^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) /* checksum.S: Sparc optimized checksum code.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) * Copyright(C) 1995 Linus Torvalds
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5) * Copyright(C) 1995 Miguel de Icaza
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) * Copyright(C) 1996 David S. Miller
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) * Copyright(C) 1997 Jakub Jelinek
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9) * derived from:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) * Linux/Alpha checksum c-code
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) * Linux/ix86 inline checksum assembly
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) * RFC1071 Computing the Internet Checksum (esp. Jacobsons m68k code)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) * David Mosberger-Tang for optimized reference c-code
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) * BSD4.4 portable checksum routine
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) #include <asm/errno.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) #include <asm/export.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) #define CSUM_BIGCHUNK(buf, offset, sum, t0, t1, t2, t3, t4, t5) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) ldd [buf + offset + 0x00], t0; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) ldd [buf + offset + 0x08], t2; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) addxcc t0, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) addxcc t1, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) ldd [buf + offset + 0x10], t4; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) addxcc t2, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) addxcc t3, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) ldd [buf + offset + 0x18], t0; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) addxcc t4, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) addxcc t5, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) addxcc t0, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) addxcc t1, sum, sum;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) #define CSUM_LASTCHUNK(buf, offset, sum, t0, t1, t2, t3) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) ldd [buf - offset - 0x08], t0; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) ldd [buf - offset - 0x00], t2; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) addxcc t0, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38) addxcc t1, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) addxcc t2, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) addxcc t3, sum, sum;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) /* Do end cruft out of band to get better cache patterns. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) csum_partial_end_cruft:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) be 1f ! caller asks %o1 & 0x8
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) andcc %o1, 4, %g0 ! nope, check for word remaining
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46) ldd [%o0], %g2 ! load two
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) addcc %g2, %o2, %o2 ! add first word to sum
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) addxcc %g3, %o2, %o2 ! add second word as well
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) add %o0, 8, %o0 ! advance buf ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) addx %g0, %o2, %o2 ! add in final carry
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) andcc %o1, 4, %g0 ! check again for word remaining
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) 1: be 1f ! nope, skip this code
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) andcc %o1, 3, %o1 ! check for trailing bytes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) ld [%o0], %g2 ! load it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) addcc %g2, %o2, %o2 ! add to sum
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) add %o0, 4, %o0 ! advance buf ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) addx %g0, %o2, %o2 ! add in final carry
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) andcc %o1, 3, %g0 ! check again for trailing bytes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) 1: be 1f ! no trailing bytes, return
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) addcc %o1, -1, %g0 ! only one byte remains?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) bne 2f ! at least two bytes more
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) subcc %o1, 2, %o1 ! only two bytes more?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) b 4f ! only one byte remains
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) or %g0, %g0, %o4 ! clear fake hword value
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) 2: lduh [%o0], %o4 ! get hword
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) be 6f ! jmp if only hword remains
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) add %o0, 2, %o0 ! advance buf ptr either way
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) sll %o4, 16, %o4 ! create upper hword
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) 4: ldub [%o0], %o5 ! get final byte
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) sll %o5, 8, %o5 ! put into place
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) or %o5, %o4, %o4 ! coalese with hword (if any)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) 6: addcc %o4, %o2, %o2 ! add to sum
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) 1: retl ! get outta here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) addx %g0, %o2, %o0 ! add final carry into retval
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) /* Also do alignment out of band to get better cache patterns. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) csum_partial_fix_alignment:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) cmp %o1, 6
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79) bl cpte - 0x4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) andcc %o0, 0x2, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81) be 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82) andcc %o0, 0x4, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) lduh [%o0 + 0x00], %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84) sub %o1, 2, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) add %o0, 2, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) sll %g2, 16, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87) addcc %g2, %o2, %o2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88) srl %o2, 16, %g3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) addx %g0, %g3, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) sll %o2, 16, %o2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) sll %g2, 16, %g3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) srl %o2, 16, %o2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93) andcc %o0, 0x4, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) or %g3, %o2, %o2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) 1: be cpa
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96) andcc %o1, 0xffffff80, %o3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) ld [%o0 + 0x00], %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) sub %o1, 4, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) addcc %g2, %o2, %o2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) add %o0, 4, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) addx %g0, %o2, %o2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) b cpa
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) andcc %o1, 0xffffff80, %o3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) /* The common case is to get called with a nicely aligned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) * buffer of size 0x20. Follow the code path for that case.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) .globl csum_partial
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) EXPORT_SYMBOL(csum_partial)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) csum_partial: /* %o0=buf, %o1=len, %o2=sum */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) andcc %o0, 0x7, %g0 ! alignment problems?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) bne csum_partial_fix_alignment ! yep, handle it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) sethi %hi(cpte - 8), %g7 ! prepare table jmp ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) andcc %o1, 0xffffff80, %o3 ! num loop iterations
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) cpa: be 3f ! none to do
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) andcc %o1, 0x70, %g1 ! clears carry flag too
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) 5: CSUM_BIGCHUNK(%o0, 0x00, %o2, %o4, %o5, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) CSUM_BIGCHUNK(%o0, 0x20, %o2, %o4, %o5, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) CSUM_BIGCHUNK(%o0, 0x40, %o2, %o4, %o5, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) CSUM_BIGCHUNK(%o0, 0x60, %o2, %o4, %o5, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) addx %g0, %o2, %o2 ! sink in final carry
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) subcc %o3, 128, %o3 ! detract from loop iters
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) bne 5b ! more to do
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) add %o0, 128, %o0 ! advance buf ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) andcc %o1, 0x70, %g1 ! clears carry flag too
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) 3: be cpte ! nope
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) andcc %o1, 0xf, %g0 ! anything left at all?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) srl %g1, 1, %o4 ! compute offset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) sub %g7, %g1, %g7 ! adjust jmp ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) sub %g7, %o4, %g7 ! final jmp ptr adjust
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) jmp %g7 + %lo(cpte - 8) ! enter the table
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) add %o0, %g1, %o0 ! advance buf ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) cptbl: CSUM_LASTCHUNK(%o0, 0x68, %o2, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) CSUM_LASTCHUNK(%o0, 0x58, %o2, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) CSUM_LASTCHUNK(%o0, 0x48, %o2, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) CSUM_LASTCHUNK(%o0, 0x38, %o2, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) CSUM_LASTCHUNK(%o0, 0x28, %o2, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) CSUM_LASTCHUNK(%o0, 0x18, %o2, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) CSUM_LASTCHUNK(%o0, 0x08, %o2, %g2, %g3, %g4, %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) addx %g0, %o2, %o2 ! fetch final carry
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) andcc %o1, 0xf, %g0 ! anything left at all?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) cpte: bne csum_partial_end_cruft ! yep, handle it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) andcc %o1, 8, %g0 ! check how much
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) cpout: retl ! get outta here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) mov %o2, %o0 ! return computed csum
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) /* Work around cpp -rob */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) #define ALLOC #alloc
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) #define EXECINSTR #execinstr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) #define EX(x,y) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) 98: x,y; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) .section __ex_table,ALLOC; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) .align 4; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) .word 98b, cc_fault; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) .text; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) .align 4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) #define EXT(start,end) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) .section __ex_table,ALLOC; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) .align 4; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) .word start, 0, end, cc_fault; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) .text; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) .align 4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) /* This aligned version executes typically in 8.5 superscalar cycles, this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) * is the best I can do. I say 8.5 because the final add will pair with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) * the next ldd in the main unrolled loop. Thus the pipe is always full.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) * If you change these macros (including order of instructions),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) * please check the fixup code below as well.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) #define CSUMCOPY_BIGCHUNK_ALIGNED(src, dst, sum, off, t0, t1, t2, t3, t4, t5, t6, t7) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) ldd [src + off + 0x00], t0; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) ldd [src + off + 0x08], t2; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) addxcc t0, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) ldd [src + off + 0x10], t4; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) addxcc t1, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) ldd [src + off + 0x18], t6; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) addxcc t2, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) std t0, [dst + off + 0x00]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) addxcc t3, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) std t2, [dst + off + 0x08]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) addxcc t4, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) std t4, [dst + off + 0x10]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) addxcc t5, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) std t6, [dst + off + 0x18]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) addxcc t6, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) addxcc t7, sum, sum;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) /* 12 superscalar cycles seems to be the limit for this case,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) * because of this we thus do all the ldd's together to get
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) * Viking MXCC into streaming mode. Ho hum...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) #define CSUMCOPY_BIGCHUNK(src, dst, sum, off, t0, t1, t2, t3, t4, t5, t6, t7) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) ldd [src + off + 0x00], t0; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) ldd [src + off + 0x08], t2; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) ldd [src + off + 0x10], t4; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) ldd [src + off + 0x18], t6; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) st t0, [dst + off + 0x00]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) addxcc t0, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) st t1, [dst + off + 0x04]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) addxcc t1, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) st t2, [dst + off + 0x08]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) addxcc t2, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) st t3, [dst + off + 0x0c]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) addxcc t3, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) st t4, [dst + off + 0x10]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) addxcc t4, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) st t5, [dst + off + 0x14]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) addxcc t5, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) st t6, [dst + off + 0x18]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) addxcc t6, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) st t7, [dst + off + 0x1c]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) addxcc t7, sum, sum;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) /* Yuck, 6 superscalar cycles... */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) #define CSUMCOPY_LASTCHUNK(src, dst, sum, off, t0, t1, t2, t3) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) ldd [src - off - 0x08], t0; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) ldd [src - off - 0x00], t2; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) addxcc t0, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) st t0, [dst - off - 0x08]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) addxcc t1, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) st t1, [dst - off - 0x04]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) addxcc t2, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) st t2, [dst - off - 0x00]; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) addxcc t3, sum, sum; \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) st t3, [dst - off + 0x04];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) /* Handle the end cruft code out of band for better cache patterns. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) cc_end_cruft:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) be 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) andcc %o3, 4, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) EX(ldd [%o0 + 0x00], %g2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) add %o1, 8, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) addcc %g2, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) add %o0, 8, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) addxcc %g3, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) EX(st %g2, [%o1 - 0x08])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) addx %g0, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) andcc %o3, 4, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) EX(st %g3, [%o1 - 0x04])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) 1: be 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) andcc %o3, 3, %o3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) EX(ld [%o0 + 0x00], %g2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) add %o1, 4, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) addcc %g2, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) EX(st %g2, [%o1 - 0x04])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) addx %g0, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) andcc %o3, 3, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) add %o0, 4, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) 1: be 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) addcc %o3, -1, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) bne 2f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) subcc %o3, 2, %o3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) b 4f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) or %g0, %g0, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) 2: EX(lduh [%o0 + 0x00], %o4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) add %o0, 2, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) EX(sth %o4, [%o1 + 0x00])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) be 6f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) add %o1, 2, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) sll %o4, 16, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) 4: EX(ldub [%o0 + 0x00], %o5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) EX(stb %o5, [%o1 + 0x00])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) sll %o5, 8, %o5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) or %o5, %o4, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) 6: addcc %o4, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) 1: retl
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) addx %g0, %g7, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) /* Also, handle the alignment code out of band. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) cc_dword_align:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) cmp %g1, 16
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) bge 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) srl %g1, 1, %o3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) 2: cmp %o3, 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) be,a ccte
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) andcc %g1, 0xf, %o3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) andcc %o3, %o0, %g0 ! Check %o0 only (%o1 has the same last 2 bits)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) be,a 2b
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) srl %o3, 1, %o3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) 1: andcc %o0, 0x1, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) bne ccslow
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) andcc %o0, 0x2, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) be 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) andcc %o0, 0x4, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) EX(lduh [%o0 + 0x00], %g4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) sub %g1, 2, %g1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288) EX(sth %g4, [%o1 + 0x00])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) add %o0, 2, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) sll %g4, 16, %g4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) addcc %g4, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) add %o1, 2, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) srl %g7, 16, %g3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) addx %g0, %g3, %g4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) sll %g7, 16, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) sll %g4, 16, %g3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) srl %g7, 16, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) andcc %o0, 0x4, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) or %g3, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) 1: be 3f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) andcc %g1, 0xffffff80, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302) EX(ld [%o0 + 0x00], %g4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303) sub %g1, 4, %g1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304) EX(st %g4, [%o1 + 0x00])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) add %o0, 4, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306) addcc %g4, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) add %o1, 4, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) addx %g0, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309) b 3f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310) andcc %g1, 0xffffff80, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312) /* Sun, you just can't beat me, you just can't. Stop trying,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) * give up. I'm serious, I am going to kick the living shit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) * out of you, game over, lights out.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) .align 8
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317) .globl __csum_partial_copy_sparc_generic
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318) EXPORT_SYMBOL(__csum_partial_copy_sparc_generic)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319) __csum_partial_copy_sparc_generic:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320) /* %o0=src, %o1=dest, %g1=len, %g7=sum */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321) xor %o0, %o1, %o4 ! get changing bits
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) andcc %o4, 3, %g0 ! check for mismatched alignment
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323) bne ccslow ! better this than unaligned/fixups
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) andcc %o0, 7, %g0 ! need to align things?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) bne cc_dword_align ! yes, we check for short lengths there
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326) andcc %g1, 0xffffff80, %g0 ! can we use unrolled loop?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 327) 3: be 3f ! nope, less than one loop remains
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 328) andcc %o1, 4, %g0 ! dest aligned on 4 or 8 byte boundary?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 329) be ccdbl + 4 ! 8 byte aligned, kick ass
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 330) 5: CSUMCOPY_BIGCHUNK(%o0,%o1,%g7,0x00,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 331) CSUMCOPY_BIGCHUNK(%o0,%o1,%g7,0x20,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 332) CSUMCOPY_BIGCHUNK(%o0,%o1,%g7,0x40,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 333) CSUMCOPY_BIGCHUNK(%o0,%o1,%g7,0x60,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 334) 10: EXT(5b, 10b) ! note for exception handling
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 335) sub %g1, 128, %g1 ! detract from length
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 336) addx %g0, %g7, %g7 ! add in last carry bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 337) andcc %g1, 0xffffff80, %g0 ! more to csum?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 338) add %o0, 128, %o0 ! advance src ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 339) bne 5b ! we did not go negative, continue looping
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 340) add %o1, 128, %o1 ! advance dest ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 341) 3: andcc %g1, 0x70, %o2 ! can use table?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 342) ccmerge:be ccte ! nope, go and check for end cruft
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 343) andcc %g1, 0xf, %o3 ! get low bits of length (clears carry btw)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 344) srl %o2, 1, %o4 ! begin negative offset computation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 345) sethi %hi(12f), %o5 ! set up table ptr end
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 346) add %o0, %o2, %o0 ! advance src ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 347) sub %o5, %o4, %o5 ! continue table calculation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 348) sll %o2, 1, %g2 ! constant multiplies are fun...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 349) sub %o5, %g2, %o5 ! some more adjustments
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 350) jmp %o5 + %lo(12f) ! jump into it, duff style, wheee...
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 351) add %o1, %o2, %o1 ! advance dest ptr (carry is clear btw)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 352) cctbl: CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x68,%g2,%g3,%g4,%g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 353) CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x58,%g2,%g3,%g4,%g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 354) CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x48,%g2,%g3,%g4,%g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 355) CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x38,%g2,%g3,%g4,%g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 356) CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x28,%g2,%g3,%g4,%g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 357) CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x18,%g2,%g3,%g4,%g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 358) CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x08,%g2,%g3,%g4,%g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 359) 12: EXT(cctbl, 12b) ! note for exception table handling
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 360) addx %g0, %g7, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 361) andcc %o3, 0xf, %g0 ! check for low bits set
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 362) ccte: bne cc_end_cruft ! something left, handle it out of band
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 363) andcc %o3, 8, %g0 ! begin checks for that code
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 364) retl ! return
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 365) mov %g7, %o0 ! give em the computed checksum
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 366) ccdbl: CSUMCOPY_BIGCHUNK_ALIGNED(%o0,%o1,%g7,0x00,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 367) CSUMCOPY_BIGCHUNK_ALIGNED(%o0,%o1,%g7,0x20,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 368) CSUMCOPY_BIGCHUNK_ALIGNED(%o0,%o1,%g7,0x40,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 369) CSUMCOPY_BIGCHUNK_ALIGNED(%o0,%o1,%g7,0x60,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 370) 11: EXT(ccdbl, 11b) ! note for exception table handling
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 371) sub %g1, 128, %g1 ! detract from length
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 372) addx %g0, %g7, %g7 ! add in last carry bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 373) andcc %g1, 0xffffff80, %g0 ! more to csum?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 374) add %o0, 128, %o0 ! advance src ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 375) bne ccdbl ! we did not go negative, continue looping
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 376) add %o1, 128, %o1 ! advance dest ptr
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 377) b ccmerge ! finish it off, above
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 378) andcc %g1, 0x70, %o2 ! can use table? (clears carry btw)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 379)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 380) ccslow: cmp %g1, 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 381) mov 0, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 382) bleu 4f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 383) andcc %o0, 1, %o5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 384) be,a 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 385) srl %g1, 1, %g4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 386) sub %g1, 1, %g1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 387) EX(ldub [%o0], %g5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 388) add %o0, 1, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 389) EX(stb %g5, [%o1])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 390) srl %g1, 1, %g4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 391) add %o1, 1, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 392) 1: cmp %g4, 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 393) be,a 3f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 394) andcc %g1, 1, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 395) andcc %o0, 2, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 396) be,a 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 397) srl %g4, 1, %g4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 398) EX(lduh [%o0], %o4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 399) sub %g1, 2, %g1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 400) srl %o4, 8, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 401) sub %g4, 1, %g4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 402) EX(stb %g2, [%o1])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 403) add %o4, %g5, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 404) EX(stb %o4, [%o1 + 1])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 405) add %o0, 2, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 406) srl %g4, 1, %g4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 407) add %o1, 2, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 408) 1: cmp %g4, 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 409) be,a 2f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 410) andcc %g1, 2, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 411) EX(ld [%o0], %o4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 412) 5: srl %o4, 24, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 413) srl %o4, 16, %g3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 414) EX(stb %g2, [%o1])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 415) srl %o4, 8, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 416) EX(stb %g3, [%o1 + 1])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 417) add %o0, 4, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 418) EX(stb %g2, [%o1 + 2])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 419) addcc %o4, %g5, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 420) EX(stb %o4, [%o1 + 3])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 421) addx %g5, %g0, %g5 ! I am now to lazy to optimize this (question it
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 422) add %o1, 4, %o1 ! is worthy). Maybe some day - with the sll/srl
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 423) subcc %g4, 1, %g4 ! tricks
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 424) bne,a 5b
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 425) EX(ld [%o0], %o4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 426) sll %g5, 16, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 427) srl %g5, 16, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 428) srl %g2, 16, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 429) andcc %g1, 2, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 430) add %g2, %g5, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 431) 2: be,a 3f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 432) andcc %g1, 1, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 433) EX(lduh [%o0], %o4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 434) andcc %g1, 1, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 435) srl %o4, 8, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 436) add %o0, 2, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 437) EX(stb %g2, [%o1])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 438) add %g5, %o4, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 439) EX(stb %o4, [%o1 + 1])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 440) add %o1, 2, %o1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 441) 3: be,a 1f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 442) sll %g5, 16, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 443) EX(ldub [%o0], %g2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 444) sll %g2, 8, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 445) EX(stb %g2, [%o1])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 446) add %g5, %o4, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 447) sll %g5, 16, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 448) 1: addcc %o4, %g5, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 449) srl %g5, 16, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 450) addx %g0, %o4, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 451) orcc %o5, %g0, %g0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 452) be 4f
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 453) srl %g5, 8, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 454) and %g5, 0xff, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 455) and %o4, 0xff, %o4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 456) sll %g2, 8, %g2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 457) or %g2, %o4, %g5
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 458) 4: addcc %g7, %g5, %g7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 459) retl
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 460) addx %g0, %g7, %o0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 461)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 462) /* We do these strange calculations for the csum_*_from_user case only, ie.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 463) * we only bother with faults on loads... */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 464)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 465) cc_fault:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 466) ret
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 467) clr %o0
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags