^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) * Copyright (C) 2008 Oracle. All rights reserved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) #include <linux/kernel.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) #include <linux/bio.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) #include <linux/file.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9) #include <linux/fs.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) #include <linux/pagemap.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) #include <linux/highmem.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) #include <linux/time.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) #include <linux/init.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) #include <linux/string.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) #include <linux/backing-dev.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) #include <linux/writeback.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) #include <linux/slab.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) #include <linux/sched/mm.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) #include <linux/log2.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) #include <crypto/hash.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) #include "misc.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) #include "ctree.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) #include "disk-io.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) #include "transaction.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) #include "btrfs_inode.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) #include "volumes.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) #include "ordered-data.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) #include "compression.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) #include "extent_io.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) #include "extent_map.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) static const char* const btrfs_compress_types[] = { "", "zlib", "lzo", "zstd" };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) const char* btrfs_compress_type2str(enum btrfs_compression_type type)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) case BTRFS_COMPRESS_ZLIB:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38) case BTRFS_COMPRESS_LZO:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) case BTRFS_COMPRESS_ZSTD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) case BTRFS_COMPRESS_NONE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) return btrfs_compress_types[type];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) bool btrfs_compress_is_valid_type(const char *str, size_t len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) for (i = 1; i < ARRAY_SIZE(btrfs_compress_types); i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) size_t comp_len = strlen(btrfs_compress_types[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) if (len < comp_len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) if (!strncmp(btrfs_compress_types[i], str, comp_len))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) static int compression_compress_pages(int type, struct list_head *ws,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) struct address_space *mapping, u64 start, struct page **pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) unsigned long *out_pages, unsigned long *total_in,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) unsigned long *total_out)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) case BTRFS_COMPRESS_ZLIB:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) return zlib_compress_pages(ws, mapping, start, pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) out_pages, total_in, total_out);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) case BTRFS_COMPRESS_LZO:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) return lzo_compress_pages(ws, mapping, start, pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) out_pages, total_in, total_out);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) case BTRFS_COMPRESS_ZSTD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) return zstd_compress_pages(ws, mapping, start, pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79) out_pages, total_in, total_out);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) case BTRFS_COMPRESS_NONE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) * This can happen when compression races with remount setting
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84) * it to 'no compress', while caller doesn't call
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) * inode_need_compress() to check if we really need to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) * compress.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88) * Not a big deal, just need to inform caller that we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) * haven't allocated any pages yet.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) *out_pages = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) return -E2BIG;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96) static int compression_decompress_bio(int type, struct list_head *ws,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) struct compressed_bio *cb)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) case BTRFS_COMPRESS_ZLIB: return zlib_decompress_bio(ws, cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) case BTRFS_COMPRESS_LZO: return lzo_decompress_bio(ws, cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) case BTRFS_COMPRESS_ZSTD: return zstd_decompress_bio(ws, cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) case BTRFS_COMPRESS_NONE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) * This can't happen, the type is validated several times
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) * before we get here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) BUG();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) static int compression_decompress(int type, struct list_head *ws,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) unsigned char *data_in, struct page *dest_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) unsigned long start_byte, size_t srclen, size_t destlen)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) case BTRFS_COMPRESS_ZLIB: return zlib_decompress(ws, data_in, dest_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) start_byte, srclen, destlen);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) case BTRFS_COMPRESS_LZO: return lzo_decompress(ws, data_in, dest_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) start_byte, srclen, destlen);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) case BTRFS_COMPRESS_ZSTD: return zstd_decompress(ws, data_in, dest_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) start_byte, srclen, destlen);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) case BTRFS_COMPRESS_NONE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) * This can't happen, the type is validated several times
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) * before we get here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) BUG();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) static int btrfs_decompress_bio(struct compressed_bio *cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) static inline int compressed_bio_size(struct btrfs_fs_info *fs_info,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) unsigned long disk_size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) return sizeof(struct compressed_bio) +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) (DIV_ROUND_UP(disk_size, fs_info->sectorsize)) * csum_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) static int check_compressed_csum(struct btrfs_inode *inode, struct bio *bio,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) u64 disk_start)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) struct btrfs_fs_info *fs_info = inode->root->fs_info;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) const u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) unsigned long i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) char *kaddr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) u8 csum[BTRFS_CSUM_SIZE];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) struct compressed_bio *cb = bio->bi_private;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) u8 *cb_sum = cb->sums;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) if (inode->flags & BTRFS_INODE_NODATASUM)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) shash->tfm = fs_info->csum_shash;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) for (i = 0; i < cb->nr_pages; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) page = cb->compressed_pages[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) kaddr = kmap_atomic(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) crypto_shash_digest(shash, kaddr, PAGE_SIZE, csum);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) kunmap_atomic(kaddr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) if (memcmp(&csum, cb_sum, csum_size)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) btrfs_print_data_csum_error(inode, disk_start,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) csum, cb_sum, cb->mirror_num);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) if (btrfs_io_bio(bio)->device)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) btrfs_dev_stat_inc_and_print(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) btrfs_io_bio(bio)->device,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) BTRFS_DEV_STAT_CORRUPTION_ERRS);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) return -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) cb_sum += csum_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) /* when we finish reading compressed pages from the disk, we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) * decompress them and then run the bio end_io routines on the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) * decompressed pages (in the inode address space).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) * This allows the checksumming and other IO error handling routines
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) * to work normally
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) * The compressed pages are freed here, and it must be run
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) * in process context
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) static void end_compressed_bio_read(struct bio *bio)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) struct compressed_bio *cb = bio->bi_private;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) struct inode *inode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) unsigned long index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) unsigned int mirror = btrfs_io_bio(bio)->mirror_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) int ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) if (bio->bi_status)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) cb->errors = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) /* if there are more bios still pending for this compressed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) * extent, just exit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) if (!refcount_dec_and_test(&cb->pending_bios))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) * Record the correct mirror_num in cb->orig_bio so that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) * read-repair can work properly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) btrfs_io_bio(cb->orig_bio)->mirror_num = mirror;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) cb->mirror_num = mirror;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) * Some IO in this cb have failed, just skip checksum as there
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) * is no way it could be correct.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) if (cb->errors == 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) goto csum_failed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) inode = cb->inode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) ret = check_compressed_csum(BTRFS_I(inode), bio,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) (u64)bio->bi_iter.bi_sector << 9);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) if (ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) goto csum_failed;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) /* ok, we're the last bio for this extent, lets start
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) * the decompression.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) ret = btrfs_decompress_bio(cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) csum_failed:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) if (ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) cb->errors = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) /* release the compressed pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) index = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) for (index = 0; index < cb->nr_pages; index++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) page = cb->compressed_pages[index];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) page->mapping = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) put_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) /* do io completion on the original bio */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) if (cb->errors) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) bio_io_error(cb->orig_bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) struct bio_vec *bvec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) struct bvec_iter_all iter_all;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) * we have verified the checksum already, set page
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) * checked so the end_io handlers know about it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) ASSERT(!bio_flagged(bio, BIO_CLONED));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) bio_for_each_segment_all(bvec, cb->orig_bio, iter_all)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) SetPageChecked(bvec->bv_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) bio_endio(cb->orig_bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) /* finally free the cb struct */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) kfree(cb->compressed_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) kfree(cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) bio_put(bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) * Clear the writeback bits on all of the file
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) * pages for a compressed write
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) static noinline void end_compressed_writeback(struct inode *inode,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) const struct compressed_bio *cb)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) unsigned long index = cb->start >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) unsigned long end_index = (cb->start + cb->len - 1) >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) struct page *pages[16];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) unsigned long nr_pages = end_index - index + 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288) if (cb->errors)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) mapping_set_error(inode->i_mapping, -EIO);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) while (nr_pages > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) ret = find_get_pages_contig(inode->i_mapping, index,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) min_t(unsigned long,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) nr_pages, ARRAY_SIZE(pages)), pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) if (ret == 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) nr_pages -= 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) index += 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) for (i = 0; i < ret; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) if (cb->errors)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302) SetPageError(pages[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303) end_page_writeback(pages[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304) put_page(pages[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306) nr_pages -= ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) index += ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309) /* the inode may be gone now */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) * do the cleanup once all the compressed pages hit the disk.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) * This will clear writeback on the file pages and free the compressed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) * pages.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317) * This also calls the writeback end hooks for the file pages so that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318) * metadata and checksums can be updated in the file.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320) static void end_compressed_bio_write(struct bio *bio)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) struct compressed_bio *cb = bio->bi_private;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323) struct inode *inode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) unsigned long index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 327) if (bio->bi_status)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 328) cb->errors = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 329)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 330) /* if there are more bios still pending for this compressed
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 331) * extent, just exit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 332) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 333) if (!refcount_dec_and_test(&cb->pending_bios))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 334) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 335)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 336) /* ok, we're the last bio for this extent, step one is to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 337) * call back into the FS and do all the end_io operations
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 338) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 339) inode = cb->inode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 340) cb->compressed_pages[0]->mapping = cb->inode->i_mapping;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 341) btrfs_writepage_endio_finish_ordered(cb->compressed_pages[0],
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 342) cb->start, cb->start + cb->len - 1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 343) !cb->errors);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 344) cb->compressed_pages[0]->mapping = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 345)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 346) end_compressed_writeback(inode, cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 347) /* note, our inode could be gone now */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 348)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 349) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 350) * release the compressed pages, these came from alloc_page and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 351) * are not attached to the inode at all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 352) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 353) index = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 354) for (index = 0; index < cb->nr_pages; index++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 355) page = cb->compressed_pages[index];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 356) page->mapping = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 357) put_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 358) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 359)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 360) /* finally free the cb struct */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 361) kfree(cb->compressed_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 362) kfree(cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 363) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 364) bio_put(bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 365) }
^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) * worker function to build and submit bios for previously compressed pages.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 369) * The corresponding pages in the inode should be marked for writeback
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 370) * and the compressed pages should have a reference on them for dropping
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 371) * when the IO is complete.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 372) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 373) * This also checksums the file bytes and gets things ready for
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 374) * the end io hooks.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 375) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 376) blk_status_t btrfs_submit_compressed_write(struct btrfs_inode *inode, u64 start,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 377) unsigned long len, u64 disk_start,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 378) unsigned long compressed_len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 379) struct page **compressed_pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 380) unsigned long nr_pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 381) unsigned int write_flags,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 382) struct cgroup_subsys_state *blkcg_css)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 383) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 384) struct btrfs_fs_info *fs_info = inode->root->fs_info;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 385) struct bio *bio = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 386) struct compressed_bio *cb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 387) unsigned long bytes_left;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 388) int pg_index = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 389) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 390) u64 first_byte = disk_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 391) blk_status_t ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 392) int skip_sum = inode->flags & BTRFS_INODE_NODATASUM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 393)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 394) WARN_ON(!PAGE_ALIGNED(start));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 395) cb = kmalloc(compressed_bio_size(fs_info, compressed_len), GFP_NOFS);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 396) if (!cb)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 397) return BLK_STS_RESOURCE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 398) refcount_set(&cb->pending_bios, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 399) cb->errors = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 400) cb->inode = &inode->vfs_inode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 401) cb->start = start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 402) cb->len = len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 403) cb->mirror_num = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 404) cb->compressed_pages = compressed_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 405) cb->compressed_len = compressed_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 406) cb->orig_bio = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 407) cb->nr_pages = nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 408)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 409) bio = btrfs_bio_alloc(first_byte);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 410) bio->bi_opf = REQ_OP_WRITE | write_flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 411) bio->bi_private = cb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 412) bio->bi_end_io = end_compressed_bio_write;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 413)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 414) if (blkcg_css) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 415) bio->bi_opf |= REQ_CGROUP_PUNT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 416) kthread_associate_blkcg(blkcg_css);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 417) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 418) refcount_set(&cb->pending_bios, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 419)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 420) /* create and submit bios for the compressed pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 421) bytes_left = compressed_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 422) for (pg_index = 0; pg_index < cb->nr_pages; pg_index++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 423) int submit = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 424)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 425) page = compressed_pages[pg_index];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 426) page->mapping = inode->vfs_inode.i_mapping;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 427) if (bio->bi_iter.bi_size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 428) submit = btrfs_bio_fits_in_stripe(page, PAGE_SIZE, bio,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 429) 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 430)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 431) page->mapping = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 432) if (submit || bio_add_page(bio, page, PAGE_SIZE, 0) <
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 433) PAGE_SIZE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 434) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 435) * inc the count before we submit the bio so
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 436) * we know the end IO handler won't happen before
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 437) * we inc the count. Otherwise, the cb might get
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 438) * freed before we're done setting it up
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 439) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 440) refcount_inc(&cb->pending_bios);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 441) ret = btrfs_bio_wq_end_io(fs_info, bio,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 442) BTRFS_WQ_ENDIO_DATA);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 443) BUG_ON(ret); /* -ENOMEM */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 444)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 445) if (!skip_sum) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 446) ret = btrfs_csum_one_bio(inode, bio, start, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 447) BUG_ON(ret); /* -ENOMEM */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 448) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 449)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 450) ret = btrfs_map_bio(fs_info, bio, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 451) if (ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 452) bio->bi_status = ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 453) bio_endio(bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 454) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 455)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 456) bio = btrfs_bio_alloc(first_byte);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 457) bio->bi_opf = REQ_OP_WRITE | write_flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 458) bio->bi_private = cb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 459) bio->bi_end_io = end_compressed_bio_write;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 460) if (blkcg_css)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 461) bio->bi_opf |= REQ_CGROUP_PUNT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 462) bio_add_page(bio, page, PAGE_SIZE, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 463) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 464) if (bytes_left < PAGE_SIZE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 465) btrfs_info(fs_info,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 466) "bytes left %lu compress len %lu nr %lu",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 467) bytes_left, cb->compressed_len, cb->nr_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 468) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 469) bytes_left -= PAGE_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 470) first_byte += PAGE_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 471) cond_resched();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 472) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 473)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 474) ret = btrfs_bio_wq_end_io(fs_info, bio, BTRFS_WQ_ENDIO_DATA);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 475) BUG_ON(ret); /* -ENOMEM */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 476)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 477) if (!skip_sum) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 478) ret = btrfs_csum_one_bio(inode, bio, start, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 479) BUG_ON(ret); /* -ENOMEM */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 480) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 481)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 482) ret = btrfs_map_bio(fs_info, bio, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 483) if (ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 484) bio->bi_status = ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 485) bio_endio(bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 486) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 487)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 488) if (blkcg_css)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 489) kthread_associate_blkcg(NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 490)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 491) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 492) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 493)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 494) static u64 bio_end_offset(struct bio *bio)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 495) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 496) struct bio_vec *last = bio_last_bvec_all(bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 497)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 498) return page_offset(last->bv_page) + last->bv_len + last->bv_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 499) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 500)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 501) static noinline int add_ra_bio_pages(struct inode *inode,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 502) u64 compressed_end,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 503) struct compressed_bio *cb)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 504) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 505) unsigned long end_index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 506) unsigned long pg_index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 507) u64 last_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 508) u64 isize = i_size_read(inode);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 509) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 510) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 511) unsigned long nr_pages = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 512) struct extent_map *em;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 513) struct address_space *mapping = inode->i_mapping;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 514) struct extent_map_tree *em_tree;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 515) struct extent_io_tree *tree;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 516) u64 end;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 517) int misses = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 518)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 519) last_offset = bio_end_offset(cb->orig_bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 520) em_tree = &BTRFS_I(inode)->extent_tree;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 521) tree = &BTRFS_I(inode)->io_tree;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 522)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 523) if (isize == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 524) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 525)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 526) end_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 527)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 528) while (last_offset < compressed_end) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 529) pg_index = last_offset >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 530)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 531) if (pg_index > end_index)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 532) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 533)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 534) page = xa_load(&mapping->i_pages, pg_index);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 535) if (page && !xa_is_value(page)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 536) misses++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 537) if (misses > 4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 538) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 539) goto next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 540) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 541)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 542) page = __page_cache_alloc(mapping_gfp_constraint(mapping,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 543) ~__GFP_FS));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 544) if (!page)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 545) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 546)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 547) if (add_to_page_cache_lru(page, mapping, pg_index, GFP_NOFS)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 548) put_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 549) goto next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 550) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 551)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 552) end = last_offset + PAGE_SIZE - 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 553) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 554) * at this point, we have a locked page in the page cache
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 555) * for these bytes in the file. But, we have to make
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 556) * sure they map to this compressed extent on disk.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 557) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 558) set_page_extent_mapped(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 559) lock_extent(tree, last_offset, end);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 560) read_lock(&em_tree->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 561) em = lookup_extent_mapping(em_tree, last_offset,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 562) PAGE_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 563) read_unlock(&em_tree->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 564)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 565) if (!em || last_offset < em->start ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 566) (last_offset + PAGE_SIZE > extent_map_end(em)) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 567) (em->block_start >> 9) != cb->orig_bio->bi_iter.bi_sector) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 568) free_extent_map(em);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 569) unlock_extent(tree, last_offset, end);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 570) unlock_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 571) put_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 572) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 573) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 574) free_extent_map(em);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 575)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 576) if (page->index == end_index) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 577) char *userpage;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 578) size_t zero_offset = offset_in_page(isize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 579)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 580) if (zero_offset) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 581) int zeros;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 582) zeros = PAGE_SIZE - zero_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 583) userpage = kmap_atomic(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 584) memset(userpage + zero_offset, 0, zeros);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 585) flush_dcache_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 586) kunmap_atomic(userpage);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 587) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 588) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 589)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 590) ret = bio_add_page(cb->orig_bio, page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 591) PAGE_SIZE, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 592)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 593) if (ret == PAGE_SIZE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 594) nr_pages++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 595) put_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 596) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 597) unlock_extent(tree, last_offset, end);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 598) unlock_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 599) put_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 600) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 601) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 602) next:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 603) last_offset += PAGE_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 604) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 605) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 606) }
^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) * for a compressed read, the bio we get passed has all the inode pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 610) * in it. We don't actually do IO on those pages but allocate new ones
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 611) * to hold the compressed pages on disk.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 612) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 613) * bio->bi_iter.bi_sector points to the compressed extent on disk
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 614) * bio->bi_io_vec points to all of the inode pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 615) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 616) * After the compressed pages are read, we copy the bytes into the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 617) * bio we were passed and then call the bio end_io calls
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 618) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 619) blk_status_t btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 620) int mirror_num, unsigned long bio_flags)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 621) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 622) struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 623) struct extent_map_tree *em_tree;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 624) struct compressed_bio *cb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 625) unsigned long compressed_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 626) unsigned long nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 627) unsigned long pg_index;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 628) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 629) struct bio *comp_bio;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 630) u64 cur_disk_byte = (u64)bio->bi_iter.bi_sector << 9;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 631) u64 em_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 632) u64 em_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 633) struct extent_map *em;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 634) blk_status_t ret = BLK_STS_RESOURCE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 635) int faili = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 636) const u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 637) u8 *sums;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 638)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 639) em_tree = &BTRFS_I(inode)->extent_tree;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 640)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 641) /* we need the actual starting offset of this extent in the file */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 642) read_lock(&em_tree->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 643) em = lookup_extent_mapping(em_tree,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 644) page_offset(bio_first_page_all(bio)),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 645) PAGE_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 646) read_unlock(&em_tree->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 647) if (!em)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 648) return BLK_STS_IOERR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 649)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 650) compressed_len = em->block_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 651) cb = kmalloc(compressed_bio_size(fs_info, compressed_len), GFP_NOFS);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 652) if (!cb)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 653) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 654)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 655) refcount_set(&cb->pending_bios, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 656) cb->errors = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 657) cb->inode = inode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 658) cb->mirror_num = mirror_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 659) sums = cb->sums;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 660)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 661) cb->start = em->orig_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 662) em_len = em->len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 663) em_start = em->start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 664)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 665) free_extent_map(em);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 666) em = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 667)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 668) cb->len = bio->bi_iter.bi_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 669) cb->compressed_len = compressed_len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 670) cb->compress_type = extent_compress_type(bio_flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 671) cb->orig_bio = bio;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 672)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 673) nr_pages = DIV_ROUND_UP(compressed_len, PAGE_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 674) cb->compressed_pages = kcalloc(nr_pages, sizeof(struct page *),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 675) GFP_NOFS);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 676) if (!cb->compressed_pages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 677) goto fail1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 678)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 679) for (pg_index = 0; pg_index < nr_pages; pg_index++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 680) cb->compressed_pages[pg_index] = alloc_page(GFP_NOFS |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 681) __GFP_HIGHMEM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 682) if (!cb->compressed_pages[pg_index]) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 683) faili = pg_index - 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 684) ret = BLK_STS_RESOURCE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 685) goto fail2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 686) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 687) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 688) faili = nr_pages - 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 689) cb->nr_pages = nr_pages;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 690)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 691) add_ra_bio_pages(inode, em_start + em_len, cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 692)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 693) /* include any pages we added in add_ra-bio_pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 694) cb->len = bio->bi_iter.bi_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 695)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 696) comp_bio = btrfs_bio_alloc(cur_disk_byte);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 697) comp_bio->bi_opf = REQ_OP_READ;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 698) comp_bio->bi_private = cb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 699) comp_bio->bi_end_io = end_compressed_bio_read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 700) refcount_set(&cb->pending_bios, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 701)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 702) for (pg_index = 0; pg_index < nr_pages; pg_index++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 703) int submit = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 704)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 705) page = cb->compressed_pages[pg_index];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 706) page->mapping = inode->i_mapping;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 707) page->index = em_start >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 708)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 709) if (comp_bio->bi_iter.bi_size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 710) submit = btrfs_bio_fits_in_stripe(page, PAGE_SIZE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 711) comp_bio, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 712)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 713) page->mapping = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 714) if (submit || bio_add_page(comp_bio, page, PAGE_SIZE, 0) <
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 715) PAGE_SIZE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 716) unsigned int nr_sectors;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 717)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 718) ret = btrfs_bio_wq_end_io(fs_info, comp_bio,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 719) BTRFS_WQ_ENDIO_DATA);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 720) BUG_ON(ret); /* -ENOMEM */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 721)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 722) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 723) * inc the count before we submit the bio so
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 724) * we know the end IO handler won't happen before
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 725) * we inc the count. Otherwise, the cb might get
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 726) * freed before we're done setting it up
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 727) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 728) refcount_inc(&cb->pending_bios);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 729)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 730) if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 731) ret = btrfs_lookup_bio_sums(inode, comp_bio,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 732) (u64)-1, sums);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 733) BUG_ON(ret); /* -ENOMEM */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 734) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 735)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 736) nr_sectors = DIV_ROUND_UP(comp_bio->bi_iter.bi_size,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 737) fs_info->sectorsize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 738) sums += csum_size * nr_sectors;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 739)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 740) ret = btrfs_map_bio(fs_info, comp_bio, mirror_num);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 741) if (ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 742) comp_bio->bi_status = ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 743) bio_endio(comp_bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 744) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 745)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 746) comp_bio = btrfs_bio_alloc(cur_disk_byte);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 747) comp_bio->bi_opf = REQ_OP_READ;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 748) comp_bio->bi_private = cb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 749) comp_bio->bi_end_io = end_compressed_bio_read;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 750)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 751) bio_add_page(comp_bio, page, PAGE_SIZE, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 752) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 753) cur_disk_byte += PAGE_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 754) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 755)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 756) ret = btrfs_bio_wq_end_io(fs_info, comp_bio, BTRFS_WQ_ENDIO_DATA);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 757) BUG_ON(ret); /* -ENOMEM */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 758)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 759) if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 760) ret = btrfs_lookup_bio_sums(inode, comp_bio, (u64)-1, sums);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 761) BUG_ON(ret); /* -ENOMEM */
^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) ret = btrfs_map_bio(fs_info, comp_bio, mirror_num);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 765) if (ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 766) comp_bio->bi_status = ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 767) bio_endio(comp_bio);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 768) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 769)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 770) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 771)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 772) fail2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 773) while (faili >= 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 774) __free_page(cb->compressed_pages[faili]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 775) faili--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 776) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 777)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 778) kfree(cb->compressed_pages);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 779) fail1:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 780) kfree(cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 781) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 782) free_extent_map(em);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 783) return ret;
^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) * Heuristic uses systematic sampling to collect data from the input data
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 788) * range, the logic can be tuned by the following constants:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 789) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 790) * @SAMPLING_READ_SIZE - how many bytes will be copied from for each sample
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 791) * @SAMPLING_INTERVAL - range from which the sampled data can be collected
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 792) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 793) #define SAMPLING_READ_SIZE (16)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 794) #define SAMPLING_INTERVAL (256)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 795)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 796) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 797) * For statistical analysis of the input data we consider bytes that form a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 798) * Galois Field of 256 objects. Each object has an attribute count, ie. how
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 799) * many times the object appeared in the sample.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 800) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 801) #define BUCKET_SIZE (256)
^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) * The size of the sample is based on a statistical sampling rule of thumb.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 805) * The common way is to perform sampling tests as long as the number of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 806) * elements in each cell is at least 5.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 807) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 808) * Instead of 5, we choose 32 to obtain more accurate results.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 809) * If the data contain the maximum number of symbols, which is 256, we obtain a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 810) * sample size bound by 8192.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 811) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 812) * For a sample of at most 8KB of data per data range: 16 consecutive bytes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 813) * from up to 512 locations.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 814) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 815) #define MAX_SAMPLE_SIZE (BTRFS_MAX_UNCOMPRESSED * \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 816) SAMPLING_READ_SIZE / SAMPLING_INTERVAL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 817)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 818) struct bucket_item {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 819) u32 count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 820) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 821)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 822) struct heuristic_ws {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 823) /* Partial copy of input data */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 824) u8 *sample;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 825) u32 sample_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 826) /* Buckets store counters for each byte value */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 827) struct bucket_item *bucket;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 828) /* Sorting buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 829) struct bucket_item *bucket_b;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 830) struct list_head list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 831) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 832)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 833) static struct workspace_manager heuristic_wsm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 834)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 835) static void free_heuristic_ws(struct list_head *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 836) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 837) struct heuristic_ws *workspace;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 838)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 839) workspace = list_entry(ws, struct heuristic_ws, list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 840)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 841) kvfree(workspace->sample);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 842) kfree(workspace->bucket);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 843) kfree(workspace->bucket_b);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 844) kfree(workspace);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 845) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 846)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 847) static struct list_head *alloc_heuristic_ws(unsigned int level)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 848) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 849) struct heuristic_ws *ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 850)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 851) ws = kzalloc(sizeof(*ws), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 852) if (!ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 853) return ERR_PTR(-ENOMEM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 854)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 855) ws->sample = kvmalloc(MAX_SAMPLE_SIZE, GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 856) if (!ws->sample)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 857) goto fail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 858)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 859) ws->bucket = kcalloc(BUCKET_SIZE, sizeof(*ws->bucket), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 860) if (!ws->bucket)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 861) goto fail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 862)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 863) ws->bucket_b = kcalloc(BUCKET_SIZE, sizeof(*ws->bucket_b), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 864) if (!ws->bucket_b)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 865) goto fail;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 866)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 867) INIT_LIST_HEAD(&ws->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 868) return &ws->list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 869) fail:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 870) free_heuristic_ws(&ws->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 871) return ERR_PTR(-ENOMEM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 872) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 873)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 874) const struct btrfs_compress_op btrfs_heuristic_compress = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 875) .workspace_manager = &heuristic_wsm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 876) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 877)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 878) static const struct btrfs_compress_op * const btrfs_compress_op[] = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 879) /* The heuristic is represented as compression type 0 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 880) &btrfs_heuristic_compress,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 881) &btrfs_zlib_compress,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 882) &btrfs_lzo_compress,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 883) &btrfs_zstd_compress,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 884) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 885)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 886) static struct list_head *alloc_workspace(int type, unsigned int level)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 887) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 888) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 889) case BTRFS_COMPRESS_NONE: return alloc_heuristic_ws(level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 890) case BTRFS_COMPRESS_ZLIB: return zlib_alloc_workspace(level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 891) case BTRFS_COMPRESS_LZO: return lzo_alloc_workspace(level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 892) case BTRFS_COMPRESS_ZSTD: return zstd_alloc_workspace(level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 893) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 894) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 895) * This can't happen, the type is validated several times
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 896) * before we get here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 897) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 898) BUG();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 899) }
^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) static void free_workspace(int type, struct list_head *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 903) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 904) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 905) case BTRFS_COMPRESS_NONE: return free_heuristic_ws(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 906) case BTRFS_COMPRESS_ZLIB: return zlib_free_workspace(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 907) case BTRFS_COMPRESS_LZO: return lzo_free_workspace(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 908) case BTRFS_COMPRESS_ZSTD: return zstd_free_workspace(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 909) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 910) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 911) * This can't happen, the type is validated several times
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 912) * before we get here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 913) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 914) BUG();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 915) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 916) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 917)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 918) static void btrfs_init_workspace_manager(int type)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 919) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 920) struct workspace_manager *wsm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 921) struct list_head *workspace;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 922)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 923) wsm = btrfs_compress_op[type]->workspace_manager;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 924) INIT_LIST_HEAD(&wsm->idle_ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 925) spin_lock_init(&wsm->ws_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 926) atomic_set(&wsm->total_ws, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 927) init_waitqueue_head(&wsm->ws_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 928)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 929) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 930) * Preallocate one workspace for each compression type so we can
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 931) * guarantee forward progress in the worst case
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 932) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 933) workspace = alloc_workspace(type, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 934) if (IS_ERR(workspace)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 935) pr_warn(
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 936) "BTRFS: cannot preallocate compression workspace, will try later\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 937) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 938) atomic_set(&wsm->total_ws, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 939) wsm->free_ws = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 940) list_add(workspace, &wsm->idle_ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 941) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 942) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 943)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 944) static void btrfs_cleanup_workspace_manager(int type)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 945) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 946) struct workspace_manager *wsman;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 947) struct list_head *ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 948)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 949) wsman = btrfs_compress_op[type]->workspace_manager;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 950) while (!list_empty(&wsman->idle_ws)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 951) ws = wsman->idle_ws.next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 952) list_del(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 953) free_workspace(type, ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 954) atomic_dec(&wsman->total_ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 955) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 956) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 957)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 958) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 959) * This finds an available workspace or allocates a new one.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 960) * If it's not possible to allocate a new one, waits until there's one.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 961) * Preallocation makes a forward progress guarantees and we do not return
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 962) * errors.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 963) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 964) struct list_head *btrfs_get_workspace(int type, unsigned int level)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 965) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 966) struct workspace_manager *wsm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 967) struct list_head *workspace;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 968) int cpus = num_online_cpus();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 969) unsigned nofs_flag;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 970) struct list_head *idle_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 971) spinlock_t *ws_lock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 972) atomic_t *total_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 973) wait_queue_head_t *ws_wait;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 974) int *free_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 975)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 976) wsm = btrfs_compress_op[type]->workspace_manager;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 977) idle_ws = &wsm->idle_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 978) ws_lock = &wsm->ws_lock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 979) total_ws = &wsm->total_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 980) ws_wait = &wsm->ws_wait;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 981) free_ws = &wsm->free_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 982)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 983) again:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 984) spin_lock(ws_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 985) if (!list_empty(idle_ws)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 986) workspace = idle_ws->next;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 987) list_del(workspace);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 988) (*free_ws)--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 989) spin_unlock(ws_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 990) return workspace;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 991)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 992) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 993) if (atomic_read(total_ws) > cpus) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 994) DEFINE_WAIT(wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 995)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 996) spin_unlock(ws_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 997) prepare_to_wait(ws_wait, &wait, TASK_UNINTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 998) if (atomic_read(total_ws) > cpus && !*free_ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 999) schedule();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1000) finish_wait(ws_wait, &wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1001) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1002) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1003) atomic_inc(total_ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1004) spin_unlock(ws_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1005)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1006) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1007) * Allocation helpers call vmalloc that can't use GFP_NOFS, so we have
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1008) * to turn it off here because we might get called from the restricted
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1009) * context of btrfs_compress_bio/btrfs_compress_pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1010) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1011) nofs_flag = memalloc_nofs_save();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1012) workspace = alloc_workspace(type, level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1013) memalloc_nofs_restore(nofs_flag);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1014)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1015) if (IS_ERR(workspace)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1016) atomic_dec(total_ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1017) wake_up(ws_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1018)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1019) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1020) * Do not return the error but go back to waiting. There's a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1021) * workspace preallocated for each type and the compression
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1022) * time is bounded so we get to a workspace eventually. This
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1023) * makes our caller's life easier.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1024) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1025) * To prevent silent and low-probability deadlocks (when the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1026) * initial preallocation fails), check if there are any
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1027) * workspaces at all.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1028) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1029) if (atomic_read(total_ws) == 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1030) static DEFINE_RATELIMIT_STATE(_rs,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1031) /* once per minute */ 60 * HZ,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1032) /* no burst */ 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1033)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1034) if (__ratelimit(&_rs)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1035) pr_warn("BTRFS: no compression workspaces, low memory, retrying\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1036) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1037) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1038) goto again;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1039) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1040) return workspace;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1041) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1042)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1043) static struct list_head *get_workspace(int type, int level)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1044) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1045) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1046) case BTRFS_COMPRESS_NONE: return btrfs_get_workspace(type, level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1047) case BTRFS_COMPRESS_ZLIB: return zlib_get_workspace(level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1048) case BTRFS_COMPRESS_LZO: return btrfs_get_workspace(type, level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1049) case BTRFS_COMPRESS_ZSTD: return zstd_get_workspace(level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1050) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1051) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1052) * This can't happen, the type is validated several times
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1053) * before we get here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1054) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1055) BUG();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1056) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1057) }
^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) * put a workspace struct back on the list or free it if we have enough
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1061) * idle ones sitting around
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1062) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1063) void btrfs_put_workspace(int type, struct list_head *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1064) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1065) struct workspace_manager *wsm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1066) struct list_head *idle_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1067) spinlock_t *ws_lock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1068) atomic_t *total_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1069) wait_queue_head_t *ws_wait;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1070) int *free_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1071)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1072) wsm = btrfs_compress_op[type]->workspace_manager;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1073) idle_ws = &wsm->idle_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1074) ws_lock = &wsm->ws_lock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1075) total_ws = &wsm->total_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1076) ws_wait = &wsm->ws_wait;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1077) free_ws = &wsm->free_ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1078)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1079) spin_lock(ws_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1080) if (*free_ws <= num_online_cpus()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1081) list_add(ws, idle_ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1082) (*free_ws)++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1083) spin_unlock(ws_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1084) goto wake;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1085) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1086) spin_unlock(ws_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1087)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1088) free_workspace(type, ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1089) atomic_dec(total_ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1090) wake:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1091) cond_wake_up(ws_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1092) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1093)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1094) static void put_workspace(int type, struct list_head *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1095) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1096) switch (type) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1097) case BTRFS_COMPRESS_NONE: return btrfs_put_workspace(type, ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1098) case BTRFS_COMPRESS_ZLIB: return btrfs_put_workspace(type, ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1099) case BTRFS_COMPRESS_LZO: return btrfs_put_workspace(type, ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1100) case BTRFS_COMPRESS_ZSTD: return zstd_put_workspace(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1101) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1102) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1103) * This can't happen, the type is validated several times
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1104) * before we get here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1105) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1106) BUG();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1107) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1108) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1109)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1110) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1111) * Adjust @level according to the limits of the compression algorithm or
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1112) * fallback to default
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1113) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1114) static unsigned int btrfs_compress_set_level(int type, unsigned level)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1115) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1116) const struct btrfs_compress_op *ops = btrfs_compress_op[type];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1117)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1118) if (level == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1119) level = ops->default_level;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1120) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1121) level = min(level, ops->max_level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1122)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1123) return level;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1124) }
^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) * Given an address space and start and length, compress the bytes into @pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1128) * that are allocated on demand.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1129) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1130) * @type_level is encoded algorithm and level, where level 0 means whatever
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1131) * default the algorithm chooses and is opaque here;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1132) * - compression algo are 0-3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1133) * - the level are bits 4-7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1134) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1135) * @out_pages is an in/out parameter, holds maximum number of pages to allocate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1136) * and returns number of actually allocated pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1137) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1138) * @total_in is used to return the number of bytes actually read. It
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1139) * may be smaller than the input length if we had to exit early because we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1140) * ran out of room in the pages array or because we cross the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1141) * max_out threshold.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1142) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1143) * @total_out is an in/out parameter, must be set to the input length and will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1144) * be also used to return the total number of compressed bytes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1145) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1146) * @max_out tells us the max number of bytes that we're allowed to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1147) * stuff into pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1148) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1149) int btrfs_compress_pages(unsigned int type_level, struct address_space *mapping,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1150) u64 start, struct page **pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1151) unsigned long *out_pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1152) unsigned long *total_in,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1153) unsigned long *total_out)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1154) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1155) int type = btrfs_compress_type(type_level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1156) int level = btrfs_compress_level(type_level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1157) struct list_head *workspace;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1158) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1159)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1160) level = btrfs_compress_set_level(type, level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1161) workspace = get_workspace(type, level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1162) ret = compression_compress_pages(type, workspace, mapping, start, pages,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1163) out_pages, total_in, total_out);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1164) put_workspace(type, workspace);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1165) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1166) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1167)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1168) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1169) * pages_in is an array of pages with compressed data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1170) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1171) * disk_start is the starting logical offset of this array in the file
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1172) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1173) * orig_bio contains the pages from the file that we want to decompress into
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1174) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1175) * srclen is the number of bytes in pages_in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1176) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1177) * The basic idea is that we have a bio that was created by readpages.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1178) * The pages in the bio are for the uncompressed data, and they may not
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1179) * be contiguous. They all correspond to the range of bytes covered by
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1180) * the compressed extent.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1181) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1182) static int btrfs_decompress_bio(struct compressed_bio *cb)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1183) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1184) struct list_head *workspace;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1185) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1186) int type = cb->compress_type;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1187)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1188) workspace = get_workspace(type, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1189) ret = compression_decompress_bio(type, workspace, cb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1190) put_workspace(type, workspace);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1191)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1192) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1193) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1194)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1195) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1196) * a less complex decompression routine. Our compressed data fits in a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1197) * single page, and we want to read a single page out of it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1198) * start_byte tells us the offset into the compressed data we're interested in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1199) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1200) int btrfs_decompress(int type, unsigned char *data_in, struct page *dest_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1201) unsigned long start_byte, size_t srclen, size_t destlen)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1202) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1203) struct list_head *workspace;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1204) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1205)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1206) workspace = get_workspace(type, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1207) ret = compression_decompress(type, workspace, data_in, dest_page,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1208) start_byte, srclen, destlen);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1209) put_workspace(type, workspace);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1210)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1211) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1212) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1213)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1214) void __init btrfs_init_compress(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1215) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1216) btrfs_init_workspace_manager(BTRFS_COMPRESS_NONE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1217) btrfs_init_workspace_manager(BTRFS_COMPRESS_ZLIB);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1218) btrfs_init_workspace_manager(BTRFS_COMPRESS_LZO);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1219) zstd_init_workspace_manager();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1220) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1221)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1222) void __cold btrfs_exit_compress(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1223) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1224) btrfs_cleanup_workspace_manager(BTRFS_COMPRESS_NONE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1225) btrfs_cleanup_workspace_manager(BTRFS_COMPRESS_ZLIB);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1226) btrfs_cleanup_workspace_manager(BTRFS_COMPRESS_LZO);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1227) zstd_cleanup_workspace_manager();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1228) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1229)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1230) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1231) * Copy uncompressed data from working buffer to pages.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1232) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1233) * buf_start is the byte offset we're of the start of our workspace buffer.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1234) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1235) * total_out is the last byte of the buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1236) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1237) int btrfs_decompress_buf2page(const char *buf, unsigned long buf_start,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1238) unsigned long total_out, u64 disk_start,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1239) struct bio *bio)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1240) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1241) unsigned long buf_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1242) unsigned long current_buf_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1243) unsigned long start_byte;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1244) unsigned long prev_start_byte;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1245) unsigned long working_bytes = total_out - buf_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1246) unsigned long bytes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1247) char *kaddr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1248) struct bio_vec bvec = bio_iter_iovec(bio, bio->bi_iter);
^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) * start byte is the first byte of the page we're currently
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1252) * copying into relative to the start of the compressed data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1253) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1254) start_byte = page_offset(bvec.bv_page) - disk_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1255)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1256) /* we haven't yet hit data corresponding to this page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1257) if (total_out <= start_byte)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1258) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1259)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1260) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1261) * the start of the data we care about is offset into
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1262) * the middle of our working buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1263) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1264) if (total_out > start_byte && buf_start < start_byte) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1265) buf_offset = start_byte - buf_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1266) working_bytes -= buf_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1267) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1268) buf_offset = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1269) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1270) current_buf_start = buf_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1271)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1272) /* copy bytes from the working buffer into the pages */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1273) while (working_bytes > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1274) bytes = min_t(unsigned long, bvec.bv_len,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1275) PAGE_SIZE - (buf_offset % PAGE_SIZE));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1276) bytes = min(bytes, working_bytes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1277)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1278) kaddr = kmap_atomic(bvec.bv_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1279) memcpy(kaddr + bvec.bv_offset, buf + buf_offset, bytes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1280) kunmap_atomic(kaddr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1281) flush_dcache_page(bvec.bv_page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1282)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1283) buf_offset += bytes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1284) working_bytes -= bytes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1285) current_buf_start += bytes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1286)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1287) /* check if we need to pick another page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1288) bio_advance(bio, bytes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1289) if (!bio->bi_iter.bi_size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1290) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1291) bvec = bio_iter_iovec(bio, bio->bi_iter);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1292) prev_start_byte = start_byte;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1293) start_byte = page_offset(bvec.bv_page) - disk_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1294)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1295) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1296) * We need to make sure we're only adjusting
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1297) * our offset into compression working buffer when
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1298) * we're switching pages. Otherwise we can incorrectly
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1299) * keep copying when we were actually done.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1300) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1301) if (start_byte != prev_start_byte) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1302) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1303) * make sure our new page is covered by this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1304) * working buffer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1305) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1306) if (total_out <= start_byte)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1307) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1308)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1309) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1310) * the next page in the biovec might not be adjacent
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1311) * to the last page, but it might still be found
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1312) * inside this working buffer. bump our offset pointer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1313) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1314) if (total_out > start_byte &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1315) current_buf_start < start_byte) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1316) buf_offset = start_byte - buf_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1317) working_bytes = total_out - start_byte;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1318) current_buf_start = buf_start + buf_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1319) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1320) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1321) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1322)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1323) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1324) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1325)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1326) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1327) * Shannon Entropy calculation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1328) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1329) * Pure byte distribution analysis fails to determine compressibility of data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1330) * Try calculating entropy to estimate the average minimum number of bits
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1331) * needed to encode the sampled data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1332) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1333) * For convenience, return the percentage of needed bits, instead of amount of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1334) * bits directly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1335) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1336) * @ENTROPY_LVL_ACEPTABLE - below that threshold, sample has low byte entropy
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1337) * and can be compressible with high probability
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1338) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1339) * @ENTROPY_LVL_HIGH - data are not compressible with high probability
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1340) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1341) * Use of ilog2() decreases precision, we lower the LVL to 5 to compensate.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1342) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1343) #define ENTROPY_LVL_ACEPTABLE (65)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1344) #define ENTROPY_LVL_HIGH (80)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1345)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1346) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1347) * For increasead precision in shannon_entropy calculation,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1348) * let's do pow(n, M) to save more digits after comma:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1349) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1350) * - maximum int bit length is 64
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1351) * - ilog2(MAX_SAMPLE_SIZE) -> 13
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1352) * - 13 * 4 = 52 < 64 -> M = 4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1353) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1354) * So use pow(n, 4).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1355) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1356) static inline u32 ilog2_w(u64 n)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1357) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1358) return ilog2(n * n * n * n);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1359) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1360)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1361) static u32 shannon_entropy(struct heuristic_ws *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1362) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1363) const u32 entropy_max = 8 * ilog2_w(2);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1364) u32 entropy_sum = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1365) u32 p, p_base, sz_base;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1366) u32 i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1367)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1368) sz_base = ilog2_w(ws->sample_size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1369) for (i = 0; i < BUCKET_SIZE && ws->bucket[i].count > 0; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1370) p = ws->bucket[i].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1371) p_base = ilog2_w(p);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1372) entropy_sum += p * (sz_base - p_base);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1373) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1374)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1375) entropy_sum /= ws->sample_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1376) return entropy_sum * 100 / entropy_max;
^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) #define RADIX_BASE 4U
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1380) #define COUNTERS_SIZE (1U << RADIX_BASE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1381)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1382) static u8 get4bits(u64 num, int shift) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1383) u8 low4bits;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1384)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1385) num >>= shift;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1386) /* Reverse order */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1387) low4bits = (COUNTERS_SIZE - 1) - (num % COUNTERS_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1388) return low4bits;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1389) }
^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) * Use 4 bits as radix base
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1393) * Use 16 u32 counters for calculating new position in buf array
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1394) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1395) * @array - array that will be sorted
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1396) * @array_buf - buffer array to store sorting results
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1397) * must be equal in size to @array
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1398) * @num - array size
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1399) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1400) static void radix_sort(struct bucket_item *array, struct bucket_item *array_buf,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1401) int num)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1402) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1403) u64 max_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1404) u64 buf_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1405) u32 counters[COUNTERS_SIZE];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1406) u32 new_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1407) u32 addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1408) int bitlen;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1409) int shift;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1410) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1411)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1412) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1413) * Try avoid useless loop iterations for small numbers stored in big
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1414) * counters. Example: 48 33 4 ... in 64bit array
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1415) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1416) max_num = array[0].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1417) for (i = 1; i < num; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1418) buf_num = array[i].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1419) if (buf_num > max_num)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1420) max_num = buf_num;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1421) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1422)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1423) buf_num = ilog2(max_num);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1424) bitlen = ALIGN(buf_num, RADIX_BASE * 2);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1425)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1426) shift = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1427) while (shift < bitlen) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1428) memset(counters, 0, sizeof(counters));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1429)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1430) for (i = 0; i < num; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1431) buf_num = array[i].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1432) addr = get4bits(buf_num, shift);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1433) counters[addr]++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1434) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1435)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1436) for (i = 1; i < COUNTERS_SIZE; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1437) counters[i] += counters[i - 1];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1438)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1439) for (i = num - 1; i >= 0; i--) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1440) buf_num = array[i].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1441) addr = get4bits(buf_num, shift);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1442) counters[addr]--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1443) new_addr = counters[addr];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1444) array_buf[new_addr] = array[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1445) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1446)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1447) shift += RADIX_BASE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1448)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1449) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1450) * Normal radix expects to move data from a temporary array, to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1451) * the main one. But that requires some CPU time. Avoid that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1452) * by doing another sort iteration to original array instead of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1453) * memcpy()
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1454) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1455) memset(counters, 0, sizeof(counters));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1456)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1457) for (i = 0; i < num; i ++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1458) buf_num = array_buf[i].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1459) addr = get4bits(buf_num, shift);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1460) counters[addr]++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1461) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1462)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1463) for (i = 1; i < COUNTERS_SIZE; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1464) counters[i] += counters[i - 1];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1465)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1466) for (i = num - 1; i >= 0; i--) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1467) buf_num = array_buf[i].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1468) addr = get4bits(buf_num, shift);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1469) counters[addr]--;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1470) new_addr = counters[addr];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1471) array[new_addr] = array_buf[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1472) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1473)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1474) shift += RADIX_BASE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1475) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1476) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1477)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1478) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1479) * Size of the core byte set - how many bytes cover 90% of the sample
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1480) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1481) * There are several types of structured binary data that use nearly all byte
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1482) * values. The distribution can be uniform and counts in all buckets will be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1483) * nearly the same (eg. encrypted data). Unlikely to be compressible.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1484) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1485) * Other possibility is normal (Gaussian) distribution, where the data could
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1486) * be potentially compressible, but we have to take a few more steps to decide
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1487) * how much.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1488) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1489) * @BYTE_CORE_SET_LOW - main part of byte values repeated frequently,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1490) * compression algo can easy fix that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1491) * @BYTE_CORE_SET_HIGH - data have uniform distribution and with high
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1492) * probability is not compressible
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1493) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1494) #define BYTE_CORE_SET_LOW (64)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1495) #define BYTE_CORE_SET_HIGH (200)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1496)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1497) static int byte_core_set_size(struct heuristic_ws *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1498) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1499) u32 i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1500) u32 coreset_sum = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1501) const u32 core_set_threshold = ws->sample_size * 90 / 100;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1502) struct bucket_item *bucket = ws->bucket;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1503)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1504) /* Sort in reverse order */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1505) radix_sort(ws->bucket, ws->bucket_b, BUCKET_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1506)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1507) for (i = 0; i < BYTE_CORE_SET_LOW; i++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1508) coreset_sum += bucket[i].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1509)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1510) if (coreset_sum > core_set_threshold)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1511) return i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1512)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1513) for (; i < BYTE_CORE_SET_HIGH && bucket[i].count > 0; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1514) coreset_sum += bucket[i].count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1515) if (coreset_sum > core_set_threshold)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1516) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1517) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1518)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1519) return i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1520) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1521)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1522) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1523) * Count byte values in buckets.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1524) * This heuristic can detect textual data (configs, xml, json, html, etc).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1525) * Because in most text-like data byte set is restricted to limited number of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1526) * possible characters, and that restriction in most cases makes data easy to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1527) * compress.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1528) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1529) * @BYTE_SET_THRESHOLD - consider all data within this byte set size:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1530) * less - compressible
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1531) * more - need additional analysis
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1532) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1533) #define BYTE_SET_THRESHOLD (64)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1534)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1535) static u32 byte_set_size(const struct heuristic_ws *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1536) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1537) u32 i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1538) u32 byte_set_size = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1539)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1540) for (i = 0; i < BYTE_SET_THRESHOLD; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1541) if (ws->bucket[i].count > 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1542) byte_set_size++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1543) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1544)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1545) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1546) * Continue collecting count of byte values in buckets. If the byte
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1547) * set size is bigger then the threshold, it's pointless to continue,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1548) * the detection technique would fail for this type of data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1549) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1550) for (; i < BUCKET_SIZE; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1551) if (ws->bucket[i].count > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1552) byte_set_size++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1553) if (byte_set_size > BYTE_SET_THRESHOLD)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1554) return byte_set_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1555) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1556) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1557)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1558) return byte_set_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1559) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1560)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1561) static bool sample_repeated_patterns(struct heuristic_ws *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1562) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1563) const u32 half_of_sample = ws->sample_size / 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1564) const u8 *data = ws->sample;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1565)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1566) return memcmp(&data[0], &data[half_of_sample], half_of_sample) == 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1567) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1568)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1569) static void heuristic_collect_sample(struct inode *inode, u64 start, u64 end,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1570) struct heuristic_ws *ws)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1571) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1572) struct page *page;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1573) u64 index, index_end;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1574) u32 i, curr_sample_pos;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1575) u8 *in_data;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1576)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1577) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1578) * Compression handles the input data by chunks of 128KiB
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1579) * (defined by BTRFS_MAX_UNCOMPRESSED)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1580) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1581) * We do the same for the heuristic and loop over the whole range.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1582) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1583) * MAX_SAMPLE_SIZE - calculated under assumption that heuristic will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1584) * process no more than BTRFS_MAX_UNCOMPRESSED at a time.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1585) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1586) if (end - start > BTRFS_MAX_UNCOMPRESSED)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1587) end = start + BTRFS_MAX_UNCOMPRESSED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1588)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1589) index = start >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1590) index_end = end >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1591)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1592) /* Don't miss unaligned end */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1593) if (!IS_ALIGNED(end, PAGE_SIZE))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1594) index_end++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1595)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1596) curr_sample_pos = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1597) while (index < index_end) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1598) page = find_get_page(inode->i_mapping, index);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1599) in_data = kmap(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1600) /* Handle case where the start is not aligned to PAGE_SIZE */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1601) i = start % PAGE_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1602) while (i < PAGE_SIZE - SAMPLING_READ_SIZE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1603) /* Don't sample any garbage from the last page */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1604) if (start > end - SAMPLING_READ_SIZE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1605) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1606) memcpy(&ws->sample[curr_sample_pos], &in_data[i],
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1607) SAMPLING_READ_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1608) i += SAMPLING_INTERVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1609) start += SAMPLING_INTERVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1610) curr_sample_pos += SAMPLING_READ_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1611) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1612) kunmap(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1613) put_page(page);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1614)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1615) index++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1616) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1617)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1618) ws->sample_size = curr_sample_pos;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1619) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1620)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1621) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1622) * Compression heuristic.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1623) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1624) * For now is's a naive and optimistic 'return true', we'll extend the logic to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1625) * quickly (compared to direct compression) detect data characteristics
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1626) * (compressible/uncompressible) to avoid wasting CPU time on uncompressible
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1627) * data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1628) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1629) * The following types of analysis can be performed:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1630) * - detect mostly zero data
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1631) * - detect data with low "byte set" size (text, etc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1632) * - detect data with low/high "core byte" set
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1633) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1634) * Return non-zero if the compression should be done, 0 otherwise.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1635) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1636) int btrfs_compress_heuristic(struct inode *inode, u64 start, u64 end)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1637) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1638) struct list_head *ws_list = get_workspace(0, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1639) struct heuristic_ws *ws;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1640) u32 i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1641) u8 byte;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1642) int ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1643)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1644) ws = list_entry(ws_list, struct heuristic_ws, list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1645)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1646) heuristic_collect_sample(inode, start, end, ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1647)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1648) if (sample_repeated_patterns(ws)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1649) ret = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1650) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1651) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1652)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1653) memset(ws->bucket, 0, sizeof(*ws->bucket)*BUCKET_SIZE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1654)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1655) for (i = 0; i < ws->sample_size; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1656) byte = ws->sample[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1657) ws->bucket[byte].count++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1658) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1659)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1660) i = byte_set_size(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1661) if (i < BYTE_SET_THRESHOLD) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1662) ret = 2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1663) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1664) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1665)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1666) i = byte_core_set_size(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1667) if (i <= BYTE_CORE_SET_LOW) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1668) ret = 3;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1669) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1670) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1671)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1672) if (i >= BYTE_CORE_SET_HIGH) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1673) ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1674) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1675) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1676)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1677) i = shannon_entropy(ws);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1678) if (i <= ENTROPY_LVL_ACEPTABLE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1679) ret = 4;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1680) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1681) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1682)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1683) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1684) * For the levels below ENTROPY_LVL_HIGH, additional analysis would be
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1685) * needed to give green light to compression.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1686) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1687) * For now just assume that compression at that level is not worth the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1688) * resources because:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1689) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1690) * 1. it is possible to defrag the data later
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1691) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1692) * 2. the data would turn out to be hardly compressible, eg. 150 byte
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1693) * values, every bucket has counter at level ~54. The heuristic would
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1694) * be confused. This can happen when data have some internal repeated
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1695) * patterns like "abbacbbc...". This can be detected by analyzing
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1696) * pairs of bytes, which is too costly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1697) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1698) if (i < ENTROPY_LVL_HIGH) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1699) ret = 5;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1700) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1701) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1702) ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1703) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1704) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1705)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1706) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1707) put_workspace(0, ws_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1708) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1709) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1710)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1711) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1712) * Convert the compression suffix (eg. after "zlib" starting with ":") to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1713) * level, unrecognized string will set the default level
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1714) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1715) unsigned int btrfs_compress_str2level(unsigned int type, const char *str)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1716) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1717) unsigned int level = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1718) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1719)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1720) if (!type)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1721) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1722)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1723) if (str[0] == ':') {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1724) ret = kstrtouint(str + 1, 10, &level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1725) if (ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1726) level = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1727) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1728)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1729) level = btrfs_compress_set_level(type, level);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1730)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1731) return level;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1732) }
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags