT_MIN_SIZE 8U /* two 32bit words */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) #ifndef CONFIG_HAVE_64BIT_ALIGNED_ACCESS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) # define RB_FORCE_8BYTE_ALIGNMENT 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) # define RB_ARCH_ALIGNMENT RB_ALIGNMENT
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) # define RB_FORCE_8BYTE_ALIGNMENT 1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) # define RB_ARCH_ALIGNMENT 8U
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) #define RB_ALIGN_DATA __aligned(RB_ARCH_ALIGNMENT)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) /* define RINGBUF_TYPE_DATA for 'case RINGBUF_TYPE_DATA:' */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) #define RINGBUF_TYPE_DATA 0 ... RINGBUF_TYPE_DATA_TYPE_LEN_MAX
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) enum {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) RB_LEN_TIME_EXTEND = 8,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) RB_LEN_TIME_STAMP = 8,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) #define skip_time_extend(event) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) ((struct ring_buffer_event *)((char *)event + RB_LEN_TIME_EXTEND))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) #define extended_time(event) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) (event->type_len >= RINGBUF_TYPE_TIME_EXTEND)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) static inline int rb_null_event(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) return event->type_len == RINGBUF_TYPE_PADDING && !event->time_delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) static void rb_event_set_padding(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) /* padding has a NULL time_delta */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) event->type_len = RINGBUF_TYPE_PADDING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) event->time_delta = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) static unsigned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) rb_event_data_length(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) unsigned length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) if (event->type_len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) length = event->type_len * RB_ALIGNMENT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) length = event->array[0];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) return length + RB_EVNT_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) * Return the length of the given event. Will return
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) * the length of the time extend if the event is a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) * time extend.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) static inline unsigned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) rb_event_length(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) switch (event->type_len) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) case RINGBUF_TYPE_PADDING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) if (rb_null_event(event))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) /* undefined */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) return event->array[0] + RB_EVNT_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) case RINGBUF_TYPE_TIME_EXTEND:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) return RB_LEN_TIME_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) case RINGBUF_TYPE_TIME_STAMP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) return RB_LEN_TIME_STAMP;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) case RINGBUF_TYPE_DATA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) return rb_event_data_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) WARN_ON_ONCE(1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) /* not hit */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) * Return total length of time extend and data,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) * or just the event length for all other events.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) static inline unsigned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) rb_event_ts_length(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) unsigned len = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) if (extended_time(event)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) /* time extends include the data event after it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) len = RB_LEN_TIME_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) event = skip_time_extend(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) return len + rb_event_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) * ring_buffer_event_length - return the length of the event
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) * @event: the event to get the length of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) * Returns the size of the data load of a data event.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) * If the event is something other than a data event, it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) * returns the size of the event itself. With the exception
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) * of a TIME EXTEND, where it still returns the size of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) * data load of the data event after it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) unsigned ring_buffer_event_length(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) unsigned length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) if (extended_time(event))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) event = skip_time_extend(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) length = rb_event_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) if (event->type_len > RINGBUF_TYPE_DATA_TYPE_LEN_MAX)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) return length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) length -= RB_EVNT_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) if (length > RB_MAX_SMALL_DATA + sizeof(event->array[0]))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) length -= sizeof(event->array[0]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) return length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) EXPORT_SYMBOL_GPL(ring_buffer_event_length);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) /* inline for ring buffer fast paths */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) static __always_inline void *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) rb_event_data(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) if (extended_time(event))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) event = skip_time_extend(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) WARN_ON_ONCE(event->type_len > RINGBUF_TYPE_DATA_TYPE_LEN_MAX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) /* If length is in len field, then array[0] has the data */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) if (event->type_len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) return (void *)&event->array[0];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) /* Otherwise length is in array[0] and array[1] has the data */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) return (void *)&event->array[1];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) * ring_buffer_event_data - return the data of the event
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) * @event: the event to get the data from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) void *ring_buffer_event_data(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) return rb_event_data(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) EXPORT_SYMBOL_GPL(ring_buffer_event_data);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) #define for_each_buffer_cpu(buffer, cpu) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) for_each_cpu(cpu, buffer->cpumask)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) #define for_each_online_buffer_cpu(buffer, cpu) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) for_each_cpu_and(cpu, buffer->cpumask, cpu_online_mask)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) #define TS_SHIFT 27
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) #define TS_MASK ((1ULL << TS_SHIFT) - 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) #define TS_DELTA_TEST (~TS_MASK)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) * ring_buffer_event_time_stamp - return the event's extended timestamp
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) * @event: the event to get the timestamp of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) * Returns the extended timestamp associated with a data event.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) * An extended time_stamp is a 64-bit timestamp represented
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) * internally in a special way that makes the best use of space
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) * contained within a ring buffer event. This function decodes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) * it and maps it to a straight u64 value.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) u64 ring_buffer_event_time_stamp(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) u64 ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303) ts = event->array[0];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304) ts <<= TS_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) ts += event->time_delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) return ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310) /* Flag when events were overwritten */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311) #define RB_MISSED_EVENTS (1 << 31)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312) /* Missed count stored at end */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) #define RB_MISSED_STORED (1 << 30)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) struct buffer_data_page {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) u64 time_stamp; /* page time stamp */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317) local_t commit; /* write committed index */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318) unsigned char data[] RB_ALIGN_DATA; /* data of buffer page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) * Note, the buffer_page list must be first. The buffer pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323) * are allocated in cache lines, which means that each buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) * page will be at the beginning of a cache line, and thus
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) * the least significant bits will be zero. We use this to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326) * add flags in the list struct pointers, to make the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 327) * lockless.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 328) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 329) struct buffer_page {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 330) struct list_head list; /* list of buffer pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 331) local_t write; /* index for next write */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 332) unsigned read; /* index for next read */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 333) local_t entries; /* entries on this page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 334) unsigned long real_end; /* real end of data */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 335) struct buffer_data_page *page; /* Actual data page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 336) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 337)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 338) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 339) * The buffer page counters, write and entries, must be reset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 340) * atomically when crossing page boundaries. To synchronize this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 341) * update, two counters are inserted into the number. One is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 342) * the actual counter for the write position or count on the page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 343) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 344) * The other is a counter of updaters. Before an update happens
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 345) * the update partition of the counter is incremented. This will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 346) * allow the updater to update the counter atomically.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 347) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 348) * The counter is 20 bits, and the state data is 12.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 349) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 350) #define RB_WRITE_MASK 0xfffff
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 351) #define RB_WRITE_INTCNT (1 << 20)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 352)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 353) static void rb_init_page(struct buffer_data_page *bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 354) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 355) local_set(&bpage->commit, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 356) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 357)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 358) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 359) * Also stolen from mm/slob.c. Thanks to Mathieu Desnoyers for pointing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 360) * this issue out.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 361) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 362) static void free_buffer_page(struct buffer_page *bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 363) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 364) free_page((unsigned long)bpage->page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 365) kfree(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 366) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 367)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 368) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 369) * We need to fit the time_stamp delta into 27 bits.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 370) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 371) static inline int test_time_stamp(u64 delta)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 372) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 373) if (delta & TS_DELTA_TEST)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 374) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 375) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 376) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 377)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 378) #define BUF_PAGE_SIZE (PAGE_SIZE - BUF_PAGE_HDR_SIZE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 379)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 380) /* Max payload is BUF_PAGE_SIZE - header (8bytes) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 381) #define BUF_MAX_DATA_SIZE (BUF_PAGE_SIZE - (sizeof(u32) * 2))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 382)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 383) int ring_buffer_print_page_header(struct trace_seq *s)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 384) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 385) struct buffer_data_page field;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 386)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 387) trace_seq_printf(s, "\tfield: u64 timestamp;\t"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 388) "offset:0;\tsize:%u;\tsigned:%u;\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 389) (unsigned int)sizeof(field.time_stamp),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 390) (unsigned int)is_signed_type(u64));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 391)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 392) trace_seq_printf(s, "\tfield: local_t commit;\t"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 393) "offset:%u;\tsize:%u;\tsigned:%u;\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 394) (unsigned int)offsetof(typeof(field), commit),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 395) (unsigned int)sizeof(field.commit),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 396) (unsigned int)is_signed_type(long));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 397)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 398) trace_seq_printf(s, "\tfield: int overwrite;\t"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 399) "offset:%u;\tsize:%u;\tsigned:%u;\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 400) (unsigned int)offsetof(typeof(field), commit),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 401) 1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 402) (unsigned int)is_signed_type(long));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 403)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 404) trace_seq_printf(s, "\tfield: char data;\t"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 405) "offset:%u;\tsize:%u;\tsigned:%u;\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 406) (unsigned int)offsetof(typeof(field), data),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 407) (unsigned int)BUF_PAGE_SIZE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 408) (unsigned int)is_signed_type(char));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 409)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 410) return !trace_seq_has_overflowed(s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 411) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 412)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 413) struct rb_irq_work {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 414) struct irq_work work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 415) wait_queue_head_t waiters;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 416) wait_queue_head_t full_waiters;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 417) bool waiters_pending;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 418) bool full_waiters_pending;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 419) bool wakeup_full;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 420) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 421)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 422) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 423) * Structure to hold event state and handle nested events.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 424) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 425) struct rb_event_info {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 426) u64 ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 427) u64 delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 428) u64 before;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 429) u64 after;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 430) unsigned long length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 431) struct buffer_page *tail_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 432) int add_timestamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 433) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 434)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 435) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 436) * Used for the add_timestamp
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 437) * NONE
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 438) * EXTEND - wants a time extend
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 439) * ABSOLUTE - the buffer requests all events to have absolute time stamps
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 440) * FORCE - force a full time stamp.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 441) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 442) enum {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 443) RB_ADD_STAMP_NONE = 0,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 444) RB_ADD_STAMP_EXTEND = BIT(1),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 445) RB_ADD_STAMP_ABSOLUTE = BIT(2),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 446) RB_ADD_STAMP_FORCE = BIT(3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 447) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 448) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 449) * Used for which event context the event is in.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 450) * TRANSITION = 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 451) * NMI = 1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 452) * IRQ = 2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 453) * SOFTIRQ = 3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 454) * NORMAL = 4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 455) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 456) * See trace_recursive_lock() comment below for more details.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 457) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 458) enum {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 459) RB_CTX_TRANSITION,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 460) RB_CTX_NMI,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 461) RB_CTX_IRQ,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 462) RB_CTX_SOFTIRQ,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 463) RB_CTX_NORMAL,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 464) RB_CTX_MAX
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 465) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 466)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 467) #if BITS_PER_LONG == 32
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 468) #define RB_TIME_32
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 469) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 470)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 471) /* To test on 64 bit machines */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 472) //#define RB_TIME_32
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 473)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 474) #ifdef RB_TIME_32
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 475)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 476) struct rb_time_struct {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 477) local_t cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 478) local_t top;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 479) local_t bottom;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 480) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 481) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 482) #include <asm/local64.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 483) struct rb_time_struct {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 484) local64_t time;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 485) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 486) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 487) typedef struct rb_time_struct rb_time_t;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 488)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 489) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 490) * head_page == tail_page && head == tail then buffer is empty.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 491) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 492) struct ring_buffer_per_cpu {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 493) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 494) atomic_t record_disabled;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 495) atomic_t resize_disabled;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 496) struct trace_buffer *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 497) raw_spinlock_t reader_lock; /* serialize readers */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 498) arch_spinlock_t lock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 499) struct lock_class_key lock_key;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 500) struct buffer_data_page *free_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 501) unsigned long nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 502) unsigned int current_context;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 503) struct list_head *pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 504) struct buffer_page *head_page; /* read from head */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 505) struct buffer_page *tail_page; /* write to tail */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 506) struct buffer_page *commit_page; /* committed pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 507) struct buffer_page *reader_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 508) unsigned long lost_events;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 509) unsigned long last_overrun;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 510) unsigned long nest;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 511) local_t entries_bytes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 512) local_t entries;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 513) local_t overrun;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 514) local_t commit_overrun;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 515) local_t dropped_events;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 516) local_t committing;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 517) local_t commits;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 518) local_t pages_touched;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 519) local_t pages_read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 520) long last_pages_touch;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 521) size_t shortest_full;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 522) unsigned long read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 523) unsigned long read_bytes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 524) rb_time_t write_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 525) rb_time_t before_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 526) u64 read_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 527) /* ring buffer pages to update, > 0 to add, < 0 to remove */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 528) long nr_pages_to_update;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 529) struct list_head new_pages; /* new pages to add */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 530) struct work_struct update_pages_work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 531) struct completion update_done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 532)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 533) struct rb_irq_work irq_work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 534) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 535)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 536) struct trace_buffer {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 537) unsigned flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 538) int cpus;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 539) atomic_t record_disabled;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 540) cpumask_var_t cpumask;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 541)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 542) struct lock_class_key *reader_lock_key;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 543)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 544) struct mutex mutex;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 545)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 546) struct ring_buffer_per_cpu **buffers;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 547)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 548) struct hlist_node node;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 549) u64 (*clock)(void);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 550)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 551) struct rb_irq_work irq_work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 552) bool time_stamp_abs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 553) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 554)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 555) struct ring_buffer_iter {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 556) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 557) unsigned long head;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 558) unsigned long next_event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 559) struct buffer_page *head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 560) struct buffer_page *cache_reader_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 561) unsigned long cache_read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 562) u64 read_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 563) u64 page_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 564) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 565) int missed_events;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 566) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 567)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 568) #ifdef RB_TIME_32
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 569)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 570) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 571) * On 32 bit machines, local64_t is very expensive. As the ring
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 572) * buffer doesn't need all the features of a true 64 bit atomic,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 573) * on 32 bit, it uses these functions (64 still uses local64_t).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 574) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 575) * For the ring buffer, 64 bit required operations for the time is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 576) * the following:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 577) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 578) * - Only need 59 bits (uses 60 to make it even).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 579) * - Reads may fail if it interrupted a modification of the time stamp.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 580) * It will succeed if it did not interrupt another write even if
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 581) * the read itself is interrupted by a write.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 582) * It returns whether it was successful or not.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 583) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 584) * - Writes always succeed and will overwrite other writes and writes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 585) * that were done by events interrupting the current write.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 586) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 587) * - A write followed by a read of the same time stamp will always succeed,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 588) * but may not contain the same value.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 589) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 590) * - A cmpxchg will fail if it interrupted another write or cmpxchg.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 591) * Other than that, it acts like a normal cmpxchg.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 592) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 593) * The 60 bit time stamp is broken up by 30 bits in a top and bottom half
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 594) * (bottom being the least significant 30 bits of the 60 bit time stamp).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 595) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 596) * The two most significant bits of each half holds a 2 bit counter (0-3).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 597) * Each update will increment this counter by one.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 598) * When reading the top and bottom, if the two counter bits match then the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 599) * top and bottom together make a valid 60 bit number.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 600) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 601) #define RB_TIME_SHIFT 30
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 602) #define RB_TIME_VAL_MASK ((1 << RB_TIME_SHIFT) - 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 603)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 604) static inline int rb_time_cnt(unsigned long val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 605) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 606) return (val >> RB_TIME_SHIFT) & 3;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 607) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 608)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 609) static inline u64 rb_time_val(unsigned long top, unsigned long bottom)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 610) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 611) u64 val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 612)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 613) val = top & RB_TIME_VAL_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 614) val <<= RB_TIME_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 615) val |= bottom & RB_TIME_VAL_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 616)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 617) return val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 618) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 619)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 620) static inline bool __rb_time_read(rb_time_t *t, u64 *ret, unsigned long *cnt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 621) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 622) unsigned long top, bottom;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 623) unsigned long c;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 624)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 625) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 626) * If the read is interrupted by a write, then the cnt will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 627) * be different. Loop until both top and bottom have been read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 628) * without interruption.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 629) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 630) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 631) c = local_read(&t->cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 632) top = local_read(&t->top);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 633) bottom = local_read(&t->bottom);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 634) } while (c != local_read(&t->cnt));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 635)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 636) *cnt = rb_time_cnt(top);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 637)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 638) /* If top and bottom counts don't match, this interrupted a write */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 639) if (*cnt != rb_time_cnt(bottom))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 640) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 641)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 642) *ret = rb_time_val(top, bottom);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 643) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 644) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 645)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 646) static bool rb_time_read(rb_time_t *t, u64 *ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 647) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 648) unsigned long cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 649)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 650) return __rb_time_read(t, ret, &cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 651) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 652)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 653) static inline unsigned long rb_time_val_cnt(unsigned long val, unsigned long cnt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 654) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 655) return (val & RB_TIME_VAL_MASK) | ((cnt & 3) << RB_TIME_SHIFT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 656) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 657)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 658) static inline void rb_time_split(u64 val, unsigned long *top, unsigned long *bottom)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 659) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 660) *top = (unsigned long)((val >> RB_TIME_SHIFT) & RB_TIME_VAL_MASK);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 661) *bottom = (unsigned long)(val & RB_TIME_VAL_MASK);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 662) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 663)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 664) static inline void rb_time_val_set(local_t *t, unsigned long val, unsigned long cnt)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 665) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 666) val = rb_time_val_cnt(val, cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 667) local_set(t, val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 668) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 669)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 670) static void rb_time_set(rb_time_t *t, u64 val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 671) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 672) unsigned long cnt, top, bottom;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 673)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 674) rb_time_split(val, &top, &bottom);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 675)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 676) /* Writes always succeed with a valid number even if it gets interrupted. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 677) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 678) cnt = local_inc_return(&t->cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 679) rb_time_val_set(&t->top, top, cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 680) rb_time_val_set(&t->bottom, bottom, cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 681) } while (cnt != local_read(&t->cnt));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 682) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 683)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 684) static inline bool
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 685) rb_time_read_cmpxchg(local_t *l, unsigned long expect, unsigned long set)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 686) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 687) unsigned long ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 688)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 689) ret = local_cmpxchg(l, expect, set);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 690) return ret == expect;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 691) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 692)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 693) static int rb_time_cmpxchg(rb_time_t *t, u64 expect, u64 set)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 694) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 695) unsigned long cnt, top, bottom;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 696) unsigned long cnt2, top2, bottom2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 697) u64 val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 698)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 699) /* The cmpxchg always fails if it interrupted an update */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 700) if (!__rb_time_read(t, &val, &cnt2))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 701) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 702)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 703) if (val != expect)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 704) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 705)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 706) cnt = local_read(&t->cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 707) if ((cnt & 3) != cnt2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 708) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 709)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 710) cnt2 = cnt + 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 711)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 712) rb_time_split(val, &top, &bottom);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 713) top = rb_time_val_cnt(top, cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 714) bottom = rb_time_val_cnt(bottom, cnt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 715)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 716) rb_time_split(set, &top2, &bottom2);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 717) top2 = rb_time_val_cnt(top2, cnt2);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 718) bottom2 = rb_time_val_cnt(bottom2, cnt2);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 719)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 720) if (!rb_time_read_cmpxchg(&t->cnt, cnt, cnt2))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 721) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 722) if (!rb_time_read_cmpxchg(&t->top, top, top2))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 723) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 724) if (!rb_time_read_cmpxchg(&t->bottom, bottom, bottom2))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 725) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 726) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 727) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 728)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 729) #else /* 64 bits */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 730)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 731) /* local64_t always succeeds */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 732)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 733) static inline bool rb_time_read(rb_time_t *t, u64 *ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 734) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 735) *ret = local64_read(&t->time);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 736) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 737) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 738) static void rb_time_set(rb_time_t *t, u64 val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 739) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 740) local64_set(&t->time, val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 741) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 742)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 743) static bool rb_time_cmpxchg(rb_time_t *t, u64 expect, u64 set)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 744) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 745) u64 val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 746) val = local64_cmpxchg(&t->time, expect, set);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 747) return val == expect;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 748) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 749) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 750)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 751) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 752) * ring_buffer_nr_pages - get the number of buffer pages in the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 753) * @buffer: The ring_buffer to get the number of pages from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 754) * @cpu: The cpu of the ring_buffer to get the number of pages from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 755) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 756) * Returns the number of pages used by a per_cpu buffer of the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 757) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 758) size_t ring_buffer_nr_pages(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 759) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 760) return buffer->buffers[cpu]->nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 761) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 762)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 763) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 764) * ring_buffer_nr_pages_dirty - get the number of used pages in the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 765) * @buffer: The ring_buffer to get the number of pages from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 766) * @cpu: The cpu of the ring_buffer to get the number of pages from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 767) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 768) * Returns the number of pages that have content in the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 769) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 770) size_t ring_buffer_nr_dirty_pages(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 771) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 772) size_t read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 773) size_t cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 774)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 775) read = local_read(&buffer->buffers[cpu]->pages_read);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 776) cnt = local_read(&buffer->buffers[cpu]->pages_touched);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 777) /* The reader can read an empty page, but not more than that */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 778) if (cnt < read) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 779) WARN_ON_ONCE(read > cnt + 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 780) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 781) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 782)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 783) return cnt - read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 784) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 785)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 786) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 787) * rb_wake_up_waiters - wake up tasks waiting for ring buffer input
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 788) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 789) * Schedules a delayed work to wake up any task that is blocked on the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 790) * ring buffer waiters queue.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 791) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 792) static void rb_wake_up_waiters(struct irq_work *work)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 793) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 794) struct rb_irq_work *rbwork = container_of(work, struct rb_irq_work, work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 795)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 796) wake_up_all(&rbwork->waiters);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 797) if (rbwork->wakeup_full) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 798) rbwork->wakeup_full = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 799) wake_up_all(&rbwork->full_waiters);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 800) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 801) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 802)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 803) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 804) * ring_buffer_wait - wait for input to the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 805) * @buffer: buffer to wait on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 806) * @cpu: the cpu buffer to wait on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 807) * @full: wait until the percentage of pages are available, if @cpu != RING_BUFFER_ALL_CPUS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 808) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 809) * If @cpu == RING_BUFFER_ALL_CPUS then the task will wake up as soon
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 810) * as data is added to any of the @buffer's cpu buffers. Otherwise
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 811) * it will wait for data to be added to a specific cpu buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 812) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 813) int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 814) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 815) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 816) DEFINE_WAIT(wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 817) struct rb_irq_work *work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 818) int ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 819)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 820) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 821) * Depending on what the caller is waiting for, either any
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 822) * data in any cpu buffer, or a specific buffer, put the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 823) * caller on the appropriate wait queue.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 824) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 825) if (cpu == RING_BUFFER_ALL_CPUS) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 826) work = &buffer->irq_work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 827) /* Full only makes sense on per cpu reads */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 828) full = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 829) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 830) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 831) return -ENODEV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 832) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 833) work = &cpu_buffer->irq_work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 834) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 835)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 836)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 837) while (true) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 838) if (full)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 839) prepare_to_wait(&work->full_waiters, &wait, TASK_INTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 840) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 841) prepare_to_wait(&work->waiters, &wait, TASK_INTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 842)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 843) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 844) * The events can happen in critical sections where
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 845) * checking a work queue can cause deadlocks.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 846) * After adding a task to the queue, this flag is set
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 847) * only to notify events to try to wake up the queue
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 848) * using irq_work.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 849) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 850) * We don't clear it even if the buffer is no longer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 851) * empty. The flag only causes the next event to run
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 852) * irq_work to do the work queue wake up. The worse
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 853) * that can happen if we race with !trace_empty() is that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 854) * an event will cause an irq_work to try to wake up
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 855) * an empty queue.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 856) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 857) * There's no reason to protect this flag either, as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 858) * the work queue and irq_work logic will do the necessary
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 859) * synchronization for the wake ups. The only thing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 860) * that is necessary is that the wake up happens after
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 861) * a task has been queued. It's OK for spurious wake ups.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 862) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 863) if (full)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 864) work->full_waiters_pending = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 865) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 866) work->waiters_pending = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 867)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 868) if (signal_pending(current)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 869) ret = -EINTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 870) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 871) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 872)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 873) if (cpu == RING_BUFFER_ALL_CPUS && !ring_buffer_empty(buffer))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 874) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 875)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 876) if (cpu != RING_BUFFER_ALL_CPUS &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 877) !ring_buffer_empty_cpu(buffer, cpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 878) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 879) bool pagebusy;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 880) size_t nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 881) size_t dirty;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 882)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 883) if (!full)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 884) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 885)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 886) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 887) pagebusy = cpu_buffer->reader_page == cpu_buffer->commit_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 888) nr_pages = cpu_buffer->nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 889) dirty = ring_buffer_nr_dirty_pages(buffer, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 890) if (!cpu_buffer->shortest_full ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 891) cpu_buffer->shortest_full < full)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 892) cpu_buffer->shortest_full = full;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 893) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 894) if (!pagebusy &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 895) (!nr_pages || (dirty * 100) > full * nr_pages))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 896) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 897) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 898)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 899) schedule();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 900) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 901)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 902) if (full)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 903) finish_wait(&work->full_waiters, &wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 904) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 905) finish_wait(&work->waiters, &wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 906)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 907) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 908) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 909)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 910) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 911) * ring_buffer_poll_wait - poll on buffer input
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 912) * @buffer: buffer to wait on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 913) * @cpu: the cpu buffer to wait on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 914) * @filp: the file descriptor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 915) * @poll_table: The poll descriptor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 916) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 917) * If @cpu == RING_BUFFER_ALL_CPUS then the task will wake up as soon
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 918) * as data is added to any of the @buffer's cpu buffers. Otherwise
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 919) * it will wait for data to be added to a specific cpu buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 920) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 921) * Returns EPOLLIN | EPOLLRDNORM if data exists in the buffers,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 922) * zero otherwise.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 923) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 924) __poll_t ring_buffer_poll_wait(struct trace_buffer *buffer, int cpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 925) struct file *filp, poll_table *poll_table)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 926) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 927) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 928) struct rb_irq_work *work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 929)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 930) if (cpu == RING_BUFFER_ALL_CPUS)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 931) work = &buffer->irq_work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 932) else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 933) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 934) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 935)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 936) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 937) work = &cpu_buffer->irq_work;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 938) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 939)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 940) poll_wait(filp, &work->waiters, poll_table);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 941) work->waiters_pending = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 942) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 943) * There's a tight race between setting the waiters_pending and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 944) * checking if the ring buffer is empty. Once the waiters_pending bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 945) * is set, the next event will wake the task up, but we can get stuck
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 946) * if there's only a single event in.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 947) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 948) * FIXME: Ideally, we need a memory barrier on the writer side as well,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 949) * but adding a memory barrier to all events will cause too much of a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 950) * performance hit in the fast path. We only need a memory barrier when
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 951) * the buffer goes from empty to having content. But as this race is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 952) * extremely small, and it's not a problem if another event comes in, we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 953) * will fix it later.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 954) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 955) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 956)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 957) if ((cpu == RING_BUFFER_ALL_CPUS && !ring_buffer_empty(buffer)) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 958) (cpu != RING_BUFFER_ALL_CPUS && !ring_buffer_empty_cpu(buffer, cpu)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 959) return EPOLLIN | EPOLLRDNORM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 960) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 961) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 962)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 963) /* buffer may be either ring_buffer or ring_buffer_per_cpu */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 964) #define RB_WARN_ON(b, cond) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 965) ({ \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 966) int _____ret = unlikely(cond); \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 967) if (_____ret) { \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 968) if (__same_type(*(b), struct ring_buffer_per_cpu)) { \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 969) struct ring_buffer_per_cpu *__b = \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 970) (void *)b; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 971) atomic_inc(&__b->buffer->record_disabled); \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 972) } else \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 973) atomic_inc(&b->record_disabled); \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 974) WARN_ON(1); \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 975) } \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 976) _____ret; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 977) })
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 978)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 979) /* Up this if you want to test the TIME_EXTENTS and normalization */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 980) #define DEBUG_SHIFT 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 981)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 982) static inline u64 rb_time_stamp(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 983) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 984) u64 ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 985)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 986) /* Skip retpolines :-( */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 987) if (IS_ENABLED(CONFIG_RETPOLINE) && likely(buffer->clock == trace_clock_local))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 988) ts = trace_clock_local();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 989) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 990) ts = buffer->clock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 991)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 992) /* shift to debug/test normalization and TIME_EXTENTS */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 993) return ts << DEBUG_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 994) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 995)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 996) u64 ring_buffer_time_stamp(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 997) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 998) u64 time;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 999)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1000) preempt_disable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1001) time = rb_time_stamp(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1002) preempt_enable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1003)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1004) return time;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1005) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1006) EXPORT_SYMBOL_GPL(ring_buffer_time_stamp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1007)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1008) void ring_buffer_normalize_time_stamp(struct trace_buffer *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1009) int cpu, u64 *ts)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1010) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1011) /* Just stupid testing the normalize function and deltas */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1012) *ts >>= DEBUG_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1013) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1014) EXPORT_SYMBOL_GPL(ring_buffer_normalize_time_stamp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1015)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1016) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1017) * Making the ring buffer lockless makes things tricky.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1018) * Although writes only happen on the CPU that they are on,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1019) * and they only need to worry about interrupts. Reads can
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1020) * happen on any CPU.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1021) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1022) * The reader page is always off the ring buffer, but when the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1023) * reader finishes with a page, it needs to swap its page with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1024) * a new one from the buffer. The reader needs to take from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1025) * the head (writes go to the tail). But if a writer is in overwrite
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1026) * mode and wraps, it must push the head page forward.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1027) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1028) * Here lies the problem.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1029) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1030) * The reader must be careful to replace only the head page, and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1031) * not another one. As described at the top of the file in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1032) * ASCII art, the reader sets its old page to point to the next
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1033) * page after head. It then sets the page after head to point to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1034) * the old reader page. But if the writer moves the head page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1035) * during this operation, the reader could end up with the tail.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1036) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1037) * We use cmpxchg to help prevent this race. We also do something
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1038) * special with the page before head. We set the LSB to 1.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1039) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1040) * When the writer must push the page forward, it will clear the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1041) * bit that points to the head page, move the head, and then set
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1042) * the bit that points to the new head page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1043) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1044) * We also don't want an interrupt coming in and moving the head
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1045) * page on another writer. Thus we use the second LSB to catch
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1046) * that too. Thus:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1047) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1048) * head->list->prev->next bit 1 bit 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1049) * ------- -------
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1050) * Normal page 0 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1051) * Points to head page 0 1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1052) * New head page 1 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1053) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1054) * Note we can not trust the prev pointer of the head page, because:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1055) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1056) * +----+ +-----+ +-----+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1057) * | |------>| T |---X--->| N |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1058) * | |<------| | | |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1059) * +----+ +-----+ +-----+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1060) * ^ ^ |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1061) * | +-----+ | |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1062) * +----------| R |----------+ |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1063) * | |<-----------+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1064) * +-----+
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1065) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1066) * Key: ---X--> HEAD flag set in pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1067) * T Tail page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1068) * R Reader page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1069) * N Next page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1070) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1071) * (see __rb_reserve_next() to see where this happens)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1072) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1073) * What the above shows is that the reader just swapped out
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1074) * the reader page with a page in the buffer, but before it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1075) * could make the new header point back to the new page added
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1076) * it was preempted by a writer. The writer moved forward onto
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1077) * the new page added by the reader and is about to move forward
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1078) * again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1079) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1080) * You can see, it is legitimate for the previous pointer of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1081) * the head (or any page) not to point back to itself. But only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1082) * temporarily.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1083) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1084)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1085) #define RB_PAGE_NORMAL 0UL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1086) #define RB_PAGE_HEAD 1UL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1087) #define RB_PAGE_UPDATE 2UL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1088)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1089)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1090) #define RB_FLAG_MASK 3UL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1091)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1092) /* PAGE_MOVED is not part of the mask */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1093) #define RB_PAGE_MOVED 4UL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1094)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1095) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1096) * rb_list_head - remove any bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1097) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1098) static struct list_head *rb_list_head(struct list_head *list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1099) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1100) unsigned long val = (unsigned long)list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1101)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1102) return (struct list_head *)(val & ~RB_FLAG_MASK);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1103) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1104)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1105) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1106) * rb_is_head_page - test if the given page is the head page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1107) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1108) * Because the reader may move the head_page pointer, we can
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1109) * not trust what the head page is (it may be pointing to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1110) * the reader page). But if the next page is a header page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1111) * its flags will be non zero.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1112) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1113) static inline int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1114) rb_is_head_page(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1115) struct buffer_page *page, struct list_head *list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1116) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1117) unsigned long val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1118)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1119) val = (unsigned long)list->next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1120)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1121) if ((val & ~RB_FLAG_MASK) != (unsigned long)&page->list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1122) return RB_PAGE_MOVED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1123)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1124) return val & RB_FLAG_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1125) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1126)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1127) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1128) * rb_is_reader_page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1129) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1130) * The unique thing about the reader page, is that, if the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1131) * writer is ever on it, the previous pointer never points
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1132) * back to the reader page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1133) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1134) static bool rb_is_reader_page(struct buffer_page *page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1135) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1136) struct list_head *list = page->list.prev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1137)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1138) return rb_list_head(list->next) != &page->list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1139) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1140)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1141) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1142) * rb_set_list_to_head - set a list_head to be pointing to head.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1143) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1144) static void rb_set_list_to_head(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1145) struct list_head *list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1146) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1147) unsigned long *ptr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1148)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1149) ptr = (unsigned long *)&list->next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1150) *ptr |= RB_PAGE_HEAD;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1151) *ptr &= ~RB_PAGE_UPDATE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1152) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1153)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1154) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1155) * rb_head_page_activate - sets up head page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1156) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1157) static void rb_head_page_activate(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1158) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1159) struct buffer_page *head;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1160)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1161) head = cpu_buffer->head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1162) if (!head)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1163) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1164)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1165) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1166) * Set the previous list pointer to have the HEAD flag.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1167) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1168) rb_set_list_to_head(cpu_buffer, head->list.prev);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1169) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1170)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1171) static void rb_list_head_clear(struct list_head *list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1172) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1173) unsigned long *ptr = (unsigned long *)&list->next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1174)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1175) *ptr &= ~RB_FLAG_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1176) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1177)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1178) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1179) * rb_head_page_deactivate - clears head page ptr (for free list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1180) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1181) static void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1182) rb_head_page_deactivate(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1183) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1184) struct list_head *hd;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1185)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1186) /* Go through the whole list and clear any pointers found. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1187) rb_list_head_clear(cpu_buffer->pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1188)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1189) list_for_each(hd, cpu_buffer->pages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1190) rb_list_head_clear(hd);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1191) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1192)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1193) static int rb_head_page_set(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1194) struct buffer_page *head,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1195) struct buffer_page *prev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1196) int old_flag, int new_flag)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1197) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1198) struct list_head *list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1199) unsigned long val = (unsigned long)&head->list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1200) unsigned long ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1201)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1202) list = &prev->list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1203)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1204) val &= ~RB_FLAG_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1205)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1206) ret = cmpxchg((unsigned long *)&list->next,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1207) val | old_flag, val | new_flag);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1208)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1209) /* check if the reader took the page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1210) if ((ret & ~RB_FLAG_MASK) != val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1211) return RB_PAGE_MOVED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1212)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1213) return ret & RB_FLAG_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1214) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1215)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1216) static int rb_head_page_set_update(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1217) struct buffer_page *head,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1218) struct buffer_page *prev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1219) int old_flag)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1220) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1221) return rb_head_page_set(cpu_buffer, head, prev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1222) old_flag, RB_PAGE_UPDATE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1223) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1224)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1225) static int rb_head_page_set_head(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1226) struct buffer_page *head,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1227) struct buffer_page *prev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1228) int old_flag)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1229) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1230) return rb_head_page_set(cpu_buffer, head, prev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1231) old_flag, RB_PAGE_HEAD);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1232) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1233)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1234) static int rb_head_page_set_normal(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1235) struct buffer_page *head,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1236) struct buffer_page *prev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1237) int old_flag)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1238) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1239) return rb_head_page_set(cpu_buffer, head, prev,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1240) old_flag, RB_PAGE_NORMAL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1241) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1242)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1243) static inline void rb_inc_page(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1244) struct buffer_page **bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1245) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1246) struct list_head *p = rb_list_head((*bpage)->list.next);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1247)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1248) *bpage = list_entry(p, struct buffer_page, list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1249) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1250)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1251) static struct buffer_page *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1252) rb_set_head_page(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1253) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1254) struct buffer_page *head;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1255) struct buffer_page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1256) struct list_head *list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1257) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1258)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1259) if (RB_WARN_ON(cpu_buffer, !cpu_buffer->head_page))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1260) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1261)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1262) /* sanity check */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1263) list = cpu_buffer->pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1264) if (RB_WARN_ON(cpu_buffer, rb_list_head(list->prev->next) != list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1265) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1266)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1267) page = head = cpu_buffer->head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1268) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1269) * It is possible that the writer moves the header behind
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1270) * where we started, and we miss in one loop.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1271) * A second loop should grab the header, but we'll do
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1272) * three loops just because I'm paranoid.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1273) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1274) for (i = 0; i < 3; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1275) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1276) if (rb_is_head_page(cpu_buffer, page, page->list.prev)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1277) cpu_buffer->head_page = page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1278) return page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1279) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1280) rb_inc_page(cpu_buffer, &page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1281) } while (page != head);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1282) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1283)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1284) RB_WARN_ON(cpu_buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1285)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1286) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1287) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1288)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1289) static int rb_head_page_replace(struct buffer_page *old,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1290) struct buffer_page *new)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1291) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1292) unsigned long *ptr = (unsigned long *)&old->list.prev->next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1293) unsigned long val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1294) unsigned long ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1295)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1296) val = *ptr & ~RB_FLAG_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1297) val |= RB_PAGE_HEAD;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1298)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1299) ret = cmpxchg(ptr, val, (unsigned long)&new->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1300)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1301) return ret == val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1302) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1303)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1304) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1305) * rb_tail_page_update - move the tail page forward
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1306) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1307) static void rb_tail_page_update(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1308) struct buffer_page *tail_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1309) struct buffer_page *next_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1310) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1311) unsigned long old_entries;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1312) unsigned long old_write;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1313)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1314) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1315) * The tail page now needs to be moved forward.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1316) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1317) * We need to reset the tail page, but without messing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1318) * with possible erasing of data brought in by interrupts
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1319) * that have moved the tail page and are currently on it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1320) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1321) * We add a counter to the write field to denote this.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1322) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1323) old_write = local_add_return(RB_WRITE_INTCNT, &next_page->write);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1324) old_entries = local_add_return(RB_WRITE_INTCNT, &next_page->entries);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1325)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1326) local_inc(&cpu_buffer->pages_touched);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1327) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1328) * Just make sure we have seen our old_write and synchronize
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1329) * with any interrupts that come in.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1330) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1331) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1332)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1333) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1334) * If the tail page is still the same as what we think
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1335) * it is, then it is up to us to update the tail
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1336) * pointer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1337) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1338) if (tail_page == READ_ONCE(cpu_buffer->tail_page)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1339) /* Zero the write counter */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1340) unsigned long val = old_write & ~RB_WRITE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1341) unsigned long eval = old_entries & ~RB_WRITE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1342)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1343) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1344) * This will only succeed if an interrupt did
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1345) * not come in and change it. In which case, we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1346) * do not want to modify it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1347) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1348) * We add (void) to let the compiler know that we do not care
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1349) * about the return value of these functions. We use the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1350) * cmpxchg to only update if an interrupt did not already
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1351) * do it for us. If the cmpxchg fails, we don't care.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1352) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1353) (void)local_cmpxchg(&next_page->write, old_write, val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1354) (void)local_cmpxchg(&next_page->entries, old_entries, eval);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1355)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1356) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1357) * No need to worry about races with clearing out the commit.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1358) * it only can increment when a commit takes place. But that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1359) * only happens in the outer most nested commit.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1360) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1361) local_set(&next_page->page->commit, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1362)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1363) /* Again, either we update tail_page or an interrupt does */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1364) (void)cmpxchg(&cpu_buffer->tail_page, tail_page, next_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1365) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1366) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1367)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1368) static int rb_check_bpage(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1369) struct buffer_page *bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1370) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1371) unsigned long val = (unsigned long)bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1372)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1373) if (RB_WARN_ON(cpu_buffer, val & RB_FLAG_MASK))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1374) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1375)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1376) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1377) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1378)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1379) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1380) * rb_check_list - make sure a pointer to a list has the last bits zero
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1381) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1382) static int rb_check_list(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1383) struct list_head *list)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1384) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1385) if (RB_WARN_ON(cpu_buffer, rb_list_head(list->prev) != list->prev))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1386) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1387) if (RB_WARN_ON(cpu_buffer, rb_list_head(list->next) != list->next))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1388) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1389) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1390) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1391)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1392) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1393) * rb_check_pages - integrity check of buffer pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1394) * @cpu_buffer: CPU buffer with pages to test
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1395) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1396) * As a safety measure we check to make sure the data pages have not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1397) * been corrupted.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1398) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1399) static int rb_check_pages(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1400) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1401) struct list_head *head = cpu_buffer->pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1402) struct buffer_page *bpage, *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1403)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1404) /* Reset the head page if it exists */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1405) if (cpu_buffer->head_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1406) rb_set_head_page(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1407)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1408) rb_head_page_deactivate(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1409)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1410) if (RB_WARN_ON(cpu_buffer, head->next->prev != head))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1411) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1412) if (RB_WARN_ON(cpu_buffer, head->prev->next != head))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1413) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1414)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1415) if (rb_check_list(cpu_buffer, head))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1416) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1417)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1418) list_for_each_entry_safe(bpage, tmp, head, list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1419) if (RB_WARN_ON(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1420) bpage->list.next->prev != &bpage->list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1421) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1422) if (RB_WARN_ON(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1423) bpage->list.prev->next != &bpage->list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1424) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1425) if (rb_check_list(cpu_buffer, &bpage->list))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1426) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1427) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1428)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1429) rb_head_page_activate(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1430)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1431) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1432) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1433)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1434) static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1435) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1436) struct buffer_page *bpage, *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1437) bool user_thread = current->mm != NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1438) gfp_t mflags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1439) long i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1440)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1441) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1442) * Check if the available memory is there first.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1443) * Note, si_mem_available() only gives us a rough estimate of available
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1444) * memory. It may not be accurate. But we don't care, we just want
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1445) * to prevent doing any allocation when it is obvious that it is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1446) * not going to succeed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1447) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1448) i = si_mem_available();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1449) if (i < nr_pages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1450) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1451)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1452) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1453) * __GFP_RETRY_MAYFAIL flag makes sure that the allocation fails
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1454) * gracefully without invoking oom-killer and the system is not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1455) * destabilized.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1456) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1457) mflags = GFP_KERNEL | __GFP_RETRY_MAYFAIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1458)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1459) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1460) * If a user thread allocates too much, and si_mem_available()
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1461) * reports there's enough memory, even though there is not.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1462) * Make sure the OOM killer kills this thread. This can happen
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1463) * even with RETRY_MAYFAIL because another task may be doing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1464) * an allocation after this task has taken all memory.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1465) * This is the task the OOM killer needs to take out during this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1466) * loop, even if it was triggered by an allocation somewhere else.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1467) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1468) if (user_thread)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1469) set_current_oom_origin();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1470) for (i = 0; i < nr_pages; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1471) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1472)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1473) bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1474) mflags, cpu_to_node(cpu));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1475) if (!bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1476) goto free_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1477)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1478) list_add(&bpage->list, pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1479)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1480) page = alloc_pages_node(cpu_to_node(cpu), mflags, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1481) if (!page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1482) goto free_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1483) bpage->page = page_address(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1484) rb_init_page(bpage->page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1485)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1486) if (user_thread && fatal_signal_pending(current))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1487) goto free_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1488) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1489) if (user_thread)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1490) clear_current_oom_origin();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1491)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1492) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1493)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1494) free_pages:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1495) list_for_each_entry_safe(bpage, tmp, pages, list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1496) list_del_init(&bpage->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1497) free_buffer_page(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1498) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1499) if (user_thread)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1500) clear_current_oom_origin();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1501)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1502) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1503) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1504)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1505) static int rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1506) unsigned long nr_pages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1507) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1508) LIST_HEAD(pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1509)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1510) WARN_ON(!nr_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1511)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1512) if (__rb_allocate_pages(nr_pages, &pages, cpu_buffer->cpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1513) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1514)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1515) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1516) * The ring buffer page list is a circular list that does not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1517) * start and end with a list head. All page list items point to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1518) * other pages.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1519) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1520) cpu_buffer->pages = pages.next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1521) list_del(&pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1522)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1523) cpu_buffer->nr_pages = nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1524)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1525) rb_check_pages(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1526)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1527) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1528) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1529)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1530) static struct ring_buffer_per_cpu *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1531) rb_allocate_cpu_buffer(struct trace_buffer *buffer, long nr_pages, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1532) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1533) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1534) struct buffer_page *bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1535) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1536) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1537)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1538) cpu_buffer = kzalloc_node(ALIGN(sizeof(*cpu_buffer), cache_line_size()),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1539) GFP_KERNEL, cpu_to_node(cpu));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1540) if (!cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1541) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1542)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1543) cpu_buffer->cpu = cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1544) cpu_buffer->buffer = buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1545) raw_spin_lock_init(&cpu_buffer->reader_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1546) lockdep_set_class(&cpu_buffer->reader_lock, buffer->reader_lock_key);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1547) cpu_buffer->lock = (arch_spinlock_t)__ARCH_SPIN_LOCK_UNLOCKED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1548) INIT_WORK(&cpu_buffer->update_pages_work, update_pages_handler);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1549) init_completion(&cpu_buffer->update_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1550) init_irq_work(&cpu_buffer->irq_work.work, rb_wake_up_waiters);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1551) init_waitqueue_head(&cpu_buffer->irq_work.waiters);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1552) init_waitqueue_head(&cpu_buffer->irq_work.full_waiters);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1553)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1554) bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1555) GFP_KERNEL, cpu_to_node(cpu));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1556) if (!bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1557) goto fail_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1558)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1559) rb_check_bpage(cpu_buffer, bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1560)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1561) cpu_buffer->reader_page = bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1562) page = alloc_pages_node(cpu_to_node(cpu), GFP_KERNEL, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1563) if (!page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1564) goto fail_free_reader;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1565) bpage->page = page_address(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1566) rb_init_page(bpage->page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1567)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1568) INIT_LIST_HEAD(&cpu_buffer->reader_page->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1569) INIT_LIST_HEAD(&cpu_buffer->new_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1570)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1571) ret = rb_allocate_pages(cpu_buffer, nr_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1572) if (ret < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1573) goto fail_free_reader;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1574)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1575) cpu_buffer->head_page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1576) = list_entry(cpu_buffer->pages, struct buffer_page, list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1577) cpu_buffer->tail_page = cpu_buffer->commit_page = cpu_buffer->head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1578)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1579) rb_head_page_activate(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1580)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1581) return cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1582)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1583) fail_free_reader:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1584) free_buffer_page(cpu_buffer->reader_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1585)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1586) fail_free_buffer:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1587) kfree(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1588) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1589) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1590)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1591) static void rb_free_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1592) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1593) struct list_head *head = cpu_buffer->pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1594) struct buffer_page *bpage, *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1595)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1596) free_buffer_page(cpu_buffer->reader_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1597)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1598) rb_head_page_deactivate(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1599)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1600) if (head) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1601) list_for_each_entry_safe(bpage, tmp, head, list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1602) list_del_init(&bpage->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1603) free_buffer_page(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1604) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1605) bpage = list_entry(head, struct buffer_page, list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1606) free_buffer_page(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1607) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1608)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1609) kfree(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1610) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1611)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1612) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1613) * __ring_buffer_alloc - allocate a new ring_buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1614) * @size: the size in bytes per cpu that is needed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1615) * @flags: attributes to set for the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1616) * @key: ring buffer reader_lock_key.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1617) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1618) * Currently the only flag that is available is the RB_FL_OVERWRITE
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1619) * flag. This flag means that the buffer will overwrite old data
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1620) * when the buffer wraps. If this flag is not set, the buffer will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1621) * drop data when the tail hits the head.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1622) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1623) struct trace_buffer *__ring_buffer_alloc(unsigned long size, unsigned flags,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1624) struct lock_class_key *key)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1625) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1626) struct trace_buffer *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1627) long nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1628) int bsize;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1629) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1630) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1631)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1632) /* keep it in its own cache line */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1633) buffer = kzalloc(ALIGN(sizeof(*buffer), cache_line_size()),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1634) GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1635) if (!buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1636) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1637)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1638) if (!zalloc_cpumask_var(&buffer->cpumask, GFP_KERNEL))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1639) goto fail_free_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1640)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1641) nr_pages = DIV_ROUND_UP(size, BUF_PAGE_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1642) buffer->flags = flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1643) buffer->clock = trace_clock_local;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1644) buffer->reader_lock_key = key;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1645)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1646) init_irq_work(&buffer->irq_work.work, rb_wake_up_waiters);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1647) init_waitqueue_head(&buffer->irq_work.waiters);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1648)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1649) /* need at least two pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1650) if (nr_pages < 2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1651) nr_pages = 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1652)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1653) buffer->cpus = nr_cpu_ids;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1654)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1655) bsize = sizeof(void *) * nr_cpu_ids;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1656) buffer->buffers = kzalloc(ALIGN(bsize, cache_line_size()),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1657) GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1658) if (!buffer->buffers)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1659) goto fail_free_cpumask;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1660)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1661) cpu = raw_smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1662) cpumask_set_cpu(cpu, buffer->cpumask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1663) buffer->buffers[cpu] = rb_allocate_cpu_buffer(buffer, nr_pages, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1664) if (!buffer->buffers[cpu])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1665) goto fail_free_buffers;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1666)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1667) ret = cpuhp_state_add_instance(CPUHP_TRACE_RB_PREPARE, &buffer->node);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1668) if (ret < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1669) goto fail_free_buffers;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1670)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1671) mutex_init(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1672)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1673) return buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1674)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1675) fail_free_buffers:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1676) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1677) if (buffer->buffers[cpu])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1678) rb_free_cpu_buffer(buffer->buffers[cpu]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1679) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1680) kfree(buffer->buffers);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1681)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1682) fail_free_cpumask:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1683) free_cpumask_var(buffer->cpumask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1684)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1685) fail_free_buffer:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1686) kfree(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1687) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1688) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1689) EXPORT_SYMBOL_GPL(__ring_buffer_alloc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1690)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1691) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1692) * ring_buffer_free - free a ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1693) * @buffer: the buffer to free.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1694) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1695) void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1696) ring_buffer_free(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1697) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1698) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1699)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1700) cpuhp_state_remove_instance(CPUHP_TRACE_RB_PREPARE, &buffer->node);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1701)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1702) for_each_buffer_cpu(buffer, cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1703) rb_free_cpu_buffer(buffer->buffers[cpu]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1704)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1705) kfree(buffer->buffers);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1706) free_cpumask_var(buffer->cpumask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1707)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1708) kfree(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1709) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1710) EXPORT_SYMBOL_GPL(ring_buffer_free);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1711)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1712) void ring_buffer_set_clock(struct trace_buffer *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1713) u64 (*clock)(void))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1714) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1715) buffer->clock = clock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1716) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1717)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1718) void ring_buffer_set_time_stamp_abs(struct trace_buffer *buffer, bool abs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1719) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1720) buffer->time_stamp_abs = abs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1721) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1722)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1723) bool ring_buffer_time_stamp_abs(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1724) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1725) return buffer->time_stamp_abs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1726) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1727)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1728) static void rb_reset_cpu(struct ring_buffer_per_cpu *cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1729)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1730) static inline unsigned long rb_page_entries(struct buffer_page *bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1731) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1732) return local_read(&bpage->entries) & RB_WRITE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1733) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1734)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1735) static inline unsigned long rb_page_write(struct buffer_page *bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1736) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1737) return local_read(&bpage->write) & RB_WRITE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1738) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1739)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1740) static int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1741) rb_remove_pages(struct ring_buffer_per_cpu *cpu_buffer, unsigned long nr_pages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1742) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1743) struct list_head *tail_page, *to_remove, *next_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1744) struct buffer_page *to_remove_page, *tmp_iter_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1745) struct buffer_page *last_page, *first_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1746) unsigned long nr_removed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1747) unsigned long head_bit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1748) int page_entries;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1749)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1750) head_bit = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1751)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1752) raw_spin_lock_irq(&cpu_buffer->reader_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1753) atomic_inc(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1754) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1755) * We don't race with the readers since we have acquired the reader
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1756) * lock. We also don't race with writers after disabling recording.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1757) * This makes it easy to figure out the first and the last page to be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1758) * removed from the list. We unlink all the pages in between including
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1759) * the first and last pages. This is done in a busy loop so that we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1760) * lose the least number of traces.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1761) * The pages are freed after we restart recording and unlock readers.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1762) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1763) tail_page = &cpu_buffer->tail_page->list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1764)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1765) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1766) * tail page might be on reader page, we remove the next page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1767) * from the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1768) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1769) if (cpu_buffer->tail_page == cpu_buffer->reader_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1770) tail_page = rb_list_head(tail_page->next);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1771) to_remove = tail_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1772)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1773) /* start of pages to remove */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1774) first_page = list_entry(rb_list_head(to_remove->next),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1775) struct buffer_page, list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1776)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1777) for (nr_removed = 0; nr_removed < nr_pages; nr_removed++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1778) to_remove = rb_list_head(to_remove)->next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1779) head_bit |= (unsigned long)to_remove & RB_PAGE_HEAD;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1780) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1781)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1782) next_page = rb_list_head(to_remove)->next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1783)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1784) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1785) * Now we remove all pages between tail_page and next_page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1786) * Make sure that we have head_bit value preserved for the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1787) * next page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1788) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1789) tail_page->next = (struct list_head *)((unsigned long)next_page |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1790) head_bit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1791) next_page = rb_list_head(next_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1792) next_page->prev = tail_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1793)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1794) /* make sure pages points to a valid page in the ring buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1795) cpu_buffer->pages = next_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1796)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1797) /* update head page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1798) if (head_bit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1799) cpu_buffer->head_page = list_entry(next_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1800) struct buffer_page, list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1801)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1802) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1803) * change read pointer to make sure any read iterators reset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1804) * themselves
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1805) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1806) cpu_buffer->read = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1807)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1808) /* pages are removed, resume tracing and then free the pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1809) atomic_dec(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1810) raw_spin_unlock_irq(&cpu_buffer->reader_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1811)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1812) RB_WARN_ON(cpu_buffer, list_empty(cpu_buffer->pages));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1813)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1814) /* last buffer page to remove */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1815) last_page = list_entry(rb_list_head(to_remove), struct buffer_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1816) list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1817) tmp_iter_page = first_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1818)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1819) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1820) cond_resched();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1821)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1822) to_remove_page = tmp_iter_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1823) rb_inc_page(cpu_buffer, &tmp_iter_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1824)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1825) /* update the counters */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1826) page_entries = rb_page_entries(to_remove_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1827) if (page_entries) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1828) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1829) * If something was added to this page, it was full
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1830) * since it is not the tail page. So we deduct the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1831) * bytes consumed in ring buffer from here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1832) * Increment overrun to account for the lost events.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1833) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1834) local_add(page_entries, &cpu_buffer->overrun);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1835) local_sub(BUF_PAGE_SIZE, &cpu_buffer->entries_bytes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1836) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1837)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1838) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1839) * We have already removed references to this list item, just
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1840) * free up the buffer_page and its page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1841) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1842) free_buffer_page(to_remove_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1843) nr_removed--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1844)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1845) } while (to_remove_page != last_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1846)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1847) RB_WARN_ON(cpu_buffer, nr_removed);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1848)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1849) return nr_removed == 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1850) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1851)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1852) static int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1853) rb_insert_pages(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1854) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1855) struct list_head *pages = &cpu_buffer->new_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1856) int retries, success;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1857)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1858) raw_spin_lock_irq(&cpu_buffer->reader_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1859) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1860) * We are holding the reader lock, so the reader page won't be swapped
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1861) * in the ring buffer. Now we are racing with the writer trying to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1862) * move head page and the tail page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1863) * We are going to adapt the reader page update process where:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1864) * 1. We first splice the start and end of list of new pages between
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1865) * the head page and its previous page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1866) * 2. We cmpxchg the prev_page->next to point from head page to the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1867) * start of new pages list.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1868) * 3. Finally, we update the head->prev to the end of new list.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1869) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1870) * We will try this process 10 times, to make sure that we don't keep
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1871) * spinning.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1872) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1873) retries = 10;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1874) success = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1875) while (retries--) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1876) struct list_head *head_page, *prev_page, *r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1877) struct list_head *last_page, *first_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1878) struct list_head *head_page_with_bit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1879)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1880) head_page = &rb_set_head_page(cpu_buffer)->list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1881) if (!head_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1882) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1883) prev_page = head_page->prev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1884)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1885) first_page = pages->next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1886) last_page = pages->prev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1887)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1888) head_page_with_bit = (struct list_head *)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1889) ((unsigned long)head_page | RB_PAGE_HEAD);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1890)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1891) last_page->next = head_page_with_bit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1892) first_page->prev = prev_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1893)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1894) r = cmpxchg(&prev_page->next, head_page_with_bit, first_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1895)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1896) if (r == head_page_with_bit) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1897) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1898) * yay, we replaced the page pointer to our new list,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1899) * now, we just have to update to head page's prev
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1900) * pointer to point to end of list
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1901) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1902) head_page->prev = last_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1903) success = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1904) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1905) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1906) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1907)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1908) if (success)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1909) INIT_LIST_HEAD(pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1910) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1911) * If we weren't successful in adding in new pages, warn and stop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1912) * tracing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1913) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1914) RB_WARN_ON(cpu_buffer, !success);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1915) raw_spin_unlock_irq(&cpu_buffer->reader_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1916)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1917) /* free pages if they weren't inserted */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1918) if (!success) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1919) struct buffer_page *bpage, *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1920) list_for_each_entry_safe(bpage, tmp, &cpu_buffer->new_pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1921) list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1922) list_del_init(&bpage->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1923) free_buffer_page(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1924) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1925) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1926) return success;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1927) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1928)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1929) static void rb_update_pages(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1930) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1931) int success;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1932)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1933) if (cpu_buffer->nr_pages_to_update > 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1934) success = rb_insert_pages(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1935) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1936) success = rb_remove_pages(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1937) -cpu_buffer->nr_pages_to_update);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1938)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1939) if (success)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1940) cpu_buffer->nr_pages += cpu_buffer->nr_pages_to_update;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1941) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1942)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1943) static void update_pages_handler(struct work_struct *work)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1944) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1945) struct ring_buffer_per_cpu *cpu_buffer = container_of(work,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1946) struct ring_buffer_per_cpu, update_pages_work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1947) rb_update_pages(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1948) complete(&cpu_buffer->update_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1949) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1950)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1951) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1952) * ring_buffer_resize - resize the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1953) * @buffer: the buffer to resize.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1954) * @size: the new size.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1955) * @cpu_id: the cpu buffer to resize
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1956) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1957) * Minimum size is 2 * BUF_PAGE_SIZE.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1958) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1959) * Returns 0 on success and < 0 on failure.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1960) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1961) int ring_buffer_resize(struct trace_buffer *buffer, unsigned long size,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1962) int cpu_id)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1963) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1964) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1965) unsigned long nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1966) int cpu, err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1967)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1968) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1969) * Always succeed at resizing a non-existent buffer:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1970) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1971) if (!buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1972) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1973)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1974) /* Make sure the requested buffer exists */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1975) if (cpu_id != RING_BUFFER_ALL_CPUS &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1976) !cpumask_test_cpu(cpu_id, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1977) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1978)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1979) nr_pages = DIV_ROUND_UP(size, BUF_PAGE_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1980)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1981) /* we need a minimum of two pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1982) if (nr_pages < 2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1983) nr_pages = 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1984)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1985) size = nr_pages * BUF_PAGE_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1986)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1987) /* prevent another thread from changing buffer sizes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1988) mutex_lock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1989)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1990)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1991) if (cpu_id == RING_BUFFER_ALL_CPUS) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1992) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1993) * Don't succeed if resizing is disabled, as a reader might be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1994) * manipulating the ring buffer and is expecting a sane state while
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1995) * this is true.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1996) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1997) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1998) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1999) if (atomic_read(&cpu_buffer->resize_disabled)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2000) err = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2001) goto out_err_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2002) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2003) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2004)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2005) /* calculate the pages to update */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2006) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2007) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2008)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2009) cpu_buffer->nr_pages_to_update = nr_pages -
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2010) cpu_buffer->nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2011) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2012) * nothing more to do for removing pages or no update
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2013) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2014) if (cpu_buffer->nr_pages_to_update <= 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2015) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2016) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2017) * to add pages, make sure all new pages can be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2018) * allocated without receiving ENOMEM
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2019) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2020) INIT_LIST_HEAD(&cpu_buffer->new_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2021) if (__rb_allocate_pages(cpu_buffer->nr_pages_to_update,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2022) &cpu_buffer->new_pages, cpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2023) /* not enough memory for new pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2024) err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2025) goto out_err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2026) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2027) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2028)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2029) get_online_cpus();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2030) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2031) * Fire off all the required work handlers
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2032) * We can't schedule on offline CPUs, but it's not necessary
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2033) * since we can change their buffer sizes without any race.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2034) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2035) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2036) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2037) if (!cpu_buffer->nr_pages_to_update)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2038) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2039)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2040) /* Can't run something on an offline CPU. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2041) if (!cpu_online(cpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2042) rb_update_pages(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2043) cpu_buffer->nr_pages_to_update = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2044) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2045) schedule_work_on(cpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2046) &cpu_buffer->update_pages_work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2047) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2048) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2049)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2050) /* wait for all the updates to complete */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2051) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2052) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2053) if (!cpu_buffer->nr_pages_to_update)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2054) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2055)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2056) if (cpu_online(cpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2057) wait_for_completion(&cpu_buffer->update_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2058) cpu_buffer->nr_pages_to_update = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2059) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2060)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2061) put_online_cpus();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2062) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2063) /* Make sure this CPU has been initialized */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2064) if (!cpumask_test_cpu(cpu_id, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2065) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2066)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2067) cpu_buffer = buffer->buffers[cpu_id];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2068)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2069) if (nr_pages == cpu_buffer->nr_pages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2070) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2071)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2072) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2073) * Don't succeed if resizing is disabled, as a reader might be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2074) * manipulating the ring buffer and is expecting a sane state while
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2075) * this is true.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2076) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2077) if (atomic_read(&cpu_buffer->resize_disabled)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2078) err = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2079) goto out_err_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2080) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2081)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2082) cpu_buffer->nr_pages_to_update = nr_pages -
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2083) cpu_buffer->nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2084)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2085) INIT_LIST_HEAD(&cpu_buffer->new_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2086) if (cpu_buffer->nr_pages_to_update > 0 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2087) __rb_allocate_pages(cpu_buffer->nr_pages_to_update,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2088) &cpu_buffer->new_pages, cpu_id)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2089) err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2090) goto out_err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2091) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2092)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2093) get_online_cpus();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2094)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2095) /* Can't run something on an offline CPU. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2096) if (!cpu_online(cpu_id))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2097) rb_update_pages(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2098) else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2099) schedule_work_on(cpu_id,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2100) &cpu_buffer->update_pages_work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2101) wait_for_completion(&cpu_buffer->update_done);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2102) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2103)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2104) cpu_buffer->nr_pages_to_update = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2105) put_online_cpus();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2106) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2107)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2108) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2109) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2110) * The ring buffer resize can happen with the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2111) * enabled, so that the update disturbs the tracing as little
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2112) * as possible. But if the buffer is disabled, we do not need
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2113) * to worry about that, and we can take the time to verify
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2114) * that the buffer is not corrupt.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2115) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2116) if (atomic_read(&buffer->record_disabled)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2117) atomic_inc(&buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2118) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2119) * Even though the buffer was disabled, we must make sure
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2120) * that it is truly disabled before calling rb_check_pages.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2121) * There could have been a race between checking
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2122) * record_disable and incrementing it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2123) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2124) synchronize_rcu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2125) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2126) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2127) rb_check_pages(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2128) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2129) atomic_dec(&buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2130) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2131)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2132) mutex_unlock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2133) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2134)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2135) out_err:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2136) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2137) struct buffer_page *bpage, *tmp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2138)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2139) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2140) cpu_buffer->nr_pages_to_update = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2141)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2142) if (list_empty(&cpu_buffer->new_pages))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2143) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2144)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2145) list_for_each_entry_safe(bpage, tmp, &cpu_buffer->new_pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2146) list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2147) list_del_init(&bpage->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2148) free_buffer_page(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2149) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2150) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2151) out_err_unlock:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2152) mutex_unlock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2153) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2154) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2155) EXPORT_SYMBOL_GPL(ring_buffer_resize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2156)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2157) void ring_buffer_change_overwrite(struct trace_buffer *buffer, int val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2158) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2159) mutex_lock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2160) if (val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2161) buffer->flags |= RB_FL_OVERWRITE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2162) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2163) buffer->flags &= ~RB_FL_OVERWRITE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2164) mutex_unlock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2165) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2166) EXPORT_SYMBOL_GPL(ring_buffer_change_overwrite);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2167)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2168) static __always_inline void *__rb_page_index(struct buffer_page *bpage, unsigned index)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2169) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2170) return bpage->page->data + index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2171) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2172)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2173) static __always_inline struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2174) rb_reader_event(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2175) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2176) return __rb_page_index(cpu_buffer->reader_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2177) cpu_buffer->reader_page->read);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2178) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2179)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2180) static __always_inline unsigned rb_page_commit(struct buffer_page *bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2181) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2182) return local_read(&bpage->page->commit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2183) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2184)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2185) static struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2186) rb_iter_head_event(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2187) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2188) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2189) struct buffer_page *iter_head_page = iter->head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2190) unsigned long commit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2191) unsigned length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2192)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2193) if (iter->head != iter->next_event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2194) return iter->event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2196) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2197) * When the writer goes across pages, it issues a cmpxchg which
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2198) * is a mb(), which will synchronize with the rmb here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2199) * (see rb_tail_page_update() and __rb_reserve_next())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2200) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2201) commit = rb_page_commit(iter_head_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2202) smp_rmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2203) event = __rb_page_index(iter_head_page, iter->head);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2204) length = rb_event_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2205)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2206) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2207) * READ_ONCE() doesn't work on functions and we don't want the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2208) * compiler doing any crazy optimizations with length.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2209) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2210) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2211)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2212) if ((iter->head + length) > commit || length > BUF_MAX_DATA_SIZE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2213) /* Writer corrupted the read? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2214) goto reset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2215)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2216) memcpy(iter->event, event, length);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2217) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2218) * If the page stamp is still the same after this rmb() then the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2219) * event was safely copied without the writer entering the page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2220) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2221) smp_rmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2222)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2223) /* Make sure the page didn't change since we read this */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2224) if (iter->page_stamp != iter_head_page->page->time_stamp ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2225) commit > rb_page_commit(iter_head_page))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2226) goto reset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2227)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2228) iter->next_event = iter->head + length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2229) return iter->event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2230) reset:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2231) /* Reset to the beginning */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2232) iter->page_stamp = iter->read_stamp = iter->head_page->page->time_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2233) iter->head = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2234) iter->next_event = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2235) iter->missed_events = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2236) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2237) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2238)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2239) /* Size is determined by what has been committed */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2240) static __always_inline unsigned rb_page_size(struct buffer_page *bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2241) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2242) return rb_page_commit(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2243) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2244)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2245) static __always_inline unsigned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2246) rb_commit_index(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2247) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2248) return rb_page_commit(cpu_buffer->commit_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2249) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2250)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2251) static __always_inline unsigned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2252) rb_event_index(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2253) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2254) unsigned long addr = (unsigned long)event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2255)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2256) return (addr & ~PAGE_MASK) - BUF_PAGE_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2257) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2258)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2259) static void rb_inc_iter(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2260) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2261) struct ring_buffer_per_cpu *cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2262)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2263) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2264) * The iterator could be on the reader page (it starts there).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2265) * But the head could have moved, since the reader was
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2266) * found. Check for this case and assign the iterator
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2267) * to the head page instead of next.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2268) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2269) if (iter->head_page == cpu_buffer->reader_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2270) iter->head_page = rb_set_head_page(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2271) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2272) rb_inc_page(cpu_buffer, &iter->head_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2273)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2274) iter->page_stamp = iter->read_stamp = iter->head_page->page->time_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2275) iter->head = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2276) iter->next_event = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2277) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2278)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2279) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2280) * rb_handle_head_page - writer hit the head page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2281) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2282) * Returns: +1 to retry page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2283) * 0 to continue
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2284) * -1 on error
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2285) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2286) static int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2287) rb_handle_head_page(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2288) struct buffer_page *tail_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2289) struct buffer_page *next_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2290) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2291) struct buffer_page *new_head;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2292) int entries;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2293) int type;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2294) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2295)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2296) entries = rb_page_entries(next_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2297)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2298) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2299) * The hard part is here. We need to move the head
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2300) * forward, and protect against both readers on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2301) * other CPUs and writers coming in via interrupts.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2302) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2303) type = rb_head_page_set_update(cpu_buffer, next_page, tail_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2304) RB_PAGE_HEAD);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2305)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2306) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2307) * type can be one of four:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2308) * NORMAL - an interrupt already moved it for us
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2309) * HEAD - we are the first to get here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2310) * UPDATE - we are the interrupt interrupting
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2311) * a current move.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2312) * MOVED - a reader on another CPU moved the next
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2313) * pointer to its reader page. Give up
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2314) * and try again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2315) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2316)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2317) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2318) case RB_PAGE_HEAD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2319) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2320) * We changed the head to UPDATE, thus
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2321) * it is our responsibility to update
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2322) * the counters.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2323) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2324) local_add(entries, &cpu_buffer->overrun);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2325) local_sub(BUF_PAGE_SIZE, &cpu_buffer->entries_bytes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2326)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2327) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2328) * The entries will be zeroed out when we move the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2329) * tail page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2330) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2331)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2332) /* still more to do */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2333) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2334)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2335) case RB_PAGE_UPDATE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2336) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2337) * This is an interrupt that interrupt the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2338) * previous update. Still more to do.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2339) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2340) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2341) case RB_PAGE_NORMAL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2342) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2343) * An interrupt came in before the update
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2344) * and processed this for us.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2345) * Nothing left to do.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2346) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2347) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2348) case RB_PAGE_MOVED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2349) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2350) * The reader is on another CPU and just did
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2351) * a swap with our next_page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2352) * Try again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2353) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2354) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2355) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2356) RB_WARN_ON(cpu_buffer, 1); /* WTF??? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2357) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2358) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2359)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2360) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2361) * Now that we are here, the old head pointer is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2362) * set to UPDATE. This will keep the reader from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2363) * swapping the head page with the reader page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2364) * The reader (on another CPU) will spin till
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2365) * we are finished.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2366) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2367) * We just need to protect against interrupts
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2368) * doing the job. We will set the next pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2369) * to HEAD. After that, we set the old pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2370) * to NORMAL, but only if it was HEAD before.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2371) * otherwise we are an interrupt, and only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2372) * want the outer most commit to reset it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2373) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2374) new_head = next_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2375) rb_inc_page(cpu_buffer, &new_head);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2376)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2377) ret = rb_head_page_set_head(cpu_buffer, new_head, next_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2378) RB_PAGE_NORMAL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2379)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2380) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2381) * Valid returns are:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2382) * HEAD - an interrupt came in and already set it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2383) * NORMAL - One of two things:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2384) * 1) We really set it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2385) * 2) A bunch of interrupts came in and moved
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2386) * the page forward again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2387) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2388) switch (ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2389) case RB_PAGE_HEAD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2390) case RB_PAGE_NORMAL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2391) /* OK */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2392) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2393) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2394) RB_WARN_ON(cpu_buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2395) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2396) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2397)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2398) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2399) * It is possible that an interrupt came in,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2400) * set the head up, then more interrupts came in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2401) * and moved it again. When we get back here,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2402) * the page would have been set to NORMAL but we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2403) * just set it back to HEAD.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2404) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2405) * How do you detect this? Well, if that happened
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2406) * the tail page would have moved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2407) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2408) if (ret == RB_PAGE_NORMAL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2409) struct buffer_page *buffer_tail_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2410)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2411) buffer_tail_page = READ_ONCE(cpu_buffer->tail_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2412) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2413) * If the tail had moved passed next, then we need
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2414) * to reset the pointer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2415) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2416) if (buffer_tail_page != tail_page &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2417) buffer_tail_page != next_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2418) rb_head_page_set_normal(cpu_buffer, new_head,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2419) next_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2420) RB_PAGE_HEAD);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2421) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2422)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2423) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2424) * If this was the outer most commit (the one that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2425) * changed the original pointer from HEAD to UPDATE),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2426) * then it is up to us to reset it to NORMAL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2427) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2428) if (type == RB_PAGE_HEAD) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2429) ret = rb_head_page_set_normal(cpu_buffer, next_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2430) tail_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2431) RB_PAGE_UPDATE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2432) if (RB_WARN_ON(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2433) ret != RB_PAGE_UPDATE))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2434) return -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2435) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2436)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2437) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2438) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2439)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2440) static inline void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2441) rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2442) unsigned long tail, struct rb_event_info *info)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2443) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2444) struct buffer_page *tail_page = info->tail_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2445) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2446) unsigned long length = info->length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2447)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2448) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2449) * Only the event that crossed the page boundary
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2450) * must fill the old tail_page with padding.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2451) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2452) if (tail >= BUF_PAGE_SIZE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2453) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2454) * If the page was filled, then we still need
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2455) * to update the real_end. Reset it to zero
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2456) * and the reader will ignore it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2457) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2458) if (tail == BUF_PAGE_SIZE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2459) tail_page->real_end = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2460)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2461) local_sub(length, &tail_page->write);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2462) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2463) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2464)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2465) event = __rb_page_index(tail_page, tail);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2466)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2467) /* account for padding bytes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2468) local_add(BUF_PAGE_SIZE - tail, &cpu_buffer->entries_bytes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2469)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2470) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2471) * Save the original length to the meta data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2472) * This will be used by the reader to add lost event
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2473) * counter.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2474) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2475) tail_page->real_end = tail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2476)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2477) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2478) * If this event is bigger than the minimum size, then
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2479) * we need to be careful that we don't subtract the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2480) * write counter enough to allow another writer to slip
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2481) * in on this page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2482) * We put in a discarded commit instead, to make sure
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2483) * that this space is not used again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2484) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2485) * If we are less than the minimum size, we don't need to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2486) * worry about it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2487) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2488) if (tail > (BUF_PAGE_SIZE - RB_EVNT_MIN_SIZE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2489) /* No room for any events */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2490)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2491) /* Mark the rest of the page with padding */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2492) rb_event_set_padding(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2493)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2494) /* Set the write back to the previous setting */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2495) local_sub(length, &tail_page->write);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2496) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2497) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2498)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2499) /* Put in a discarded event */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2500) event->array[0] = (BUF_PAGE_SIZE - tail) - RB_EVNT_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2501) event->type_len = RINGBUF_TYPE_PADDING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2502) /* time delta must be non zero */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2503) event->time_delta = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2504)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2505) /* Set write to end of buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2506) length = (tail + length) - BUF_PAGE_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2507) local_sub(length, &tail_page->write);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2508) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2509)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2510) static inline void rb_end_commit(struct ring_buffer_per_cpu *cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2511)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2512) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2513) * This is the slow path, force gcc not to inline it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2514) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2515) static noinline struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2516) rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2517) unsigned long tail, struct rb_event_info *info)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2518) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2519) struct buffer_page *tail_page = info->tail_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2520) struct buffer_page *commit_page = cpu_buffer->commit_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2521) struct trace_buffer *buffer = cpu_buffer->buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2522) struct buffer_page *next_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2523) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2524)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2525) next_page = tail_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2526)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2527) rb_inc_page(cpu_buffer, &next_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2528)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2529) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2530) * If for some reason, we had an interrupt storm that made
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2531) * it all the way around the buffer, bail, and warn
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2532) * about it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2533) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2534) if (unlikely(next_page == commit_page)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2535) local_inc(&cpu_buffer->commit_overrun);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2536) goto out_reset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2537) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2538)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2539) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2540) * This is where the fun begins!
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2541) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2542) * We are fighting against races between a reader that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2543) * could be on another CPU trying to swap its reader
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2544) * page with the buffer head.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2545) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2546) * We are also fighting against interrupts coming in and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2547) * moving the head or tail on us as well.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2548) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2549) * If the next page is the head page then we have filled
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2550) * the buffer, unless the commit page is still on the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2551) * reader page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2552) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2553) if (rb_is_head_page(cpu_buffer, next_page, &tail_page->list)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2554)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2555) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2556) * If the commit is not on the reader page, then
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2557) * move the header page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2558) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2559) if (!rb_is_reader_page(cpu_buffer->commit_page)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2560) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2561) * If we are not in overwrite mode,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2562) * this is easy, just stop here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2563) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2564) if (!(buffer->flags & RB_FL_OVERWRITE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2565) local_inc(&cpu_buffer->dropped_events);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2566) goto out_reset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2567) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2568)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2569) ret = rb_handle_head_page(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2570) tail_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2571) next_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2572) if (ret < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2573) goto out_reset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2574) if (ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2575) goto out_again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2576) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2577) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2578) * We need to be careful here too. The
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2579) * commit page could still be on the reader
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2580) * page. We could have a small buffer, and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2581) * have filled up the buffer with events
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2582) * from interrupts and such, and wrapped.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2583) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2584) * Note, if the tail page is also the on the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2585) * reader_page, we let it move out.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2586) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2587) if (unlikely((cpu_buffer->commit_page !=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2588) cpu_buffer->tail_page) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2589) (cpu_buffer->commit_page ==
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2590) cpu_buffer->reader_page))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2591) local_inc(&cpu_buffer->commit_overrun);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2592) goto out_reset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2593) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2594) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2595) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2596)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2597) rb_tail_page_update(cpu_buffer, tail_page, next_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2598)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2599) out_again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2600)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2601) rb_reset_tail(cpu_buffer, tail, info);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2602)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2603) /* Commit what we have for now. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2604) rb_end_commit(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2605) /* rb_end_commit() decs committing */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2606) local_inc(&cpu_buffer->committing);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2607)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2608) /* fail and let the caller try again */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2609) return ERR_PTR(-EAGAIN);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2610)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2611) out_reset:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2612) /* reset write */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2613) rb_reset_tail(cpu_buffer, tail, info);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2614)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2615) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2616) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2617)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2618) /* Slow path */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2619) static struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2620) rb_add_time_stamp(struct ring_buffer_event *event, u64 delta, bool abs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2621) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2622) if (abs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2623) event->type_len = RINGBUF_TYPE_TIME_STAMP;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2624) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2625) event->type_len = RINGBUF_TYPE_TIME_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2626)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2627) /* Not the first event on the page, or not delta? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2628) if (abs || rb_event_index(event)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2629) event->time_delta = delta & TS_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2630) event->array[0] = delta >> TS_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2631) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2632) /* nope, just zero it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2633) event->time_delta = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2634) event->array[0] = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2635) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2636)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2637) return skip_time_extend(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2638) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2639)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2640) static inline bool rb_event_is_commit(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2641) struct ring_buffer_event *event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2642)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2643) #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2644) static inline bool sched_clock_stable(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2645) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2646) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2647) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2648) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2649)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2650) static void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2651) rb_check_timestamp(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2652) struct rb_event_info *info)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2653) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2654) u64 write_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2655)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2656) WARN_ONCE(1, "Delta way too big! %llu ts=%llu before=%llu after=%llu write stamp=%llu\n%s",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2657) (unsigned long long)info->delta,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2658) (unsigned long long)info->ts,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2659) (unsigned long long)info->before,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2660) (unsigned long long)info->after,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2661) (unsigned long long)(rb_time_read(&cpu_buffer->write_stamp, &write_stamp) ? write_stamp : 0),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2662) sched_clock_stable() ? "" :
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2663) "If you just came from a suspend/resume,\n"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2664) "please switch to the trace global clock:\n"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2665) " echo global > /sys/kernel/debug/tracing/trace_clock\n"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2666) "or add trace_clock=global to the kernel command line\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2667) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2668)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2669) static void rb_add_timestamp(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2670) struct ring_buffer_event **event,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2671) struct rb_event_info *info,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2672) u64 *delta,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2673) unsigned int *length)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2674) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2675) bool abs = info->add_timestamp &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2676) (RB_ADD_STAMP_FORCE | RB_ADD_STAMP_ABSOLUTE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2677)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2678) if (unlikely(info->delta > (1ULL << 59))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2679) /* did the clock go backwards */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2680) if (info->before == info->after && info->before > info->ts) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2681) /* not interrupted */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2682) static int once;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2683)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2684) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2685) * This is possible with a recalibrating of the TSC.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2686) * Do not produce a call stack, but just report it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2687) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2688) if (!once) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2689) once++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2690) pr_warn("Ring buffer clock went backwards: %llu -> %llu\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2691) info->before, info->ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2692) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2693) } else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2694) rb_check_timestamp(cpu_buffer, info);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2695) if (!abs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2696) info->delta = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2697) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2698) *event = rb_add_time_stamp(*event, info->delta, abs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2699) *length -= RB_LEN_TIME_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2700) *delta = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2701) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2702)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2703) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2704) * rb_update_event - update event type and data
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2705) * @cpu_buffer: The per cpu buffer of the @event
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2706) * @event: the event to update
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2707) * @info: The info to update the @event with (contains length and delta)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2708) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2709) * Update the type and data fields of the @event. The length
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2710) * is the actual size that is written to the ring buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2711) * and with this, we can determine what to place into the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2712) * data field.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2713) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2714) static void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2715) rb_update_event(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2716) struct ring_buffer_event *event,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2717) struct rb_event_info *info)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2718) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2719) unsigned length = info->length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2720) u64 delta = info->delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2721)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2722) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2723) * If we need to add a timestamp, then we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2724) * add it to the start of the reserved space.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2725) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2726) if (unlikely(info->add_timestamp))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2727) rb_add_timestamp(cpu_buffer, &event, info, &delta, &length);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2728)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2729) event->time_delta = delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2730) length -= RB_EVNT_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2731) if (length > RB_MAX_SMALL_DATA || RB_FORCE_8BYTE_ALIGNMENT) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2732) event->type_len = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2733) event->array[0] = length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2734) } else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2735) event->type_len = DIV_ROUND_UP(length, RB_ALIGNMENT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2736) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2737)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2738) static unsigned rb_calculate_event_length(unsigned length)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2739) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2740) struct ring_buffer_event event; /* Used only for sizeof array */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2741)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2742) /* zero length can cause confusions */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2743) if (!length)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2744) length++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2745)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2746) if (length > RB_MAX_SMALL_DATA || RB_FORCE_8BYTE_ALIGNMENT)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2747) length += sizeof(event.array[0]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2748)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2749) length += RB_EVNT_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2750) length = ALIGN(length, RB_ARCH_ALIGNMENT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2751)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2752) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2753) * In case the time delta is larger than the 27 bits for it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2754) * in the header, we need to add a timestamp. If another
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2755) * event comes in when trying to discard this one to increase
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2756) * the length, then the timestamp will be added in the allocated
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2757) * space of this event. If length is bigger than the size needed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2758) * for the TIME_EXTEND, then padding has to be used. The events
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2759) * length must be either RB_LEN_TIME_EXTEND, or greater than or equal
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2760) * to RB_LEN_TIME_EXTEND + 8, as 8 is the minimum size for padding.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2761) * As length is a multiple of 4, we only need to worry if it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2762) * is 12 (RB_LEN_TIME_EXTEND + 4).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2763) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2764) if (length == RB_LEN_TIME_EXTEND + RB_ALIGNMENT)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2765) length += RB_ALIGNMENT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2766)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2767) return length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2768) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2769)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2770) static __always_inline bool
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2771) rb_event_is_commit(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2772) struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2773) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2774) unsigned long addr = (unsigned long)event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2775) unsigned long index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2776)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2777) index = rb_event_index(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2778) addr &= PAGE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2779)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2780) return cpu_buffer->commit_page->page == (void *)addr &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2781) rb_commit_index(cpu_buffer) == index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2782) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2783)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2784) static u64 rb_time_delta(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2785) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2786) switch (event->type_len) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2787) case RINGBUF_TYPE_PADDING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2788) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2789)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2790) case RINGBUF_TYPE_TIME_EXTEND:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2791) return ring_buffer_event_time_stamp(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2792)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2793) case RINGBUF_TYPE_TIME_STAMP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2794) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2795)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2796) case RINGBUF_TYPE_DATA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2797) return event->time_delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2798) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2799) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2800) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2801) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2802)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2803) static inline int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2804) rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2805) struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2806) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2807) unsigned long new_index, old_index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2808) struct buffer_page *bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2809) unsigned long index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2810) unsigned long addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2811) u64 write_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2812) u64 delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2813)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2814) new_index = rb_event_index(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2815) old_index = new_index + rb_event_ts_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2816) addr = (unsigned long)event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2817) addr &= PAGE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2818)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2819) bpage = READ_ONCE(cpu_buffer->tail_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2820)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2821) delta = rb_time_delta(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2822)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2823) if (!rb_time_read(&cpu_buffer->write_stamp, &write_stamp))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2824) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2825)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2826) /* Make sure the write stamp is read before testing the location */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2827) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2828)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2829) if (bpage->page == (void *)addr && rb_page_write(bpage) == old_index) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2830) unsigned long write_mask =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2831) local_read(&bpage->write) & ~RB_WRITE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2832) unsigned long event_length = rb_event_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2833)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2834) /* Something came in, can't discard */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2835) if (!rb_time_cmpxchg(&cpu_buffer->write_stamp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2836) write_stamp, write_stamp - delta))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2837) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2838)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2839) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2840) * It's possible that the event time delta is zero
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2841) * (has the same time stamp as the previous event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2842) * in which case write_stamp and before_stamp could
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2843) * be the same. In such a case, force before_stamp
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2844) * to be different than write_stamp. It doesn't
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2845) * matter what it is, as long as its different.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2846) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2847) if (!delta)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2848) rb_time_set(&cpu_buffer->before_stamp, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2849)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2850) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2851) * If an event were to come in now, it would see that the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2852) * write_stamp and the before_stamp are different, and assume
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2853) * that this event just added itself before updating
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2854) * the write stamp. The interrupting event will fix the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2855) * write stamp for us, and use the before stamp as its delta.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2856) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2857)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2858) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2859) * This is on the tail page. It is possible that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2860) * a write could come in and move the tail page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2861) * and write to the next page. That is fine
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2862) * because we just shorten what is on this page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2863) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2864) old_index += write_mask;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2865) new_index += write_mask;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2866) index = local_cmpxchg(&bpage->write, old_index, new_index);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2867) if (index == old_index) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2868) /* update counters */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2869) local_sub(event_length, &cpu_buffer->entries_bytes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2870) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2871) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2872) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2873)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2874) /* could not discard */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2875) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2876) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2877)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2878) static void rb_start_commit(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2879) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2880) local_inc(&cpu_buffer->committing);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2881) local_inc(&cpu_buffer->commits);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2882) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2883)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2884) static __always_inline void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2885) rb_set_commit_to_write(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2886) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2887) unsigned long max_count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2888)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2889) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2890) * We only race with interrupts and NMIs on this CPU.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2891) * If we own the commit event, then we can commit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2892) * all others that interrupted us, since the interruptions
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2893) * are in stack format (they finish before they come
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2894) * back to us). This allows us to do a simple loop to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2895) * assign the commit to the tail.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2896) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2897) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2898) max_count = cpu_buffer->nr_pages * 100;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2899)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2900) while (cpu_buffer->commit_page != READ_ONCE(cpu_buffer->tail_page)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2901) if (RB_WARN_ON(cpu_buffer, !(--max_count)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2902) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2903) if (RB_WARN_ON(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2904) rb_is_reader_page(cpu_buffer->tail_page)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2905) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2906) local_set(&cpu_buffer->commit_page->page->commit,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2907) rb_page_write(cpu_buffer->commit_page));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2908) rb_inc_page(cpu_buffer, &cpu_buffer->commit_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2909) /* add barrier to keep gcc from optimizing too much */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2910) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2911) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2912) while (rb_commit_index(cpu_buffer) !=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2913) rb_page_write(cpu_buffer->commit_page)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2914)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2915) local_set(&cpu_buffer->commit_page->page->commit,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2916) rb_page_write(cpu_buffer->commit_page));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2917) RB_WARN_ON(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2918) local_read(&cpu_buffer->commit_page->page->commit) &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2919) ~RB_WRITE_MASK);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2920) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2921) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2922)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2923) /* again, keep gcc from optimizing */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2924) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2925)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2926) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2927) * If an interrupt came in just after the first while loop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2928) * and pushed the tail page forward, we will be left with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2929) * a dangling commit that will never go forward.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2930) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2931) if (unlikely(cpu_buffer->commit_page != READ_ONCE(cpu_buffer->tail_page)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2932) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2933) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2934)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2935) static __always_inline void rb_end_commit(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2936) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2937) unsigned long commits;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2938)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2939) if (RB_WARN_ON(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2940) !local_read(&cpu_buffer->committing)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2941) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2942)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2943) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2944) commits = local_read(&cpu_buffer->commits);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2945) /* synchronize with interrupts */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2946) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2947) if (local_read(&cpu_buffer->committing) == 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2948) rb_set_commit_to_write(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2949)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2950) local_dec(&cpu_buffer->committing);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2951)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2952) /* synchronize with interrupts */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2953) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2954)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2955) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2956) * Need to account for interrupts coming in between the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2957) * updating of the commit page and the clearing of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2958) * committing counter.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2959) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2960) if (unlikely(local_read(&cpu_buffer->commits) != commits) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2961) !local_read(&cpu_buffer->committing)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2962) local_inc(&cpu_buffer->committing);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2963) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2964) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2965) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2966)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2967) static inline void rb_event_discard(struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2968) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2969) if (extended_time(event))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2970) event = skip_time_extend(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2971)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2972) /* array[0] holds the actual length for the discarded event */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2973) event->array[0] = rb_event_data_length(event) - RB_EVNT_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2974) event->type_len = RINGBUF_TYPE_PADDING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2975) /* time delta must be non zero */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2976) if (!event->time_delta)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2977) event->time_delta = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2978) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2979)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2980) static void rb_commit(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2981) struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2982) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2983) local_inc(&cpu_buffer->entries);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2984) rb_end_commit(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2985) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2986)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2987) static __always_inline void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2988) rb_wakeups(struct trace_buffer *buffer, struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2989) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2990) size_t nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2991) size_t dirty;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2992) size_t full;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2993)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2994) if (buffer->irq_work.waiters_pending) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2995) buffer->irq_work.waiters_pending = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2996) /* irq_work_queue() supplies it's own memory barriers */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2997) irq_work_queue(&buffer->irq_work.work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2998) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2999)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3000) if (cpu_buffer->irq_work.waiters_pending) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3001) cpu_buffer->irq_work.waiters_pending = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3002) /* irq_work_queue() supplies it's own memory barriers */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3003) irq_work_queue(&cpu_buffer->irq_work.work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3004) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3005)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3006) if (cpu_buffer->last_pages_touch == local_read(&cpu_buffer->pages_touched))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3007) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3008)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3009) if (cpu_buffer->reader_page == cpu_buffer->commit_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3010) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3011)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3012) if (!cpu_buffer->irq_work.full_waiters_pending)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3013) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3014)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3015) cpu_buffer->last_pages_touch = local_read(&cpu_buffer->pages_touched);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3016)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3017) full = cpu_buffer->shortest_full;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3018) nr_pages = cpu_buffer->nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3019) dirty = ring_buffer_nr_dirty_pages(buffer, cpu_buffer->cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3020) if (full && nr_pages && (dirty * 100) <= full * nr_pages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3021) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3022)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3023) cpu_buffer->irq_work.wakeup_full = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3024) cpu_buffer->irq_work.full_waiters_pending = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3025) /* irq_work_queue() supplies it's own memory barriers */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3026) irq_work_queue(&cpu_buffer->irq_work.work);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3027) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3028)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3029) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3030) * The lock and unlock are done within a preempt disable section.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3031) * The current_context per_cpu variable can only be modified
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3032) * by the current task between lock and unlock. But it can
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3033) * be modified more than once via an interrupt. To pass this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3034) * information from the lock to the unlock without having to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3035) * access the 'in_interrupt()' functions again (which do show
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3036) * a bit of overhead in something as critical as function tracing,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3037) * we use a bitmask trick.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3038) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3039) * bit 1 = NMI context
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3040) * bit 2 = IRQ context
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3041) * bit 3 = SoftIRQ context
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3042) * bit 4 = normal context.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3043) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3044) * This works because this is the order of contexts that can
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3045) * preempt other contexts. A SoftIRQ never preempts an IRQ
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3046) * context.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3047) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3048) * When the context is determined, the corresponding bit is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3049) * checked and set (if it was set, then a recursion of that context
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3050) * happened).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3051) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3052) * On unlock, we need to clear this bit. To do so, just subtract
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3053) * 1 from the current_context and AND it to itself.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3054) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3055) * (binary)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3056) * 101 - 1 = 100
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3057) * 101 & 100 = 100 (clearing bit zero)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3058) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3059) * 1010 - 1 = 1001
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3060) * 1010 & 1001 = 1000 (clearing bit 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3061) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3062) * The least significant bit can be cleared this way, and it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3063) * just so happens that it is the same bit corresponding to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3064) * the current context.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3065) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3066) * Now the TRANSITION bit breaks the above slightly. The TRANSITION bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3067) * is set when a recursion is detected at the current context, and if
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3068) * the TRANSITION bit is already set, it will fail the recursion.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3069) * This is needed because there's a lag between the changing of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3070) * interrupt context and updating the preempt count. In this case,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3071) * a false positive will be found. To handle this, one extra recursion
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3072) * is allowed, and this is done by the TRANSITION bit. If the TRANSITION
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3073) * bit is already set, then it is considered a recursion and the function
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3074) * ends. Otherwise, the TRANSITION bit is set, and that bit is returned.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3075) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3076) * On the trace_recursive_unlock(), the TRANSITION bit will be the first
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3077) * to be cleared. Even if it wasn't the context that set it. That is,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3078) * if an interrupt comes in while NORMAL bit is set and the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3079) * is called before preempt_count() is updated, since the check will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3080) * be on the NORMAL bit, the TRANSITION bit will then be set. If an
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3081) * NMI then comes in, it will set the NMI bit, but when the NMI code
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3082) * does the trace_recursive_unlock() it will clear the TRANSTION bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3083) * and leave the NMI bit set. But this is fine, because the interrupt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3084) * code that set the TRANSITION bit will then clear the NMI bit when it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3085) * calls trace_recursive_unlock(). If another NMI comes in, it will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3086) * set the TRANSITION bit and continue.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3087) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3088) * Note: The TRANSITION bit only handles a single transition between context.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3089) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3090)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3091) static __always_inline int
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3092) trace_recursive_lock(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3093) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3094) unsigned int val = cpu_buffer->current_context;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3095) unsigned long pc = preempt_count();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3096) int bit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3097)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3098) if (!(pc & (NMI_MASK | HARDIRQ_MASK | SOFTIRQ_OFFSET)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3099) bit = RB_CTX_NORMAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3100) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3101) bit = pc & NMI_MASK ? RB_CTX_NMI :
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3102) pc & HARDIRQ_MASK ? RB_CTX_IRQ : RB_CTX_SOFTIRQ;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3103)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3104) if (unlikely(val & (1 << (bit + cpu_buffer->nest)))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3105) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3106) * It is possible that this was called by transitioning
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3107) * between interrupt context, and preempt_count() has not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3108) * been updated yet. In this case, use the TRANSITION bit.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3109) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3110) bit = RB_CTX_TRANSITION;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3111) if (val & (1 << (bit + cpu_buffer->nest)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3112) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3113) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3114)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3115) val |= (1 << (bit + cpu_buffer->nest));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3116) cpu_buffer->current_context = val;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3117)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3118) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3119) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3120)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3121) static __always_inline void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3122) trace_recursive_unlock(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3123) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3124) cpu_buffer->current_context &=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3125) cpu_buffer->current_context - (1 << cpu_buffer->nest);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3126) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3127)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3128) /* The recursive locking above uses 5 bits */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3129) #define NESTED_BITS 5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3130)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3131) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3132) * ring_buffer_nest_start - Allow to trace while nested
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3133) * @buffer: The ring buffer to modify
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3134) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3135) * The ring buffer has a safety mechanism to prevent recursion.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3136) * But there may be a case where a trace needs to be done while
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3137) * tracing something else. In this case, calling this function
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3138) * will allow this function to nest within a currently active
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3139) * ring_buffer_lock_reserve().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3140) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3141) * Call this function before calling another ring_buffer_lock_reserve() and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3142) * call ring_buffer_nest_end() after the nested ring_buffer_unlock_commit().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3143) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3144) void ring_buffer_nest_start(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3145) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3146) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3147) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3148)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3149) /* Enabled by ring_buffer_nest_end() */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3150) preempt_disable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3151) cpu = raw_smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3152) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3153) /* This is the shift value for the above recursive locking */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3154) cpu_buffer->nest += NESTED_BITS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3155) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3156)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3157) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3158) * ring_buffer_nest_end - Allow to trace while nested
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3159) * @buffer: The ring buffer to modify
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3160) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3161) * Must be called after ring_buffer_nest_start() and after the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3162) * ring_buffer_unlock_commit().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3163) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3164) void ring_buffer_nest_end(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3165) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3166) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3167) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3168)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3169) /* disabled by ring_buffer_nest_start() */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3170) cpu = raw_smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3171) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3172) /* This is the shift value for the above recursive locking */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3173) cpu_buffer->nest -= NESTED_BITS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3174) preempt_enable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3175) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3176)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3177) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3178) * ring_buffer_unlock_commit - commit a reserved
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3179) * @buffer: The buffer to commit to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3180) * @event: The event pointer to commit.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3181) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3182) * This commits the data to the ring buffer, and releases any locks held.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3183) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3184) * Must be paired with ring_buffer_lock_reserve.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3185) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3186) int ring_buffer_unlock_commit(struct trace_buffer *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3187) struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3188) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3189) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3190) int cpu = raw_smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3191)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3192) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3193)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3194) rb_commit(cpu_buffer, event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3196) rb_wakeups(buffer, cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3197)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3198) trace_recursive_unlock(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3199)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3200) preempt_enable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3201)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3202) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3203) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3204) EXPORT_SYMBOL_GPL(ring_buffer_unlock_commit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3205)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3206) static struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3207) __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3208) struct rb_event_info *info)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3209) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3210) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3211) struct buffer_page *tail_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3212) unsigned long tail, write, w;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3213) bool a_ok;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3214) bool b_ok;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3215)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3216) /* Don't let the compiler play games with cpu_buffer->tail_page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3217) tail_page = info->tail_page = READ_ONCE(cpu_buffer->tail_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3218)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3219) /*A*/ w = local_read(&tail_page->write) & RB_WRITE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3220) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3221) b_ok = rb_time_read(&cpu_buffer->before_stamp, &info->before);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3222) a_ok = rb_time_read(&cpu_buffer->write_stamp, &info->after);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3223) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3224) info->ts = rb_time_stamp(cpu_buffer->buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3225)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3226) if ((info->add_timestamp & RB_ADD_STAMP_ABSOLUTE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3227) info->delta = info->ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3228) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3229) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3230) * If interrupting an event time update, we may need an
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3231) * absolute timestamp.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3232) * Don't bother if this is the start of a new page (w == 0).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3233) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3234) if (unlikely(!a_ok || !b_ok || (info->before != info->after && w))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3235) info->add_timestamp |= RB_ADD_STAMP_FORCE | RB_ADD_STAMP_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3236) info->length += RB_LEN_TIME_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3237) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3238) info->delta = info->ts - info->after;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3239) if (unlikely(test_time_stamp(info->delta))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3240) info->add_timestamp |= RB_ADD_STAMP_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3241) info->length += RB_LEN_TIME_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3242) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3243) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3244) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3245)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3246) /*B*/ rb_time_set(&cpu_buffer->before_stamp, info->ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3247)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3248) /*C*/ write = local_add_return(info->length, &tail_page->write);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3249)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3250) /* set write to only the index of the write */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3251) write &= RB_WRITE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3252)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3253) tail = write - info->length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3254)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3255) /* See if we shot pass the end of this buffer page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3256) if (unlikely(write > BUF_PAGE_SIZE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3257) /* before and after may now different, fix it up*/
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3258) b_ok = rb_time_read(&cpu_buffer->before_stamp, &info->before);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3259) a_ok = rb_time_read(&cpu_buffer->write_stamp, &info->after);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3260) if (a_ok && b_ok && info->before != info->after)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3261) (void)rb_time_cmpxchg(&cpu_buffer->before_stamp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3262) info->before, info->after);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3263) return rb_move_tail(cpu_buffer, tail, info);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3264) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3265)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3266) if (likely(tail == w)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3267) u64 save_before;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3268) bool s_ok;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3269)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3270) /* Nothing interrupted us between A and C */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3271) /*D*/ rb_time_set(&cpu_buffer->write_stamp, info->ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3272) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3273) /*E*/ s_ok = rb_time_read(&cpu_buffer->before_stamp, &save_before);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3274) RB_WARN_ON(cpu_buffer, !s_ok);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3275) if (likely(!(info->add_timestamp &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3276) (RB_ADD_STAMP_FORCE | RB_ADD_STAMP_ABSOLUTE))))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3277) /* This did not interrupt any time update */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3278) info->delta = info->ts - info->after;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3279) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3280) /* Just use full timestamp for inerrupting event */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3281) info->delta = info->ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3282) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3283) if (unlikely(info->ts != save_before)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3284) /* SLOW PATH - Interrupted between C and E */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3285)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3286) a_ok = rb_time_read(&cpu_buffer->write_stamp, &info->after);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3287) RB_WARN_ON(cpu_buffer, !a_ok);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3288)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3289) /* Write stamp must only go forward */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3290) if (save_before > info->after) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3291) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3292) * We do not care about the result, only that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3293) * it gets updated atomically.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3294) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3295) (void)rb_time_cmpxchg(&cpu_buffer->write_stamp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3296) info->after, save_before);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3297) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3298) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3299) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3300) u64 ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3301) /* SLOW PATH - Interrupted between A and C */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3302) a_ok = rb_time_read(&cpu_buffer->write_stamp, &info->after);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3303) /* Was interrupted before here, write_stamp must be valid */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3304) RB_WARN_ON(cpu_buffer, !a_ok);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3305) ts = rb_time_stamp(cpu_buffer->buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3306) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3307) /*E*/ if (write == (local_read(&tail_page->write) & RB_WRITE_MASK) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3308) info->after < ts &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3309) rb_time_cmpxchg(&cpu_buffer->write_stamp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3310) info->after, ts)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3311) /* Nothing came after this event between C and E */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3312) info->delta = ts - info->after;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3313) info->ts = ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3314) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3315) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3316) * Interrupted beween C and E:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3317) * Lost the previous events time stamp. Just set the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3318) * delta to zero, and this will be the same time as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3319) * the event this event interrupted. And the events that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3320) * came after this will still be correct (as they would
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3321) * have built their delta on the previous event.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3322) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3323) info->delta = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3324) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3325) info->add_timestamp &= ~RB_ADD_STAMP_FORCE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3326) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3327)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3328) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3329) * If this is the first commit on the page, then it has the same
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3330) * timestamp as the page itself.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3331) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3332) if (unlikely(!tail && !(info->add_timestamp &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3333) (RB_ADD_STAMP_FORCE | RB_ADD_STAMP_ABSOLUTE))))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3334) info->delta = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3335)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3336) /* We reserved something on the buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3337)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3338) event = __rb_page_index(tail_page, tail);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3339) rb_update_event(cpu_buffer, event, info);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3340)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3341) local_inc(&tail_page->entries);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3342)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3343) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3344) * If this is the first commit on the page, then update
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3345) * its timestamp.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3346) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3347) if (unlikely(!tail))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3348) tail_page->page->time_stamp = info->ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3349)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3350) /* account for these added bytes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3351) local_add(info->length, &cpu_buffer->entries_bytes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3352)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3353) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3354) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3355)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3356) static __always_inline struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3357) rb_reserve_next_event(struct trace_buffer *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3358) struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3359) unsigned long length)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3360) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3361) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3362) struct rb_event_info info;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3363) int nr_loops = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3364) int add_ts_default;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3365)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3366) rb_start_commit(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3367) /* The commit page can not change after this */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3368)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3369) #ifdef CONFIG_RING_BUFFER_ALLOW_SWAP
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3370) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3371) * Due to the ability to swap a cpu buffer from a buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3372) * it is possible it was swapped before we committed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3373) * (committing stops a swap). We check for it here and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3374) * if it happened, we have to fail the write.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3375) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3376) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3377) if (unlikely(READ_ONCE(cpu_buffer->buffer) != buffer)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3378) local_dec(&cpu_buffer->committing);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3379) local_dec(&cpu_buffer->commits);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3380) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3381) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3382) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3383)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3384) info.length = rb_calculate_event_length(length);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3385)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3386) if (ring_buffer_time_stamp_abs(cpu_buffer->buffer)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3387) add_ts_default = RB_ADD_STAMP_ABSOLUTE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3388) info.length += RB_LEN_TIME_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3389) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3390) add_ts_default = RB_ADD_STAMP_NONE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3391) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3392)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3393) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3394) info.add_timestamp = add_ts_default;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3395) info.delta = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3396)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3397) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3398) * We allow for interrupts to reenter here and do a trace.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3399) * If one does, it will cause this original code to loop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3400) * back here. Even with heavy interrupts happening, this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3401) * should only happen a few times in a row. If this happens
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3402) * 1000 times in a row, there must be either an interrupt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3403) * storm or we have something buggy.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3404) * Bail!
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3405) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3406) if (RB_WARN_ON(cpu_buffer, ++nr_loops > 1000))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3407) goto out_fail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3408)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3409) event = __rb_reserve_next(cpu_buffer, &info);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3410)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3411) if (unlikely(PTR_ERR(event) == -EAGAIN)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3412) if (info.add_timestamp & (RB_ADD_STAMP_FORCE | RB_ADD_STAMP_EXTEND))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3413) info.length -= RB_LEN_TIME_EXTEND;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3414) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3415) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3416)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3417) if (likely(event))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3418) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3419) out_fail:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3420) rb_end_commit(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3421) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3422) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3423)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3424) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3425) * ring_buffer_lock_reserve - reserve a part of the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3426) * @buffer: the ring buffer to reserve from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3427) * @length: the length of the data to reserve (excluding event header)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3428) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3429) * Returns a reserved event on the ring buffer to copy directly to.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3430) * The user of this interface will need to get the body to write into
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3431) * and can use the ring_buffer_event_data() interface.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3432) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3433) * The length is the length of the data needed, not the event length
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3434) * which also includes the event header.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3435) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3436) * Must be paired with ring_buffer_unlock_commit, unless NULL is returned.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3437) * If NULL is returned, then nothing has been allocated or locked.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3438) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3439) struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3440) ring_buffer_lock_reserve(struct trace_buffer *buffer, unsigned long length)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3441) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3442) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3443) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3444) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3445)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3446) /* If we are tracing schedule, we don't want to recurse */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3447) preempt_disable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3448)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3449) if (unlikely(atomic_read(&buffer->record_disabled)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3450) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3451)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3452) cpu = raw_smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3453)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3454) if (unlikely(!cpumask_test_cpu(cpu, buffer->cpumask)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3455) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3456)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3457) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3458)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3459) if (unlikely(atomic_read(&cpu_buffer->record_disabled)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3460) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3461)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3462) if (unlikely(length > BUF_MAX_DATA_SIZE))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3463) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3464)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3465) if (unlikely(trace_recursive_lock(cpu_buffer)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3466) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3467)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3468) event = rb_reserve_next_event(buffer, cpu_buffer, length);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3469) if (!event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3470) goto out_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3471)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3472) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3473)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3474) out_unlock:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3475) trace_recursive_unlock(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3476) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3477) preempt_enable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3478) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3479) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3480) EXPORT_SYMBOL_GPL(ring_buffer_lock_reserve);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3481)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3482) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3483) * Decrement the entries to the page that an event is on.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3484) * The event does not even need to exist, only the pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3485) * to the page it is on. This may only be called before the commit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3486) * takes place.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3487) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3488) static inline void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3489) rb_decrement_entry(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3490) struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3491) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3492) unsigned long addr = (unsigned long)event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3493) struct buffer_page *bpage = cpu_buffer->commit_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3494) struct buffer_page *start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3495)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3496) addr &= PAGE_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3497)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3498) /* Do the likely case first */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3499) if (likely(bpage->page == (void *)addr)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3500) local_dec(&bpage->entries);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3501) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3502) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3503)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3504) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3505) * Because the commit page may be on the reader page we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3506) * start with the next page and check the end loop there.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3507) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3508) rb_inc_page(cpu_buffer, &bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3509) start = bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3510) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3511) if (bpage->page == (void *)addr) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3512) local_dec(&bpage->entries);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3513) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3514) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3515) rb_inc_page(cpu_buffer, &bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3516) } while (bpage != start);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3517)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3518) /* commit not part of this buffer?? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3519) RB_WARN_ON(cpu_buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3520) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3521)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3522) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3523) * ring_buffer_commit_discard - discard an event that has not been committed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3524) * @buffer: the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3525) * @event: non committed event to discard
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3526) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3527) * Sometimes an event that is in the ring buffer needs to be ignored.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3528) * This function lets the user discard an event in the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3529) * and then that event will not be read later.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3530) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3531) * This function only works if it is called before the item has been
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3532) * committed. It will try to free the event from the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3533) * if another event has not been added behind it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3534) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3535) * If another event has been added behind it, it will set the event
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3536) * up as discarded, and perform the commit.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3537) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3538) * If this function is called, do not call ring_buffer_unlock_commit on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3539) * the event.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3540) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3541) void ring_buffer_discard_commit(struct trace_buffer *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3542) struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3543) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3544) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3545) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3546)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3547) /* The event is discarded regardless */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3548) rb_event_discard(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3549)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3550) cpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3551) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3552)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3553) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3554) * This must only be called if the event has not been
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3555) * committed yet. Thus we can assume that preemption
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3556) * is still disabled.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3557) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3558) RB_WARN_ON(buffer, !local_read(&cpu_buffer->committing));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3559)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3560) rb_decrement_entry(cpu_buffer, event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3561) if (rb_try_to_discard(cpu_buffer, event))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3562) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3563)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3564) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3565) rb_end_commit(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3566)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3567) trace_recursive_unlock(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3568)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3569) preempt_enable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3570)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3571) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3572) EXPORT_SYMBOL_GPL(ring_buffer_discard_commit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3573)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3574) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3575) * ring_buffer_write - write data to the buffer without reserving
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3576) * @buffer: The ring buffer to write to.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3577) * @length: The length of the data being written (excluding the event header)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3578) * @data: The data to write to the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3579) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3580) * This is like ring_buffer_lock_reserve and ring_buffer_unlock_commit as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3581) * one function. If you already have the data to write to the buffer, it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3582) * may be easier to simply call this function.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3583) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3584) * Note, like ring_buffer_lock_reserve, the length is the length of the data
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3585) * and not the length of the event which would hold the header.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3586) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3587) int ring_buffer_write(struct trace_buffer *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3588) unsigned long length,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3589) void *data)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3590) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3591) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3592) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3593) void *body;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3594) int ret = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3595) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3596)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3597) preempt_disable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3598)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3599) if (atomic_read(&buffer->record_disabled))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3600) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3601)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3602) cpu = raw_smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3603)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3604) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3605) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3606)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3607) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3608)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3609) if (atomic_read(&cpu_buffer->record_disabled))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3610) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3611)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3612) if (length > BUF_MAX_DATA_SIZE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3613) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3614)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3615) if (unlikely(trace_recursive_lock(cpu_buffer)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3616) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3617)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3618) event = rb_reserve_next_event(buffer, cpu_buffer, length);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3619) if (!event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3620) goto out_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3621)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3622) body = rb_event_data(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3623)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3624) memcpy(body, data, length);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3625)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3626) rb_commit(cpu_buffer, event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3627)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3628) rb_wakeups(buffer, cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3629)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3630) ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3631)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3632) out_unlock:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3633) trace_recursive_unlock(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3634)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3635) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3636) preempt_enable_notrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3637)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3638) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3639) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3640) EXPORT_SYMBOL_GPL(ring_buffer_write);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3641)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3642) static bool rb_per_cpu_empty(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3643) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3644) struct buffer_page *reader = cpu_buffer->reader_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3645) struct buffer_page *head = rb_set_head_page(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3646) struct buffer_page *commit = cpu_buffer->commit_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3647)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3648) /* In case of error, head will be NULL */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3649) if (unlikely(!head))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3650) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3651)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3652) /* Reader should exhaust content in reader page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3653) if (reader->read != rb_page_commit(reader))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3654) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3655)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3656) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3657) * If writers are committing on the reader page, knowing all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3658) * committed content has been read, the ring buffer is empty.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3659) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3660) if (commit == reader)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3661) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3662)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3663) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3664) * If writers are committing on a page other than reader page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3665) * and head page, there should always be content to read.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3666) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3667) if (commit != head)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3668) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3669)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3670) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3671) * Writers are committing on the head page, we just need
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3672) * to care about there're committed data, and the reader will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3673) * swap reader page with head page when it is to read data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3674) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3675) return rb_page_commit(commit) == 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3676) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3677)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3678) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3679) * ring_buffer_record_disable - stop all writes into the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3680) * @buffer: The ring buffer to stop writes to.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3681) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3682) * This prevents all writes to the buffer. Any attempt to write
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3683) * to the buffer after this will fail and return NULL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3684) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3685) * The caller should call synchronize_rcu() after this.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3686) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3687) void ring_buffer_record_disable(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3688) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3689) atomic_inc(&buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3690) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3691) EXPORT_SYMBOL_GPL(ring_buffer_record_disable);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3692)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3693) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3694) * ring_buffer_record_enable - enable writes to the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3695) * @buffer: The ring buffer to enable writes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3696) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3697) * Note, multiple disables will need the same number of enables
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3698) * to truly enable the writing (much like preempt_disable).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3699) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3700) void ring_buffer_record_enable(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3701) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3702) atomic_dec(&buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3703) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3704) EXPORT_SYMBOL_GPL(ring_buffer_record_enable);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3705)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3706) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3707) * ring_buffer_record_off - stop all writes into the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3708) * @buffer: The ring buffer to stop writes to.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3709) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3710) * This prevents all writes to the buffer. Any attempt to write
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3711) * to the buffer after this will fail and return NULL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3712) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3713) * This is different than ring_buffer_record_disable() as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3714) * it works like an on/off switch, where as the disable() version
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3715) * must be paired with a enable().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3716) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3717) void ring_buffer_record_off(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3718) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3719) unsigned int rd;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3720) unsigned int new_rd;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3721)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3722) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3723) rd = atomic_read(&buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3724) new_rd = rd | RB_BUFFER_OFF;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3725) } while (atomic_cmpxchg(&buffer->record_disabled, rd, new_rd) != rd);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3726) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3727) EXPORT_SYMBOL_GPL(ring_buffer_record_off);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3728)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3729) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3730) * ring_buffer_record_on - restart writes into the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3731) * @buffer: The ring buffer to start writes to.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3732) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3733) * This enables all writes to the buffer that was disabled by
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3734) * ring_buffer_record_off().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3735) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3736) * This is different than ring_buffer_record_enable() as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3737) * it works like an on/off switch, where as the enable() version
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3738) * must be paired with a disable().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3739) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3740) void ring_buffer_record_on(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3741) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3742) unsigned int rd;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3743) unsigned int new_rd;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3744)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3745) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3746) rd = atomic_read(&buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3747) new_rd = rd & ~RB_BUFFER_OFF;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3748) } while (atomic_cmpxchg(&buffer->record_disabled, rd, new_rd) != rd);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3749) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3750) EXPORT_SYMBOL_GPL(ring_buffer_record_on);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3751)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3752) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3753) * ring_buffer_record_is_on - return true if the ring buffer can write
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3754) * @buffer: The ring buffer to see if write is enabled
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3755) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3756) * Returns true if the ring buffer is in a state that it accepts writes.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3757) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3758) bool ring_buffer_record_is_on(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3759) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3760) return !atomic_read(&buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3761) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3762)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3763) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3764) * ring_buffer_record_is_set_on - return true if the ring buffer is set writable
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3765) * @buffer: The ring buffer to see if write is set enabled
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3766) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3767) * Returns true if the ring buffer is set writable by ring_buffer_record_on().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3768) * Note that this does NOT mean it is in a writable state.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3769) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3770) * It may return true when the ring buffer has been disabled by
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3771) * ring_buffer_record_disable(), as that is a temporary disabling of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3772) * the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3773) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3774) bool ring_buffer_record_is_set_on(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3775) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3776) return !(atomic_read(&buffer->record_disabled) & RB_BUFFER_OFF);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3777) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3778)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3779) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3780) * ring_buffer_record_disable_cpu - stop all writes into the cpu_buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3781) * @buffer: The ring buffer to stop writes to.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3782) * @cpu: The CPU buffer to stop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3783) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3784) * This prevents all writes to the buffer. Any attempt to write
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3785) * to the buffer after this will fail and return NULL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3786) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3787) * The caller should call synchronize_rcu() after this.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3788) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3789) void ring_buffer_record_disable_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3790) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3791) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3792)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3793) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3794) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3795)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3796) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3797) atomic_inc(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3798) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3799) EXPORT_SYMBOL_GPL(ring_buffer_record_disable_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3800)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3801) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3802) * ring_buffer_record_enable_cpu - enable writes to the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3803) * @buffer: The ring buffer to enable writes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3804) * @cpu: The CPU to enable.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3805) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3806) * Note, multiple disables will need the same number of enables
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3807) * to truly enable the writing (much like preempt_disable).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3808) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3809) void ring_buffer_record_enable_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3810) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3811) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3812)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3813) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3814) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3815)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3816) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3817) atomic_dec(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3818) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3819) EXPORT_SYMBOL_GPL(ring_buffer_record_enable_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3820)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3821) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3822) * The total entries in the ring buffer is the running counter
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3823) * of entries entered into the ring buffer, minus the sum of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3824) * the entries read from the ring buffer and the number of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3825) * entries that were overwritten.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3826) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3827) static inline unsigned long
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3828) rb_num_of_entries(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3829) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3830) return local_read(&cpu_buffer->entries) -
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3831) (local_read(&cpu_buffer->overrun) + cpu_buffer->read);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3832) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3833)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3834) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3835) * ring_buffer_oldest_event_ts - get the oldest event timestamp from the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3836) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3837) * @cpu: The per CPU buffer to read from.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3838) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3839) u64 ring_buffer_oldest_event_ts(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3840) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3841) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3842) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3843) struct buffer_page *bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3844) u64 ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3845)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3846) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3847) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3848)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3849) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3850) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3851) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3852) * if the tail is on reader_page, oldest time stamp is on the reader
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3853) * page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3854) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3855) if (cpu_buffer->tail_page == cpu_buffer->reader_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3856) bpage = cpu_buffer->reader_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3857) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3858) bpage = rb_set_head_page(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3859) if (bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3860) ret = bpage->page->time_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3861) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3862)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3863) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3864) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3865) EXPORT_SYMBOL_GPL(ring_buffer_oldest_event_ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3866)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3867) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3868) * ring_buffer_bytes_cpu - get the number of bytes consumed in a cpu buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3869) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3870) * @cpu: The per CPU buffer to read from.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3871) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3872) unsigned long ring_buffer_bytes_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3873) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3874) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3875) unsigned long ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3876)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3877) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3878) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3879)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3880) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3881) ret = local_read(&cpu_buffer->entries_bytes) - cpu_buffer->read_bytes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3882)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3883) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3884) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3885) EXPORT_SYMBOL_GPL(ring_buffer_bytes_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3886)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3887) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3888) * ring_buffer_entries_cpu - get the number of entries in a cpu buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3889) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3890) * @cpu: The per CPU buffer to get the entries from.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3891) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3892) unsigned long ring_buffer_entries_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3893) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3894) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3895)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3896) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3897) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3898)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3899) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3900)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3901) return rb_num_of_entries(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3902) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3903) EXPORT_SYMBOL_GPL(ring_buffer_entries_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3904)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3905) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3906) * ring_buffer_overrun_cpu - get the number of overruns caused by the ring
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3907) * buffer wrapping around (only if RB_FL_OVERWRITE is on).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3908) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3909) * @cpu: The per CPU buffer to get the number of overruns from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3910) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3911) unsigned long ring_buffer_overrun_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3912) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3913) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3914) unsigned long ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3915)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3916) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3917) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3918)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3919) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3920) ret = local_read(&cpu_buffer->overrun);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3921)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3922) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3923) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3924) EXPORT_SYMBOL_GPL(ring_buffer_overrun_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3925)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3926) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3927) * ring_buffer_commit_overrun_cpu - get the number of overruns caused by
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3928) * commits failing due to the buffer wrapping around while there are uncommitted
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3929) * events, such as during an interrupt storm.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3930) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3931) * @cpu: The per CPU buffer to get the number of overruns from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3932) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3933) unsigned long
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3934) ring_buffer_commit_overrun_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3935) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3936) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3937) unsigned long ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3938)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3939) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3940) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3941)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3942) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3943) ret = local_read(&cpu_buffer->commit_overrun);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3944)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3945) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3946) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3947) EXPORT_SYMBOL_GPL(ring_buffer_commit_overrun_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3948)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3949) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3950) * ring_buffer_dropped_events_cpu - get the number of dropped events caused by
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3951) * the ring buffer filling up (only if RB_FL_OVERWRITE is off).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3952) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3953) * @cpu: The per CPU buffer to get the number of overruns from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3954) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3955) unsigned long
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3956) ring_buffer_dropped_events_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3957) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3958) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3959) unsigned long ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3960)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3961) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3962) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3963)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3964) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3965) ret = local_read(&cpu_buffer->dropped_events);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3966)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3967) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3968) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3969) EXPORT_SYMBOL_GPL(ring_buffer_dropped_events_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3970)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3971) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3972) * ring_buffer_read_events_cpu - get the number of events successfully read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3973) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3974) * @cpu: The per CPU buffer to get the number of events read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3975) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3976) unsigned long
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3977) ring_buffer_read_events_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3978) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3979) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3980)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3981) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3982) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3983)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3984) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3985) return cpu_buffer->read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3986) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3987) EXPORT_SYMBOL_GPL(ring_buffer_read_events_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3988)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3989) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3990) * ring_buffer_entries - get the number of entries in a buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3991) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3992) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3993) * Returns the total number of entries in the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3994) * (all CPU entries)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3995) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3996) unsigned long ring_buffer_entries(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3997) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3998) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3999) unsigned long entries = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4000) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4001)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4002) /* if you care about this being correct, lock the buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4003) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4004) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4005) entries += rb_num_of_entries(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4006) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4007)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4008) return entries;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4009) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4010) EXPORT_SYMBOL_GPL(ring_buffer_entries);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4011)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4012) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4013) * ring_buffer_overruns - get the number of overruns in buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4014) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4015) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4016) * Returns the total number of overruns in the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4017) * (all CPU entries)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4018) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4019) unsigned long ring_buffer_overruns(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4020) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4021) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4022) unsigned long overruns = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4023) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4024)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4025) /* if you care about this being correct, lock the buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4026) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4027) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4028) overruns += local_read(&cpu_buffer->overrun);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4029) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4030)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4031) return overruns;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4032) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4033) EXPORT_SYMBOL_GPL(ring_buffer_overruns);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4034)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4035) static void rb_iter_reset(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4036) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4037) struct ring_buffer_per_cpu *cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4038)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4039) /* Iterator usage is expected to have record disabled */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4040) iter->head_page = cpu_buffer->reader_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4041) iter->head = cpu_buffer->reader_page->read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4042) iter->next_event = iter->head;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4043)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4044) iter->cache_reader_page = iter->head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4045) iter->cache_read = cpu_buffer->read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4046)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4047) if (iter->head) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4048) iter->read_stamp = cpu_buffer->read_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4049) iter->page_stamp = cpu_buffer->reader_page->page->time_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4050) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4051) iter->read_stamp = iter->head_page->page->time_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4052) iter->page_stamp = iter->read_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4053) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4054) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4055)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4056) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4057) * ring_buffer_iter_reset - reset an iterator
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4058) * @iter: The iterator to reset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4059) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4060) * Resets the iterator, so that it will start from the beginning
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4061) * again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4062) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4063) void ring_buffer_iter_reset(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4064) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4065) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4066) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4067)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4068) if (!iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4069) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4070)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4071) cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4072)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4073) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4074) rb_iter_reset(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4075) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4076) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4077) EXPORT_SYMBOL_GPL(ring_buffer_iter_reset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4078)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4079) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4080) * ring_buffer_iter_empty - check if an iterator has no more to read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4081) * @iter: The iterator to check
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4082) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4083) int ring_buffer_iter_empty(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4084) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4085) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4086) struct buffer_page *reader;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4087) struct buffer_page *head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4088) struct buffer_page *commit_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4089) struct buffer_page *curr_commit_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4090) unsigned commit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4091) u64 curr_commit_ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4092) u64 commit_ts;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4093)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4094) cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4095) reader = cpu_buffer->reader_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4096) head_page = cpu_buffer->head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4097) commit_page = cpu_buffer->commit_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4098) commit_ts = commit_page->page->time_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4099)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4100) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4101) * When the writer goes across pages, it issues a cmpxchg which
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4102) * is a mb(), which will synchronize with the rmb here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4103) * (see rb_tail_page_update())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4104) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4105) smp_rmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4106) commit = rb_page_commit(commit_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4107) /* We want to make sure that the commit page doesn't change */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4108) smp_rmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4109)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4110) /* Make sure commit page didn't change */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4111) curr_commit_page = READ_ONCE(cpu_buffer->commit_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4112) curr_commit_ts = READ_ONCE(curr_commit_page->page->time_stamp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4113)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4114) /* If the commit page changed, then there's more data */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4115) if (curr_commit_page != commit_page ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4116) curr_commit_ts != commit_ts)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4117) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4118)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4119) /* Still racy, as it may return a false positive, but that's OK */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4120) return ((iter->head_page == commit_page && iter->head >= commit) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4121) (iter->head_page == reader && commit_page == head_page &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4122) head_page->read == commit &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4123) iter->head == rb_page_commit(cpu_buffer->reader_page)));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4124) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4125) EXPORT_SYMBOL_GPL(ring_buffer_iter_empty);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4126)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4127) static void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4128) rb_update_read_stamp(struct ring_buffer_per_cpu *cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4129) struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4130) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4131) u64 delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4132)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4133) switch (event->type_len) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4134) case RINGBUF_TYPE_PADDING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4135) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4136)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4137) case RINGBUF_TYPE_TIME_EXTEND:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4138) delta = ring_buffer_event_time_stamp(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4139) cpu_buffer->read_stamp += delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4140) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4141)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4142) case RINGBUF_TYPE_TIME_STAMP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4143) delta = ring_buffer_event_time_stamp(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4144) cpu_buffer->read_stamp = delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4145) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4146)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4147) case RINGBUF_TYPE_DATA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4148) cpu_buffer->read_stamp += event->time_delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4149) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4150)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4151) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4152) RB_WARN_ON(cpu_buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4153) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4154) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4155) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4156)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4157) static void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4158) rb_update_iter_read_stamp(struct ring_buffer_iter *iter,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4159) struct ring_buffer_event *event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4160) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4161) u64 delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4162)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4163) switch (event->type_len) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4164) case RINGBUF_TYPE_PADDING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4165) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4166)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4167) case RINGBUF_TYPE_TIME_EXTEND:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4168) delta = ring_buffer_event_time_stamp(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4169) iter->read_stamp += delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4170) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4171)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4172) case RINGBUF_TYPE_TIME_STAMP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4173) delta = ring_buffer_event_time_stamp(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4174) iter->read_stamp = delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4175) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4176)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4177) case RINGBUF_TYPE_DATA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4178) iter->read_stamp += event->time_delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4179) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4180)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4181) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4182) RB_WARN_ON(iter->cpu_buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4183) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4184) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4185) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4186)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4187) static struct buffer_page *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4188) rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4189) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4190) struct buffer_page *reader = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4191) unsigned long overwrite;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4192) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4193) int nr_loops = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4194) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4196) local_irq_save(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4197) arch_spin_lock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4198)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4199) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4200) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4201) * This should normally only loop twice. But because the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4202) * start of the reader inserts an empty page, it causes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4203) * a case where we will loop three times. There should be no
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4204) * reason to loop four times (that I know of).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4205) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4206) if (RB_WARN_ON(cpu_buffer, ++nr_loops > 3)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4207) reader = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4208) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4209) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4210)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4211) reader = cpu_buffer->reader_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4212)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4213) /* If there's more to read, return this page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4214) if (cpu_buffer->reader_page->read < rb_page_size(reader))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4215) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4216)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4217) /* Never should we have an index greater than the size */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4218) if (RB_WARN_ON(cpu_buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4219) cpu_buffer->reader_page->read > rb_page_size(reader)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4220) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4221)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4222) /* check if we caught up to the tail */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4223) reader = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4224) if (cpu_buffer->commit_page == cpu_buffer->reader_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4225) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4226)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4227) /* Don't bother swapping if the ring buffer is empty */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4228) if (rb_num_of_entries(cpu_buffer) == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4229) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4230)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4231) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4232) * Reset the reader page to size zero.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4233) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4234) local_set(&cpu_buffer->reader_page->write, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4235) local_set(&cpu_buffer->reader_page->entries, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4236) local_set(&cpu_buffer->reader_page->page->commit, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4237) cpu_buffer->reader_page->real_end = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4238)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4239) spin:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4240) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4241) * Splice the empty reader page into the list around the head.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4242) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4243) reader = rb_set_head_page(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4244) if (!reader)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4245) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4246) cpu_buffer->reader_page->list.next = rb_list_head(reader->list.next);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4247) cpu_buffer->reader_page->list.prev = reader->list.prev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4248)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4249) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4250) * cpu_buffer->pages just needs to point to the buffer, it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4251) * has no specific buffer page to point to. Lets move it out
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4252) * of our way so we don't accidentally swap it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4253) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4254) cpu_buffer->pages = reader->list.prev;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4255)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4256) /* The reader page will be pointing to the new head */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4257) rb_set_list_to_head(cpu_buffer, &cpu_buffer->reader_page->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4258)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4259) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4260) * We want to make sure we read the overruns after we set up our
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4261) * pointers to the next object. The writer side does a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4262) * cmpxchg to cross pages which acts as the mb on the writer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4263) * side. Note, the reader will constantly fail the swap
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4264) * while the writer is updating the pointers, so this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4265) * guarantees that the overwrite recorded here is the one we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4266) * want to compare with the last_overrun.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4267) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4268) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4269) overwrite = local_read(&(cpu_buffer->overrun));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4270)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4271) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4272) * Here's the tricky part.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4273) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4274) * We need to move the pointer past the header page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4275) * But we can only do that if a writer is not currently
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4276) * moving it. The page before the header page has the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4277) * flag bit '1' set if it is pointing to the page we want.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4278) * but if the writer is in the process of moving it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4279) * than it will be '2' or already moved '0'.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4280) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4281)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4282) ret = rb_head_page_replace(reader, cpu_buffer->reader_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4283)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4284) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4285) * If we did not convert it, then we must try again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4286) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4287) if (!ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4288) goto spin;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4289)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4290) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4291) * Yay! We succeeded in replacing the page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4292) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4293) * Now make the new head point back to the reader page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4294) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4295) rb_list_head(reader->list.next)->prev = &cpu_buffer->reader_page->list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4296) rb_inc_page(cpu_buffer, &cpu_buffer->head_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4297)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4298) local_inc(&cpu_buffer->pages_read);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4299)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4300) /* Finally update the reader page to the new head */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4301) cpu_buffer->reader_page = reader;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4302) cpu_buffer->reader_page->read = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4303)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4304) if (overwrite != cpu_buffer->last_overrun) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4305) cpu_buffer->lost_events = overwrite - cpu_buffer->last_overrun;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4306) cpu_buffer->last_overrun = overwrite;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4307) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4308)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4309) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4310)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4311) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4312) /* Update the read_stamp on the first event */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4313) if (reader && reader->read == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4314) cpu_buffer->read_stamp = reader->page->time_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4315)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4316) arch_spin_unlock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4317) local_irq_restore(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4318)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4319) return reader;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4320) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4321)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4322) static void rb_advance_reader(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4323) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4324) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4325) struct buffer_page *reader;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4326) unsigned length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4327)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4328) reader = rb_get_reader_page(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4329)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4330) /* This function should not be called when buffer is empty */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4331) if (RB_WARN_ON(cpu_buffer, !reader))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4332) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4333)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4334) event = rb_reader_event(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4335)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4336) if (event->type_len <= RINGBUF_TYPE_DATA_TYPE_LEN_MAX)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4337) cpu_buffer->read++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4338)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4339) rb_update_read_stamp(cpu_buffer, event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4340)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4341) length = rb_event_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4342) cpu_buffer->reader_page->read += length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4343) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4344)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4345) static void rb_advance_iter(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4346) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4347) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4348)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4349) cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4350)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4351) /* If head == next_event then we need to jump to the next event */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4352) if (iter->head == iter->next_event) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4353) /* If the event gets overwritten again, there's nothing to do */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4354) if (rb_iter_head_event(iter) == NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4355) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4356) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4357)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4358) iter->head = iter->next_event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4359)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4360) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4361) * Check if we are at the end of the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4362) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4363) if (iter->next_event >= rb_page_size(iter->head_page)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4364) /* discarded commits can make the page empty */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4365) if (iter->head_page == cpu_buffer->commit_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4366) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4367) rb_inc_iter(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4368) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4369) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4370)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4371) rb_update_iter_read_stamp(iter, iter->event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4372) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4373)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4374) static int rb_lost_events(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4375) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4376) return cpu_buffer->lost_events;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4377) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4378)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4379) static struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4380) rb_buffer_peek(struct ring_buffer_per_cpu *cpu_buffer, u64 *ts,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4381) unsigned long *lost_events)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4382) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4383) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4384) struct buffer_page *reader;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4385) int nr_loops = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4386)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4387) if (ts)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4388) *ts = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4389) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4390) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4391) * We repeat when a time extend is encountered.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4392) * Since the time extend is always attached to a data event,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4393) * we should never loop more than once.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4394) * (We never hit the following condition more than twice).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4395) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4396) if (RB_WARN_ON(cpu_buffer, ++nr_loops > 2))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4397) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4398)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4399) reader = rb_get_reader_page(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4400) if (!reader)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4401) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4402)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4403) event = rb_reader_event(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4404)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4405) switch (event->type_len) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4406) case RINGBUF_TYPE_PADDING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4407) if (rb_null_event(event))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4408) RB_WARN_ON(cpu_buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4409) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4410) * Because the writer could be discarding every
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4411) * event it creates (which would probably be bad)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4412) * if we were to go back to "again" then we may never
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4413) * catch up, and will trigger the warn on, or lock
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4414) * the box. Return the padding, and we will release
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4415) * the current locks, and try again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4416) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4417) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4418)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4419) case RINGBUF_TYPE_TIME_EXTEND:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4420) /* Internal data, OK to advance */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4421) rb_advance_reader(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4422) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4423)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4424) case RINGBUF_TYPE_TIME_STAMP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4425) if (ts) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4426) *ts = ring_buffer_event_time_stamp(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4427) ring_buffer_normalize_time_stamp(cpu_buffer->buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4428) cpu_buffer->cpu, ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4429) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4430) /* Internal data, OK to advance */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4431) rb_advance_reader(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4432) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4433)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4434) case RINGBUF_TYPE_DATA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4435) if (ts && !(*ts)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4436) *ts = cpu_buffer->read_stamp + event->time_delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4437) ring_buffer_normalize_time_stamp(cpu_buffer->buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4438) cpu_buffer->cpu, ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4439) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4440) if (lost_events)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4441) *lost_events = rb_lost_events(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4442) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4443)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4444) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4445) RB_WARN_ON(cpu_buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4446) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4447)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4448) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4449) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4450) EXPORT_SYMBOL_GPL(ring_buffer_peek);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4451)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4452) static struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4453) rb_iter_peek(struct ring_buffer_iter *iter, u64 *ts)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4454) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4455) struct trace_buffer *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4456) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4457) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4458) int nr_loops = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4459)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4460) if (ts)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4461) *ts = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4462)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4463) cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4464) buffer = cpu_buffer->buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4465)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4466) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4467) * Check if someone performed a consuming read to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4468) * the buffer. A consuming read invalidates the iterator
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4469) * and we need to reset the iterator in this case.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4470) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4471) if (unlikely(iter->cache_read != cpu_buffer->read ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4472) iter->cache_reader_page != cpu_buffer->reader_page))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4473) rb_iter_reset(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4474)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4475) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4476) if (ring_buffer_iter_empty(iter))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4477) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4478)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4479) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4480) * As the writer can mess with what the iterator is trying
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4481) * to read, just give up if we fail to get an event after
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4482) * three tries. The iterator is not as reliable when reading
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4483) * the ring buffer with an active write as the consumer is.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4484) * Do not warn if the three failures is reached.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4485) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4486) if (++nr_loops > 3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4487) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4488)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4489) if (rb_per_cpu_empty(cpu_buffer))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4490) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4491)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4492) if (iter->head >= rb_page_size(iter->head_page)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4493) rb_inc_iter(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4494) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4495) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4496)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4497) event = rb_iter_head_event(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4498) if (!event)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4499) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4500)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4501) switch (event->type_len) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4502) case RINGBUF_TYPE_PADDING:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4503) if (rb_null_event(event)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4504) rb_inc_iter(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4505) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4506) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4507) rb_advance_iter(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4508) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4509)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4510) case RINGBUF_TYPE_TIME_EXTEND:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4511) /* Internal data, OK to advance */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4512) rb_advance_iter(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4513) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4514)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4515) case RINGBUF_TYPE_TIME_STAMP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4516) if (ts) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4517) *ts = ring_buffer_event_time_stamp(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4518) ring_buffer_normalize_time_stamp(cpu_buffer->buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4519) cpu_buffer->cpu, ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4520) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4521) /* Internal data, OK to advance */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4522) rb_advance_iter(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4523) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4524)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4525) case RINGBUF_TYPE_DATA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4526) if (ts && !(*ts)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4527) *ts = iter->read_stamp + event->time_delta;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4528) ring_buffer_normalize_time_stamp(buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4529) cpu_buffer->cpu, ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4530) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4531) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4532)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4533) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4534) RB_WARN_ON(cpu_buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4535) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4536)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4537) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4538) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4539) EXPORT_SYMBOL_GPL(ring_buffer_iter_peek);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4540)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4541) static inline bool rb_reader_lock(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4542) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4543) if (likely(!in_nmi())) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4544) raw_spin_lock(&cpu_buffer->reader_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4545) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4546) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4547)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4548) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4549) * If an NMI die dumps out the content of the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4550) * trylock must be used to prevent a deadlock if the NMI
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4551) * preempted a task that holds the ring buffer locks. If
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4552) * we get the lock then all is fine, if not, then continue
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4553) * to do the read, but this can corrupt the ring buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4554) * so it must be permanently disabled from future writes.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4555) * Reading from NMI is a oneshot deal.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4556) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4557) if (raw_spin_trylock(&cpu_buffer->reader_lock))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4558) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4559)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4560) /* Continue without locking, but disable the ring buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4561) atomic_inc(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4562) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4563) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4564)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4565) static inline void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4566) rb_reader_unlock(struct ring_buffer_per_cpu *cpu_buffer, bool locked)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4567) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4568) if (likely(locked))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4569) raw_spin_unlock(&cpu_buffer->reader_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4570) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4571) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4572)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4573) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4574) * ring_buffer_peek - peek at the next event to be read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4575) * @buffer: The ring buffer to read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4576) * @cpu: The cpu to peak at
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4577) * @ts: The timestamp counter of this event.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4578) * @lost_events: a variable to store if events were lost (may be NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4579) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4580) * This will return the event that will be read next, but does
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4581) * not consume the data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4582) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4583) struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4584) ring_buffer_peek(struct trace_buffer *buffer, int cpu, u64 *ts,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4585) unsigned long *lost_events)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4586) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4587) struct ring_buffer_per_cpu *cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4588) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4589) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4590) bool dolock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4591)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4592) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4593) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4594)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4595) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4596) local_irq_save(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4597) dolock = rb_reader_lock(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4598) event = rb_buffer_peek(cpu_buffer, ts, lost_events);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4599) if (event && event->type_len == RINGBUF_TYPE_PADDING)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4600) rb_advance_reader(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4601) rb_reader_unlock(cpu_buffer, dolock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4602) local_irq_restore(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4603)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4604) if (event && event->type_len == RINGBUF_TYPE_PADDING)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4605) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4606)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4607) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4608) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4609)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4610) /** ring_buffer_iter_dropped - report if there are dropped events
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4611) * @iter: The ring buffer iterator
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4612) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4613) * Returns true if there was dropped events since the last peek.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4614) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4615) bool ring_buffer_iter_dropped(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4616) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4617) bool ret = iter->missed_events != 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4618)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4619) iter->missed_events = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4620) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4621) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4622) EXPORT_SYMBOL_GPL(ring_buffer_iter_dropped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4623)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4624) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4625) * ring_buffer_iter_peek - peek at the next event to be read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4626) * @iter: The ring buffer iterator
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4627) * @ts: The timestamp counter of this event.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4628) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4629) * This will return the event that will be read next, but does
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4630) * not increment the iterator.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4631) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4632) struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4633) ring_buffer_iter_peek(struct ring_buffer_iter *iter, u64 *ts)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4634) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4635) struct ring_buffer_per_cpu *cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4636) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4637) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4638)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4639) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4640) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4641) event = rb_iter_peek(iter, ts);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4642) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4643)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4644) if (event && event->type_len == RINGBUF_TYPE_PADDING)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4645) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4646)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4647) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4648) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4649)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4650) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4651) * ring_buffer_consume - return an event and consume it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4652) * @buffer: The ring buffer to get the next event from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4653) * @cpu: the cpu to read the buffer from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4654) * @ts: a variable to store the timestamp (may be NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4655) * @lost_events: a variable to store if events were lost (may be NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4656) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4657) * Returns the next event in the ring buffer, and that event is consumed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4658) * Meaning, that sequential reads will keep returning a different event,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4659) * and eventually empty the ring buffer if the producer is slower.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4660) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4661) struct ring_buffer_event *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4662) ring_buffer_consume(struct trace_buffer *buffer, int cpu, u64 *ts,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4663) unsigned long *lost_events)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4664) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4665) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4666) struct ring_buffer_event *event = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4667) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4668) bool dolock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4669)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4670) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4671) /* might be called in atomic */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4672) preempt_disable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4673)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4674) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4675) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4676)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4677) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4678) local_irq_save(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4679) dolock = rb_reader_lock(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4680)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4681) event = rb_buffer_peek(cpu_buffer, ts, lost_events);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4682) if (event) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4683) cpu_buffer->lost_events = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4684) rb_advance_reader(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4685) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4686)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4687) rb_reader_unlock(cpu_buffer, dolock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4688) local_irq_restore(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4689)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4690) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4691) preempt_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4692)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4693) if (event && event->type_len == RINGBUF_TYPE_PADDING)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4694) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4695)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4696) return event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4697) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4698) EXPORT_SYMBOL_GPL(ring_buffer_consume);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4699)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4700) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4701) * ring_buffer_read_prepare - Prepare for a non consuming read of the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4702) * @buffer: The ring buffer to read from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4703) * @cpu: The cpu buffer to iterate over
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4704) * @flags: gfp flags to use for memory allocation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4705) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4706) * This performs the initial preparations necessary to iterate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4707) * through the buffer. Memory is allocated, buffer recording
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4708) * is disabled, and the iterator pointer is returned to the caller.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4709) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4710) * Disabling buffer recording prevents the reading from being
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4711) * corrupted. This is not a consuming read, so a producer is not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4712) * expected.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4713) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4714) * After a sequence of ring_buffer_read_prepare calls, the user is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4715) * expected to make at least one call to ring_buffer_read_prepare_sync.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4716) * Afterwards, ring_buffer_read_start is invoked to get things going
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4717) * for real.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4718) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4719) * This overall must be paired with ring_buffer_read_finish.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4720) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4721) struct ring_buffer_iter *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4722) ring_buffer_read_prepare(struct trace_buffer *buffer, int cpu, gfp_t flags)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4723) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4724) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4725) struct ring_buffer_iter *iter;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4726)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4727) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4728) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4729)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4730) iter = kzalloc(sizeof(*iter), flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4731) if (!iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4732) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4733)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4734) iter->event = kmalloc(BUF_MAX_DATA_SIZE, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4735) if (!iter->event) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4736) kfree(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4737) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4738) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4739)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4740) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4741)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4742) iter->cpu_buffer = cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4743)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4744) atomic_inc(&cpu_buffer->resize_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4745)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4746) return iter;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4747) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4748) EXPORT_SYMBOL_GPL(ring_buffer_read_prepare);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4749)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4750) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4751) * ring_buffer_read_prepare_sync - Synchronize a set of prepare calls
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4752) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4753) * All previously invoked ring_buffer_read_prepare calls to prepare
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4754) * iterators will be synchronized. Afterwards, read_buffer_read_start
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4755) * calls on those iterators are allowed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4756) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4757) void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4758) ring_buffer_read_prepare_sync(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4759) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4760) synchronize_rcu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4761) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4762) EXPORT_SYMBOL_GPL(ring_buffer_read_prepare_sync);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4763)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4764) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4765) * ring_buffer_read_start - start a non consuming read of the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4766) * @iter: The iterator returned by ring_buffer_read_prepare
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4767) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4768) * This finalizes the startup of an iteration through the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4769) * The iterator comes from a call to ring_buffer_read_prepare and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4770) * an intervening ring_buffer_read_prepare_sync must have been
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4771) * performed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4772) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4773) * Must be paired with ring_buffer_read_finish.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4774) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4775) void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4776) ring_buffer_read_start(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4777) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4778) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4779) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4780)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4781) if (!iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4782) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4783)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4784) cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4785)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4786) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4787) arch_spin_lock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4788) rb_iter_reset(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4789) arch_spin_unlock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4790) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4791) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4792) EXPORT_SYMBOL_GPL(ring_buffer_read_start);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4793)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4794) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4795) * ring_buffer_read_finish - finish reading the iterator of the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4796) * @iter: The iterator retrieved by ring_buffer_start
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4797) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4798) * This re-enables the recording to the buffer, and frees the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4799) * iterator.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4800) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4801) void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4802) ring_buffer_read_finish(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4803) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4804) struct ring_buffer_per_cpu *cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4805) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4806)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4807) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4808) * Ring buffer is disabled from recording, here's a good place
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4809) * to check the integrity of the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4810) * Must prevent readers from trying to read, as the check
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4811) * clears the HEAD page and readers require it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4812) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4813) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4814) rb_check_pages(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4815) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4816)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4817) atomic_dec(&cpu_buffer->resize_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4818) kfree(iter->event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4819) kfree(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4820) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4821) EXPORT_SYMBOL_GPL(ring_buffer_read_finish);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4822)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4823) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4824) * ring_buffer_iter_advance - advance the iterator to the next location
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4825) * @iter: The ring buffer iterator
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4826) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4827) * Move the location of the iterator such that the next read will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4828) * be the next location of the iterator.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4829) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4830) void ring_buffer_iter_advance(struct ring_buffer_iter *iter)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4831) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4832) struct ring_buffer_per_cpu *cpu_buffer = iter->cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4833) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4834)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4835) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4836)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4837) rb_advance_iter(iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4838)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4839) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4840) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4841) EXPORT_SYMBOL_GPL(ring_buffer_iter_advance);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4842)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4843) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4844) * ring_buffer_size - return the size of the ring buffer (in bytes)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4845) * @buffer: The ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4846) * @cpu: The CPU to get ring buffer size from.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4847) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4848) unsigned long ring_buffer_size(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4849) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4850) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4851) * Earlier, this method returned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4852) * BUF_PAGE_SIZE * buffer->nr_pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4853) * Since the nr_pages field is now removed, we have converted this to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4854) * return the per cpu buffer value.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4855) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4856) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4857) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4858)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4859) return BUF_PAGE_SIZE * buffer->buffers[cpu]->nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4860) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4861) EXPORT_SYMBOL_GPL(ring_buffer_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4862)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4863) static void
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4864) rb_reset_cpu(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4865) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4866) rb_head_page_deactivate(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4867)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4868) cpu_buffer->head_page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4869) = list_entry(cpu_buffer->pages, struct buffer_page, list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4870) local_set(&cpu_buffer->head_page->write, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4871) local_set(&cpu_buffer->head_page->entries, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4872) local_set(&cpu_buffer->head_page->page->commit, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4873)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4874) cpu_buffer->head_page->read = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4875)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4876) cpu_buffer->tail_page = cpu_buffer->head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4877) cpu_buffer->commit_page = cpu_buffer->head_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4878)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4879) INIT_LIST_HEAD(&cpu_buffer->reader_page->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4880) INIT_LIST_HEAD(&cpu_buffer->new_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4881) local_set(&cpu_buffer->reader_page->write, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4882) local_set(&cpu_buffer->reader_page->entries, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4883) local_set(&cpu_buffer->reader_page->page->commit, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4884) cpu_buffer->reader_page->read = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4885)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4886) local_set(&cpu_buffer->entries_bytes, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4887) local_set(&cpu_buffer->overrun, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4888) local_set(&cpu_buffer->commit_overrun, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4889) local_set(&cpu_buffer->dropped_events, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4890) local_set(&cpu_buffer->entries, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4891) local_set(&cpu_buffer->committing, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4892) local_set(&cpu_buffer->commits, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4893) local_set(&cpu_buffer->pages_touched, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4894) local_set(&cpu_buffer->pages_read, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4895) cpu_buffer->last_pages_touch = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4896) cpu_buffer->shortest_full = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4897) cpu_buffer->read = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4898) cpu_buffer->read_bytes = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4899)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4900) rb_time_set(&cpu_buffer->write_stamp, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4901) rb_time_set(&cpu_buffer->before_stamp, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4902)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4903) cpu_buffer->lost_events = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4904) cpu_buffer->last_overrun = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4905)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4906) rb_head_page_activate(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4907) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4908)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4909) /* Must have disabled the cpu buffer then done a synchronize_rcu */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4910) static void reset_disabled_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4911) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4912) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4913)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4914) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4915)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4916) if (RB_WARN_ON(cpu_buffer, local_read(&cpu_buffer->committing)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4917) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4918)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4919) arch_spin_lock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4920)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4921) rb_reset_cpu(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4922)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4923) arch_spin_unlock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4924)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4925) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4926) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4927) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4928)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4929) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4930) * ring_buffer_reset_cpu - reset a ring buffer per CPU buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4931) * @buffer: The ring buffer to reset a per cpu buffer of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4932) * @cpu: The CPU buffer to be reset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4933) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4934) void ring_buffer_reset_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4935) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4936) struct ring_buffer_per_cpu *cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4937)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4938) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4939) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4940)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4941) /* prevent another thread from changing buffer sizes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4942) mutex_lock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4943)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4944) atomic_inc(&cpu_buffer->resize_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4945) atomic_inc(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4946)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4947) /* Make sure all commits have finished */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4948) synchronize_rcu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4949)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4950) reset_disabled_cpu_buffer(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4951)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4952) atomic_dec(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4953) atomic_dec(&cpu_buffer->resize_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4954)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4955) mutex_unlock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4956) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4957) EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4958)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4959) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4960) * ring_buffer_reset_cpu - reset a ring buffer per CPU buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4961) * @buffer: The ring buffer to reset a per cpu buffer of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4962) * @cpu: The CPU buffer to be reset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4963) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4964) void ring_buffer_reset_online_cpus(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4965) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4966) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4967) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4968)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4969) /* prevent another thread from changing buffer sizes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4970) mutex_lock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4971)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4972) for_each_online_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4973) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4974)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4975) atomic_inc(&cpu_buffer->resize_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4976) atomic_inc(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4977) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4978)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4979) /* Make sure all commits have finished */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4980) synchronize_rcu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4981)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4982) for_each_online_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4983) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4984)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4985) reset_disabled_cpu_buffer(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4986)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4987) atomic_dec(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4988) atomic_dec(&cpu_buffer->resize_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4989) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4990)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4991) mutex_unlock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4992) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4993)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4994) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4995) * ring_buffer_reset - reset a ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4996) * @buffer: The ring buffer to reset all cpu buffers
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4997) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4998) void ring_buffer_reset(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4999) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5000) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5001) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5002)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5003) /* prevent another thread from changing buffer sizes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5004) mutex_lock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5005)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5006) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5007) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5008)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5009) atomic_inc(&cpu_buffer->resize_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5010) atomic_inc(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5011) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5012)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5013) /* Make sure all commits have finished */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5014) synchronize_rcu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5015)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5016) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5017) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5018)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5019) reset_disabled_cpu_buffer(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5020)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5021) atomic_dec(&cpu_buffer->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5022) atomic_dec(&cpu_buffer->resize_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5023) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5024)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5025) mutex_unlock(&buffer->mutex);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5026) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5027) EXPORT_SYMBOL_GPL(ring_buffer_reset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5028)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5029) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5030) * rind_buffer_empty - is the ring buffer empty?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5031) * @buffer: The ring buffer to test
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5032) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5033) bool ring_buffer_empty(struct trace_buffer *buffer)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5034) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5035) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5036) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5037) bool dolock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5038) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5039) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5040)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5041) /* yes this is racy, but if you don't like the race, lock the buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5042) for_each_buffer_cpu(buffer, cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5043) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5044) local_irq_save(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5045) dolock = rb_reader_lock(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5046) ret = rb_per_cpu_empty(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5047) rb_reader_unlock(cpu_buffer, dolock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5048) local_irq_restore(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5049)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5050) if (!ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5051) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5052) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5053)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5054) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5055) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5056) EXPORT_SYMBOL_GPL(ring_buffer_empty);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5057)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5058) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5059) * ring_buffer_empty_cpu - is a cpu buffer of a ring buffer empty?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5060) * @buffer: The ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5061) * @cpu: The CPU buffer to test
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5062) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5063) bool ring_buffer_empty_cpu(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5064) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5065) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5066) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5067) bool dolock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5068) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5069)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5070) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5071) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5072)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5073) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5074) local_irq_save(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5075) dolock = rb_reader_lock(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5076) ret = rb_per_cpu_empty(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5077) rb_reader_unlock(cpu_buffer, dolock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5078) local_irq_restore(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5079)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5080) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5081) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5082) EXPORT_SYMBOL_GPL(ring_buffer_empty_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5083)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5084) #ifdef CONFIG_RING_BUFFER_ALLOW_SWAP
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5085) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5086) * ring_buffer_swap_cpu - swap a CPU buffer between two ring buffers
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5087) * @buffer_a: One buffer to swap with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5088) * @buffer_b: The other buffer to swap with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5089) * @cpu: the CPU of the buffers to swap
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5090) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5091) * This function is useful for tracers that want to take a "snapshot"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5092) * of a CPU buffer and has another back up buffer lying around.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5093) * it is expected that the tracer handles the cpu buffer not being
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5094) * used at the moment.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5095) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5096) int ring_buffer_swap_cpu(struct trace_buffer *buffer_a,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5097) struct trace_buffer *buffer_b, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5098) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5099) struct ring_buffer_per_cpu *cpu_buffer_a;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5100) struct ring_buffer_per_cpu *cpu_buffer_b;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5101) int ret = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5102)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5103) if (!cpumask_test_cpu(cpu, buffer_a->cpumask) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5104) !cpumask_test_cpu(cpu, buffer_b->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5105) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5106)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5107) cpu_buffer_a = buffer_a->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5108) cpu_buffer_b = buffer_b->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5109)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5110) /* At least make sure the two buffers are somewhat the same */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5111) if (cpu_buffer_a->nr_pages != cpu_buffer_b->nr_pages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5112) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5113)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5114) ret = -EAGAIN;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5115)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5116) if (atomic_read(&buffer_a->record_disabled))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5117) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5118)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5119) if (atomic_read(&buffer_b->record_disabled))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5120) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5121)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5122) if (atomic_read(&cpu_buffer_a->record_disabled))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5123) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5124)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5125) if (atomic_read(&cpu_buffer_b->record_disabled))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5126) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5127)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5128) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5129) * We can't do a synchronize_rcu here because this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5130) * function can be called in atomic context.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5131) * Normally this will be called from the same CPU as cpu.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5132) * If not it's up to the caller to protect this.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5133) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5134) atomic_inc(&cpu_buffer_a->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5135) atomic_inc(&cpu_buffer_b->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5136)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5137) ret = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5138) if (local_read(&cpu_buffer_a->committing))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5139) goto out_dec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5140) if (local_read(&cpu_buffer_b->committing))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5141) goto out_dec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5142)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5143) buffer_a->buffers[cpu] = cpu_buffer_b;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5144) buffer_b->buffers[cpu] = cpu_buffer_a;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5145)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5146) cpu_buffer_b->buffer = buffer_a;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5147) cpu_buffer_a->buffer = buffer_b;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5148)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5149) ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5150)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5151) out_dec:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5152) atomic_dec(&cpu_buffer_a->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5153) atomic_dec(&cpu_buffer_b->record_disabled);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5154) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5155) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5156) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5157) EXPORT_SYMBOL_GPL(ring_buffer_swap_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5158) #endif /* CONFIG_RING_BUFFER_ALLOW_SWAP */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5159)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5160) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5161) * ring_buffer_alloc_read_page - allocate a page to read from buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5162) * @buffer: the buffer to allocate for.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5163) * @cpu: the cpu buffer to allocate.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5164) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5165) * This function is used in conjunction with ring_buffer_read_page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5166) * When reading a full page from the ring buffer, these functions
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5167) * can be used to speed up the process. The calling function should
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5168) * allocate a few pages first with this function. Then when it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5169) * needs to get pages from the ring buffer, it passes the result
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5170) * of this function into ring_buffer_read_page, which will swap
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5171) * the page that was allocated, with the read page of the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5172) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5173) * Returns:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5174) * The page allocated, or ERR_PTR
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5175) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5176) void *ring_buffer_alloc_read_page(struct trace_buffer *buffer, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5177) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5178) struct ring_buffer_per_cpu *cpu_buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5179) struct buffer_data_page *bpage = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5180) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5181) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5182)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5183) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5184) return ERR_PTR(-ENODEV);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5185)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5186) cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5187) local_irq_save(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5188) arch_spin_lock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5189)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5190) if (cpu_buffer->free_page) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5191) bpage = cpu_buffer->free_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5192) cpu_buffer->free_page = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5193) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5194)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5195) arch_spin_unlock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5196) local_irq_restore(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5197)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5198) if (bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5199) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5200)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5201) page = alloc_pages_node(cpu_to_node(cpu),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5202) GFP_KERNEL | __GFP_NORETRY, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5203) if (!page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5204) return ERR_PTR(-ENOMEM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5205)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5206) bpage = page_address(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5207)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5208) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5209) rb_init_page(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5210)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5211) return bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5212) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5213) EXPORT_SYMBOL_GPL(ring_buffer_alloc_read_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5214)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5215) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5216) * ring_buffer_free_read_page - free an allocated read page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5217) * @buffer: the buffer the page was allocate for
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5218) * @cpu: the cpu buffer the page came from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5219) * @data: the page to free
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5220) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5221) * Free a page allocated from ring_buffer_alloc_read_page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5222) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5223) void ring_buffer_free_read_page(struct trace_buffer *buffer, int cpu, void *data)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5224) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5225) struct ring_buffer_per_cpu *cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5226) struct buffer_data_page *bpage = data;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5227) struct page *page = virt_to_page(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5228) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5229)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5230) /* If the page is still in use someplace else, we can't reuse it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5231) if (page_ref_count(page) > 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5232) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5233)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5234) local_irq_save(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5235) arch_spin_lock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5236)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5237) if (!cpu_buffer->free_page) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5238) cpu_buffer->free_page = bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5239) bpage = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5240) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5241)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5242) arch_spin_unlock(&cpu_buffer->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5243) local_irq_restore(flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5244)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5245) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5246) free_page((unsigned long)bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5247) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5248) EXPORT_SYMBOL_GPL(ring_buffer_free_read_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5249)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5250) /**
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5251) * ring_buffer_read_page - extract a page from the ring buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5252) * @buffer: buffer to extract from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5253) * @data_page: the page to use allocated from ring_buffer_alloc_read_page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5254) * @len: amount to extract
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5255) * @cpu: the cpu of the buffer to extract
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5256) * @full: should the extraction only happen when the page is full.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5257) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5258) * This function will pull out a page from the ring buffer and consume it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5259) * @data_page must be the address of the variable that was returned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5260) * from ring_buffer_alloc_read_page. This is because the page might be used
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5261) * to swap with a page in the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5262) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5263) * for example:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5264) * rpage = ring_buffer_alloc_read_page(buffer, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5265) * if (IS_ERR(rpage))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5266) * return PTR_ERR(rpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5267) * ret = ring_buffer_read_page(buffer, &rpage, len, cpu, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5268) * if (ret >= 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5269) * process_page(rpage, ret);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5270) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5271) * When @full is set, the function will not return true unless
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5272) * the writer is off the reader page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5273) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5274) * Note: it is up to the calling functions to handle sleeps and wakeups.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5275) * The ring buffer can be used anywhere in the kernel and can not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5276) * blindly call wake_up. The layer that uses the ring buffer must be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5277) * responsible for that.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5278) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5279) * Returns:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5280) * >=0 if data has been transferred, returns the offset of consumed data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5281) * <0 if no data has been transferred.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5282) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5283) int ring_buffer_read_page(struct trace_buffer *buffer,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5284) void **data_page, size_t len, int cpu, int full)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5285) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5286) struct ring_buffer_per_cpu *cpu_buffer = buffer->buffers[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5287) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5288) struct buffer_data_page *bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5289) struct buffer_page *reader;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5290) unsigned long missed_events;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5291) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5292) unsigned int commit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5293) unsigned int read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5294) u64 save_timestamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5295) int ret = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5296)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5297) if (!cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5298) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5299)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5300) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5301) * If len is not big enough to hold the page header, then
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5302) * we can not copy anything.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5303) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5304) if (len <= BUF_PAGE_HDR_SIZE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5305) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5306)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5307) len -= BUF_PAGE_HDR_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5308)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5309) if (!data_page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5310) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5311)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5312) bpage = *data_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5313) if (!bpage)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5314) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5315)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5316) raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5317)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5318) reader = rb_get_reader_page(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5319) if (!reader)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5320) goto out_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5321)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5322) event = rb_reader_event(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5323)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5324) read = reader->read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5325) commit = rb_page_commit(reader);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5326)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5327) /* Check if any events were dropped */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5328) missed_events = cpu_buffer->lost_events;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5329)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5330) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5331) * If this page has been partially read or
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5332) * if len is not big enough to read the rest of the page or
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5333) * a writer is still on the page, then
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5334) * we must copy the data from the page to the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5335) * Otherwise, we can simply swap the page with the one passed in.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5336) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5337) if (read || (len < (commit - read)) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5338) cpu_buffer->reader_page == cpu_buffer->commit_page) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5339) struct buffer_data_page *rpage = cpu_buffer->reader_page->page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5340) unsigned int rpos = read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5341) unsigned int pos = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5342) unsigned int size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5343)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5344) if (full)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5345) goto out_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5346)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5347) if (len > (commit - read))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5348) len = (commit - read);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5349)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5350) /* Always keep the time extend and data together */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5351) size = rb_event_ts_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5352)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5353) if (len < size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5354) goto out_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5355)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5356) /* save the current timestamp, since the user will need it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5357) save_timestamp = cpu_buffer->read_stamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5358)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5359) /* Need to copy one event at a time */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5360) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5361) /* We need the size of one event, because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5362) * rb_advance_reader only advances by one event,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5363) * whereas rb_event_ts_length may include the size of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5364) * one or two events.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5365) * We have already ensured there's enough space if this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5366) * is a time extend. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5367) size = rb_event_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5368) memcpy(bpage->data + pos, rpage->data + rpos, size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5369)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5370) len -= size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5371)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5372) rb_advance_reader(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5373) rpos = reader->read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5374) pos += size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5375)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5376) if (rpos >= commit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5377) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5378)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5379) event = rb_reader_event(cpu_buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5380) /* Always keep the time extend and data together */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5381) size = rb_event_ts_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5382) } while (len >= size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5383)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5384) /* update bpage */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5385) local_set(&bpage->commit, pos);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5386) bpage->time_stamp = save_timestamp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5387)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5388) /* we copied everything to the beginning */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5389) read = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5390) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5391) /* update the entry counter */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5392) cpu_buffer->read += rb_page_entries(reader);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5393) cpu_buffer->read_bytes += BUF_PAGE_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5394)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5395) /* swap the pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5396) rb_init_page(bpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5397) bpage = reader->page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5398) reader->page = *data_page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5399) local_set(&reader->write, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5400) local_set(&reader->entries, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5401) reader->read = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5402) *data_page = bpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5403)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5404) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5405) * Use the real_end for the data size,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5406) * This gives us a chance to store the lost events
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5407) * on the page.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5408) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5409) if (reader->real_end)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5410) local_set(&bpage->commit, reader->real_end);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5411) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5412) ret = read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5413)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5414) cpu_buffer->lost_events = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5415)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5416) commit = local_read(&bpage->commit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5417) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5418) * Set a flag in the commit field if we lost events
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5419) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5420) if (missed_events) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5421) /* If there is room at the end of the page to save the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5422) * missed events, then record it there.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5423) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5424) if (BUF_PAGE_SIZE - commit >= sizeof(missed_events)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5425) memcpy(&bpage->data[commit], &missed_events,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5426) sizeof(missed_events));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5427) local_add(RB_MISSED_STORED, &bpage->commit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5428) commit += sizeof(missed_events);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5429) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5430) local_add(RB_MISSED_EVENTS, &bpage->commit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5431) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5432)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5433) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5434) * This page may be off to user land. Zero it out here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5435) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5436) if (commit < BUF_PAGE_SIZE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5437) memset(&bpage->data[commit], 0, BUF_PAGE_SIZE - commit);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5438)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5439) out_unlock:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5440) raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5441)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5442) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5443) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5444) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5445) EXPORT_SYMBOL_GPL(ring_buffer_read_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5446)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5447) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5448) * We only allocate new buffers, never free them if the CPU goes down.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5449) * If we were to free the buffer, then the user would lose any trace that was in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5450) * the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5451) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5452) int trace_rb_cpu_prepare(unsigned int cpu, struct hlist_node *node)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5453) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5454) struct trace_buffer *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5455) long nr_pages_same;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5456) int cpu_i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5457) unsigned long nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5458)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5459) buffer = container_of(node, struct trace_buffer, node);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5460) if (cpumask_test_cpu(cpu, buffer->cpumask))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5461) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5462)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5463) nr_pages = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5464) nr_pages_same = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5465) /* check if all cpu sizes are same */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5466) for_each_buffer_cpu(buffer, cpu_i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5467) /* fill in the size from first enabled cpu */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5468) if (nr_pages == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5469) nr_pages = buffer->buffers[cpu_i]->nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5470) if (nr_pages != buffer->buffers[cpu_i]->nr_pages) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5471) nr_pages_same = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5472) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5473) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5474) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5475) /* allocate minimum pages, user can later expand it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5476) if (!nr_pages_same)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5477) nr_pages = 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5478) buffer->buffers[cpu] =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5479) rb_allocate_cpu_buffer(buffer, nr_pages, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5480) if (!buffer->buffers[cpu]) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5481) WARN(1, "failed to allocate ring buffer on CPU %u\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5482) cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5483) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5484) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5485) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5486) cpumask_set_cpu(cpu, buffer->cpumask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5487) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5488) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5489)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5490) #ifdef CONFIG_RING_BUFFER_STARTUP_TEST
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5491) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5492) * This is a basic integrity check of the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5493) * Late in the boot cycle this test will run when configured in.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5494) * It will kick off a thread per CPU that will go into a loop
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5495) * writing to the per cpu ring buffer various sizes of data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5496) * Some of the data will be large items, some small.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5497) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5498) * Another thread is created that goes into a spin, sending out
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5499) * IPIs to the other CPUs to also write into the ring buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5500) * this is to test the nesting ability of the buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5501) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5502) * Basic stats are recorded and reported. If something in the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5503) * ring buffer should happen that's not expected, a big warning
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5504) * is displayed and all ring buffers are disabled.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5505) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5506) static struct task_struct *rb_threads[NR_CPUS] __initdata;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5507)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5508) struct rb_test_data {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5509) struct trace_buffer *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5510) unsigned long events;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5511) unsigned long bytes_written;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5512) unsigned long bytes_alloc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5513) unsigned long bytes_dropped;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5514) unsigned long events_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5515) unsigned long bytes_written_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5516) unsigned long bytes_alloc_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5517) unsigned long bytes_dropped_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5518) int min_size_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5519) int max_size_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5520) int max_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5521) int min_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5522) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5523) int cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5524) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5525)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5526) static struct rb_test_data rb_data[NR_CPUS] __initdata;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5527)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5528) /* 1 meg per cpu */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5529) #define RB_TEST_BUFFER_SIZE 1048576
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5530)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5531) static char rb_string[] __initdata =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5532) "abcdefghijklmnopqrstuvwxyz1234567890!@#$%^&*()?+\\"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5533) "?+|:';\",.<>/?abcdefghijklmnopqrstuvwxyz1234567890"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5534) "!@#$%^&*()?+\\?+|:';\",.<>/?abcdefghijklmnopqrstuv";
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5535)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5536) static bool rb_test_started __initdata;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5537)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5538) struct rb_item {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5539) int size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5540) char str[];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5541) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5542)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5543) static __init int rb_write_something(struct rb_test_data *data, bool nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5544) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5545) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5546) struct rb_item *item;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5547) bool started;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5548) int event_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5549) int size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5550) int len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5551) int cnt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5552)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5553) /* Have nested writes different that what is written */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5554) cnt = data->cnt + (nested ? 27 : 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5555)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5556) /* Multiply cnt by ~e, to make some unique increment */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5557) size = (cnt * 68 / 25) % (sizeof(rb_string) - 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5558)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5559) len = size + sizeof(struct rb_item);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5560)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5561) started = rb_test_started;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5562) /* read rb_test_started before checking buffer enabled */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5563) smp_rmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5564)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5565) event = ring_buffer_lock_reserve(data->buffer, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5566) if (!event) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5567) /* Ignore dropped events before test starts. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5568) if (started) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5569) if (nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5570) data->bytes_dropped += len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5571) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5572) data->bytes_dropped_nested += len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5573) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5574) return len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5575) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5576)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5577) event_len = ring_buffer_event_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5578)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5579) if (RB_WARN_ON(data->buffer, event_len < len))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5580) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5581)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5582) item = ring_buffer_event_data(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5583) item->size = size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5584) memcpy(item->str, rb_string, size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5585)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5586) if (nested) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5587) data->bytes_alloc_nested += event_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5588) data->bytes_written_nested += len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5589) data->events_nested++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5590) if (!data->min_size_nested || len < data->min_size_nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5591) data->min_size_nested = len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5592) if (len > data->max_size_nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5593) data->max_size_nested = len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5594) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5595) data->bytes_alloc += event_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5596) data->bytes_written += len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5597) data->events++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5598) if (!data->min_size || len < data->min_size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5599) data->max_size = len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5600) if (len > data->max_size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5601) data->max_size = len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5602) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5603)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5604) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5605) ring_buffer_unlock_commit(data->buffer, event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5606)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5607) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5608) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5609)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5610) static __init int rb_test(void *arg)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5611) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5612) struct rb_test_data *data = arg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5613)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5614) while (!kthread_should_stop()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5615) rb_write_something(data, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5616) data->cnt++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5617)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5618) set_current_state(TASK_INTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5619) /* Now sleep between a min of 100-300us and a max of 1ms */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5620) usleep_range(((data->cnt % 3) + 1) * 100, 1000);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5621) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5622)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5623) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5624) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5625)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5626) static __init void rb_ipi(void *ignore)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5627) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5628) struct rb_test_data *data;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5629) int cpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5630)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5631) data = &rb_data[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5632) rb_write_something(data, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5633) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5634)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5635) static __init int rb_hammer_test(void *arg)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5636) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5637) while (!kthread_should_stop()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5638)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5639) /* Send an IPI to all cpus to write data! */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5640) smp_call_function(rb_ipi, NULL, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5641) /* No sleep, but for non preempt, let others run */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5642) schedule();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5643) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5644)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5645) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5646) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5647)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5648) static __init int test_ringbuffer(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5649) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5650) struct task_struct *rb_hammer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5651) struct trace_buffer *buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5652) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5653) int ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5654)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5655) if (security_locked_down(LOCKDOWN_TRACEFS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5656) pr_warn("Lockdown is enabled, skipping ring buffer tests\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5657) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5658) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5659)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5660) pr_info("Running ring buffer tests...\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5661)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5662) buffer = ring_buffer_alloc(RB_TEST_BUFFER_SIZE, RB_FL_OVERWRITE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5663) if (WARN_ON(!buffer))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5664) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5665)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5666) /* Disable buffer so that threads can't write to it yet */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5667) ring_buffer_record_off(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5668)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5669) for_each_online_cpu(cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5670) rb_data[cpu].buffer = buffer;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5671) rb_data[cpu].cpu = cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5672) rb_data[cpu].cnt = cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5673) rb_threads[cpu] = kthread_create(rb_test, &rb_data[cpu],
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5674) "rbtester/%d", cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5675) if (WARN_ON(IS_ERR(rb_threads[cpu]))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5676) pr_cont("FAILED\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5677) ret = PTR_ERR(rb_threads[cpu]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5678) goto out_free;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5679) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5680)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5681) kthread_bind(rb_threads[cpu], cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5682) wake_up_process(rb_threads[cpu]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5683) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5684)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5685) /* Now create the rb hammer! */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5686) rb_hammer = kthread_run(rb_hammer_test, NULL, "rbhammer");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5687) if (WARN_ON(IS_ERR(rb_hammer))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5688) pr_cont("FAILED\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5689) ret = PTR_ERR(rb_hammer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5690) goto out_free;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5691) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5692)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5693) ring_buffer_record_on(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5694) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5695) * Show buffer is enabled before setting rb_test_started.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5696) * Yes there's a small race window where events could be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5697) * dropped and the thread wont catch it. But when a ring
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5698) * buffer gets enabled, there will always be some kind of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5699) * delay before other CPUs see it. Thus, we don't care about
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5700) * those dropped events. We care about events dropped after
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5701) * the threads see that the buffer is active.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5702) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5703) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5704) rb_test_started = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5705)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5706) set_current_state(TASK_INTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5707) /* Just run for 10 seconds */;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5708) schedule_timeout(10 * HZ);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5709)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5710) kthread_stop(rb_hammer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5711)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5712) out_free:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5713) for_each_online_cpu(cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5714) if (!rb_threads[cpu])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5715) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5716) kthread_stop(rb_threads[cpu]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5717) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5718) if (ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5719) ring_buffer_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5720) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5721) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5722)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5723) /* Report! */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5724) pr_info("finished\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5725) for_each_online_cpu(cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5726) struct ring_buffer_event *event;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5727) struct rb_test_data *data = &rb_data[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5728) struct rb_item *item;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5729) unsigned long total_events;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5730) unsigned long total_dropped;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5731) unsigned long total_written;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5732) unsigned long total_alloc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5733) unsigned long total_read = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5734) unsigned long total_size = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5735) unsigned long total_len = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5736) unsigned long total_lost = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5737) unsigned long lost;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5738) int big_event_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5739) int small_event_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5740)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5741) ret = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5742)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5743) total_events = data->events + data->events_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5744) total_written = data->bytes_written + data->bytes_written_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5745) total_alloc = data->bytes_alloc + data->bytes_alloc_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5746) total_dropped = data->bytes_dropped + data->bytes_dropped_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5747)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5748) big_event_size = data->max_size + data->max_size_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5749) small_event_size = data->min_size + data->min_size_nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5750)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5751) pr_info("CPU %d:\n", cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5752) pr_info(" events: %ld\n", total_events);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5753) pr_info(" dropped bytes: %ld\n", total_dropped);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5754) pr_info(" alloced bytes: %ld\n", total_alloc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5755) pr_info(" written bytes: %ld\n", total_written);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5756) pr_info(" biggest event: %d\n", big_event_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5757) pr_info(" smallest event: %d\n", small_event_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5758)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5759) if (RB_WARN_ON(buffer, total_dropped))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5760) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5761)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5762) ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5763)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5764) while ((event = ring_buffer_consume(buffer, cpu, NULL, &lost))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5765) total_lost += lost;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5766) item = ring_buffer_event_data(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5767) total_len += ring_buffer_event_length(event);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5768) total_size += item->size + sizeof(struct rb_item);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5769) if (memcmp(&item->str[0], rb_string, item->size) != 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5770) pr_info("FAILED!\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5771) pr_info("buffer had: %.*s\n", item->size, item->str);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5772) pr_info("expected: %.*s\n", item->size, rb_string);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5773) RB_WARN_ON(buffer, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5774) ret = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5775) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5776) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5777) total_read++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5778) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5779) if (ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5780) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5781)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5782) ret = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5783)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5784) pr_info(" read events: %ld\n", total_read);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5785) pr_info(" lost events: %ld\n", total_lost);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5786) pr_info(" total events: %ld\n", total_lost + total_read);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5787) pr_info(" recorded len bytes: %ld\n", total_len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5788) pr_info(" recorded size bytes: %ld\n", total_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5789) if (total_lost)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5790) pr_info(" With dropped events, record len and size may not match\n"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5791) " alloced and written from above\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5792) if (!total_lost) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5793) if (RB_WARN_ON(buffer, total_len != total_alloc ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5794) total_size != total_written))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5795) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5796) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5797) if (RB_WARN_ON(buffer, total_lost + total_read != total_events))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5798) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5799)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5800) ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5801) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5802) if (!ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5803) pr_info("Ring buffer PASSED!\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5804)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5805) ring_buffer_free(buffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5806) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5807) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5808)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5809) late_initcall(test_ringbuffer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5810) #endif /* CONFIG_RING_BUFFER_STARTUP_TEST */
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags