AuxXxilium/linux_dsm_epyc7002
1
0
Fork 0
mirror of https://github.com/AuxXxilium/linux_dsm_epyc7002.git synced 2024-12-11 23:16:43 +07:00
Code Issues Actions Packages Projects Releases Wiki Activity
0a06d42566
linux_dsm_epyc7002/arch/sparc/lib/checksum_32.S
Greg Kroah-Hartman b24413180f License cleanup: add SPDX GPL-2.0 license identifier to files with no license 
Many source files in the tree are missing licensing information, which
makes it harder for compliance tools to determine the correct license.

By default all files without license information are under the default
license of the kernel, which is GPL version 2.

Update the files which contain no license information with the 'GPL-2.0'
SPDX license identifier.  The SPDX identifier is a legally binding
shorthand, which can be used instead of the full boiler plate text.

This patch is based on work done by Thomas Gleixner and Kate Stewart and
Philippe Ombredanne.

How this work was done:

Patches were generated and checked against linux-4.14-rc6 for a subset of
the use cases:
 - file had no licensing information it it.
 - file was a */uapi/* one with no licensing information in it,
 - file was a */uapi/* one with existing licensing information,

Further patches will be generated in subsequent months to fix up cases
where non-standard license headers were used, and references to license
had to be inferred by heuristics based on keywords.

The analysis to determine which SPDX License Identifier to be applied to
a file was done in a spreadsheet of side by side results from of the
output of two independent scanners (ScanCode & Windriver) producing SPDX
tag:value files created by Philippe Ombredanne.  Philippe prepared the
base worksheet, and did an initial spot review of a few 1000 files.

The 4.13 kernel was the starting point of the analysis with 60,537 files
assessed.  Kate Stewart did a file by file comparison of the scanner
results in the spreadsheet to determine which SPDX license identifier(s)
to be applied to the file. She confirmed any determination that was not
immediately clear with lawyers working with the Linux Foundation.

Criteria used to select files for SPDX license identifier tagging was:
 - Files considered eligible had to be source code files.
 - Make and config files were included as candidates if they contained >5
   lines of source
 - File already had some variant of a license header in it (even if <5
   lines).

All documentation files were explicitly excluded.

The following heuristics were used to determine which SPDX license
identifiers to apply.

 - when both scanners couldn't find any license traces, file was
   considered to have no license information in it, and the top level
   COPYING file license applied.

   For non */uapi/* files that summary was:

   SPDX license identifier                            # files
   ---------------------------------------------------|-------
   GPL-2.0                                              11139

   and resulted in the first patch in this series.

   If that file was a */uapi/* path one, it was "GPL-2.0 WITH
   Linux-syscall-note" otherwise it was "GPL-2.0".  Results of that was:

   SPDX license identifier                            # files
   ---------------------------------------------------|-------
   GPL-2.0 WITH Linux-syscall-note                        930

   and resulted in the second patch in this series.

 - if a file had some form of licensing information in it, and was one
   of the */uapi/* ones, it was denoted with the Linux-syscall-note if
   any GPL family license was found in the file or had no licensing in
   it (per prior point).  Results summary:

   SPDX license identifier                            # files
   ---------------------------------------------------|------
   GPL-2.0 WITH Linux-syscall-note                       270
   GPL-2.0+ WITH Linux-syscall-note                      169
   ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause)    21
   ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)    17
   LGPL-2.1+ WITH Linux-syscall-note                      15
   GPL-1.0+ WITH Linux-syscall-note                       14
   ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause)    5
   LGPL-2.0+ WITH Linux-syscall-note                       4
   LGPL-2.1 WITH Linux-syscall-note                        3
   ((GPL-2.0 WITH Linux-syscall-note) OR MIT)              3
   ((GPL-2.0 WITH Linux-syscall-note) AND MIT)             1

   and that resulted in the third patch in this series.

 - when the two scanners agreed on the detected license(s), that became
   the concluded license(s).

 - when there was disagreement between the two scanners (one detected a
   license but the other didn't, or they both detected different
   licenses) a manual inspection of the file occurred.

 - In most cases a manual inspection of the information in the file
   resulted in a clear resolution of the license that should apply (and
   which scanner probably needed to revisit its heuristics).

 - When it was not immediately clear, the license identifier was
   confirmed with lawyers working with the Linux Foundation.

 - If there was any question as to the appropriate license identifier,
   the file was flagged for further research and to be revisited later
   in time.

In total, over 70 hours of logged manual review was done on the
spreadsheet to determine the SPDX license identifiers to apply to the
source files by Kate, Philippe, Thomas and, in some cases, confirmation
by lawyers working with the Linux Foundation.

Kate also obtained a third independent scan of the 4.13 code base from
FOSSology, and compared selected files where the other two scanners
disagreed against that SPDX file, to see if there was new insights.  The
Windriver scanner is based on an older version of FOSSology in part, so
they are related.

Thomas did random spot checks in about 500 files from the spreadsheets
for the uapi headers and agreed with SPDX license identifier in the
files he inspected. For the non-uapi files Thomas did random spot checks
in about 15000 files.

In initial set of patches against 4.14-rc6, 3 files were found to have
copy/paste license identifier errors, and have been fixed to reflect the
correct identifier.

Additionally Philippe spent 10 hours this week doing a detailed manual
inspection and review of the 12,461 patched files from the initial patch
version early this week with:
 - a full scancode scan run, collecting the matched texts, detected
   license ids and scores
 - reviewing anything where there was a license detected (about 500+
   files) to ensure that the applied SPDX license was correct
 - reviewing anything where there was no detection but the patch license
   was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied
   SPDX license was correct

This produced a worksheet with 20 files needing minor correction.  This
worksheet was then exported into 3 different .csv files for the
different types of files to be modified.

These .csv files were then reviewed by Greg.  Thomas wrote a script to
parse the csv files and add the proper SPDX tag to the file, in the
format that the file expected.  This script was further refined by Greg
based on the output to detect more types of files automatically and to
distinguish between header and source .c files (which need different
comment types.)  Finally Greg ran the script using the .csv files to
generate the patches.

Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-02 11:10:55 +01:00

594 lines
17 KiB
ArmAsm
Raw Blame History

/* SPDX-License-Identifier: GPL-2.0 */
/* checksum.S: Sparc optimized checksum code.
*
* Copyright(C) 1995 Linus Torvalds
* Copyright(C) 1995 Miguel de Icaza
* Copyright(C) 1996 David S. Miller
* Copyright(C) 1997 Jakub Jelinek
*
* derived from:
* Linux/Alpha checksum c-code
* Linux/ix86 inline checksum assembly
* RFC1071 Computing the Internet Checksum (esp. Jacobsons m68k code)
* David Mosberger-Tang for optimized reference c-code
* BSD4.4 portable checksum routine
*/
#include <asm/errno.h>
#include <asm/export.h>
#define CSUM_BIGCHUNK(buf, offset, sum, t0, t1, t2, t3, t4, t5) \
ldd [buf + offset + 0x00], t0; \
ldd [buf + offset + 0x08], t2; \
addxcc t0, sum, sum; \
addxcc t1, sum, sum; \
ldd [buf + offset + 0x10], t4; \
addxcc t2, sum, sum; \
addxcc t3, sum, sum; \
ldd [buf + offset + 0x18], t0; \
addxcc t4, sum, sum; \
addxcc t5, sum, sum; \
addxcc t0, sum, sum; \
addxcc t1, sum, sum;
#define CSUM_LASTCHUNK(buf, offset, sum, t0, t1, t2, t3) \
ldd [buf - offset - 0x08], t0; \
ldd [buf - offset - 0x00], t2; \
addxcc t0, sum, sum; \
addxcc t1, sum, sum; \
addxcc t2, sum, sum; \
addxcc t3, sum, sum;
/* Do end cruft out of band to get better cache patterns. */
csum_partial_end_cruft:
be 1f ! caller asks %o1 & 0x8
andcc %o1, 4, %g0 ! nope, check for word remaining
ldd [%o0], %g2 ! load two
addcc %g2, %o2, %o2 ! add first word to sum
addxcc %g3, %o2, %o2 ! add second word as well
add %o0, 8, %o0 ! advance buf ptr
addx %g0, %o2, %o2 ! add in final carry
andcc %o1, 4, %g0 ! check again for word remaining
1: be 1f ! nope, skip this code
andcc %o1, 3, %o1 ! check for trailing bytes
ld [%o0], %g2 ! load it
addcc %g2, %o2, %o2 ! add to sum
add %o0, 4, %o0 ! advance buf ptr
addx %g0, %o2, %o2 ! add in final carry
andcc %o1, 3, %g0 ! check again for trailing bytes
1: be 1f ! no trailing bytes, return
addcc %o1, -1, %g0 ! only one byte remains?
bne 2f ! at least two bytes more
subcc %o1, 2, %o1 ! only two bytes more?
b 4f ! only one byte remains
or %g0, %g0, %o4 ! clear fake hword value
2: lduh [%o0], %o4 ! get hword
be 6f ! jmp if only hword remains
add %o0, 2, %o0 ! advance buf ptr either way
sll %o4, 16, %o4 ! create upper hword
4: ldub [%o0], %o5 ! get final byte
sll %o5, 8, %o5 ! put into place
or %o5, %o4, %o4 ! coalese with hword (if any)
6: addcc %o4, %o2, %o2 ! add to sum
1: retl ! get outta here
addx %g0, %o2, %o0 ! add final carry into retval
/* Also do alignment out of band to get better cache patterns. */
csum_partial_fix_alignment:
cmp %o1, 6
bl cpte - 0x4
andcc %o0, 0x2, %g0
be 1f
andcc %o0, 0x4, %g0
lduh [%o0 + 0x00], %g2
sub %o1, 2, %o1
add %o0, 2, %o0
sll %g2, 16, %g2
addcc %g2, %o2, %o2
srl %o2, 16, %g3
addx %g0, %g3, %g2
sll %o2, 16, %o2
sll %g2, 16, %g3
srl %o2, 16, %o2
andcc %o0, 0x4, %g0
or %g3, %o2, %o2
1: be cpa
andcc %o1, 0xffffff80, %o3
ld [%o0 + 0x00], %g2
sub %o1, 4, %o1
addcc %g2, %o2, %o2
add %o0, 4, %o0
addx %g0, %o2, %o2
b cpa
andcc %o1, 0xffffff80, %o3
/* The common case is to get called with a nicely aligned
* buffer of size 0x20. Follow the code path for that case.
*/
.globl csum_partial
EXPORT_SYMBOL(csum_partial)
csum_partial: /* %o0=buf, %o1=len, %o2=sum */
andcc %o0, 0x7, %g0 ! alignment problems?
bne csum_partial_fix_alignment ! yep, handle it
sethi %hi(cpte - 8), %g7 ! prepare table jmp ptr
andcc %o1, 0xffffff80, %o3 ! num loop iterations
cpa: be 3f ! none to do
andcc %o1, 0x70, %g1 ! clears carry flag too
5: CSUM_BIGCHUNK(%o0, 0x00, %o2, %o4, %o5, %g2, %g3, %g4, %g5)
CSUM_BIGCHUNK(%o0, 0x20, %o2, %o4, %o5, %g2, %g3, %g4, %g5)
CSUM_BIGCHUNK(%o0, 0x40, %o2, %o4, %o5, %g2, %g3, %g4, %g5)
CSUM_BIGCHUNK(%o0, 0x60, %o2, %o4, %o5, %g2, %g3, %g4, %g5)
addx %g0, %o2, %o2 ! sink in final carry
subcc %o3, 128, %o3 ! detract from loop iters
bne 5b ! more to do
add %o0, 128, %o0 ! advance buf ptr
andcc %o1, 0x70, %g1 ! clears carry flag too
3: be cpte ! nope
andcc %o1, 0xf, %g0 ! anything left at all?
srl %g1, 1, %o4 ! compute offset
sub %g7, %g1, %g7 ! adjust jmp ptr
sub %g7, %o4, %g7 ! final jmp ptr adjust
jmp %g7 + %lo(cpte - 8) ! enter the table
add %o0, %g1, %o0 ! advance buf ptr
cptbl: CSUM_LASTCHUNK(%o0, 0x68, %o2, %g2, %g3, %g4, %g5)
CSUM_LASTCHUNK(%o0, 0x58, %o2, %g2, %g3, %g4, %g5)
CSUM_LASTCHUNK(%o0, 0x48, %o2, %g2, %g3, %g4, %g5)
CSUM_LASTCHUNK(%o0, 0x38, %o2, %g2, %g3, %g4, %g5)
CSUM_LASTCHUNK(%o0, 0x28, %o2, %g2, %g3, %g4, %g5)
CSUM_LASTCHUNK(%o0, 0x18, %o2, %g2, %g3, %g4, %g5)
CSUM_LASTCHUNK(%o0, 0x08, %o2, %g2, %g3, %g4, %g5)
addx %g0, %o2, %o2 ! fetch final carry
andcc %o1, 0xf, %g0 ! anything left at all?
cpte: bne csum_partial_end_cruft ! yep, handle it
andcc %o1, 8, %g0 ! check how much
cpout: retl ! get outta here
mov %o2, %o0 ! return computed csum
.globl __csum_partial_copy_start, __csum_partial_copy_end
__csum_partial_copy_start:
/* Work around cpp -rob */
#define ALLOC #alloc
#define EXECINSTR #execinstr
#define EX(x,y,a,b) \
98: x,y; \
.section .fixup,ALLOC,EXECINSTR; \
.align 4; \
99: ba 30f; \
a, b, %o3; \
.section __ex_table,ALLOC; \
.align 4; \
.word 98b, 99b; \
.text; \
.align 4
#define EX2(x,y) \
98: x,y; \
.section __ex_table,ALLOC; \
.align 4; \
.word 98b, 30f; \
.text; \
.align 4
#define EX3(x,y) \
98: x,y; \
.section __ex_table,ALLOC; \
.align 4; \
.word 98b, 96f; \
.text; \
.align 4
#define EXT(start,end,handler) \
.section __ex_table,ALLOC; \
.align 4; \
.word start, 0, end, handler; \
.text; \
.align 4
/* This aligned version executes typically in 8.5 superscalar cycles, this
* is the best I can do. I say 8.5 because the final add will pair with
* the next ldd in the main unrolled loop. Thus the pipe is always full.
* If you change these macros (including order of instructions),
* please check the fixup code below as well.
*/
#define CSUMCOPY_BIGCHUNK_ALIGNED(src, dst, sum, off, t0, t1, t2, t3, t4, t5, t6, t7) \
ldd [src + off + 0x00], t0; \
ldd [src + off + 0x08], t2; \
addxcc t0, sum, sum; \
ldd [src + off + 0x10], t4; \
addxcc t1, sum, sum; \
ldd [src + off + 0x18], t6; \
addxcc t2, sum, sum; \
std t0, [dst + off + 0x00]; \
addxcc t3, sum, sum; \
std t2, [dst + off + 0x08]; \
addxcc t4, sum, sum; \
std t4, [dst + off + 0x10]; \
addxcc t5, sum, sum; \
std t6, [dst + off + 0x18]; \
addxcc t6, sum, sum; \
addxcc t7, sum, sum;
/* 12 superscalar cycles seems to be the limit for this case,
* because of this we thus do all the ldd's together to get
* Viking MXCC into streaming mode. Ho hum...
*/
#define CSUMCOPY_BIGCHUNK(src, dst, sum, off, t0, t1, t2, t3, t4, t5, t6, t7) \
ldd [src + off + 0x00], t0; \
ldd [src + off + 0x08], t2; \
ldd [src + off + 0x10], t4; \
ldd [src + off + 0x18], t6; \
st t0, [dst + off + 0x00]; \
addxcc t0, sum, sum; \
st t1, [dst + off + 0x04]; \
addxcc t1, sum, sum; \
st t2, [dst + off + 0x08]; \
addxcc t2, sum, sum; \
st t3, [dst + off + 0x0c]; \
addxcc t3, sum, sum; \
st t4, [dst + off + 0x10]; \
addxcc t4, sum, sum; \
st t5, [dst + off + 0x14]; \
addxcc t5, sum, sum; \
st t6, [dst + off + 0x18]; \
addxcc t6, sum, sum; \
st t7, [dst + off + 0x1c]; \
addxcc t7, sum, sum;
/* Yuck, 6 superscalar cycles... */
#define CSUMCOPY_LASTCHUNK(src, dst, sum, off, t0, t1, t2, t3) \
ldd [src - off - 0x08], t0; \
ldd [src - off - 0x00], t2; \
addxcc t0, sum, sum; \
st t0, [dst - off - 0x08]; \
addxcc t1, sum, sum; \
st t1, [dst - off - 0x04]; \
addxcc t2, sum, sum; \
st t2, [dst - off - 0x00]; \
addxcc t3, sum, sum; \
st t3, [dst - off + 0x04];
/* Handle the end cruft code out of band for better cache patterns. */
cc_end_cruft:
be 1f
andcc %o3, 4, %g0
EX(ldd [%o0 + 0x00], %g2, and %o3, 0xf)
add %o1, 8, %o1
addcc %g2, %g7, %g7
add %o0, 8, %o0
addxcc %g3, %g7, %g7
EX2(st %g2, [%o1 - 0x08])
addx %g0, %g7, %g7
andcc %o3, 4, %g0
EX2(st %g3, [%o1 - 0x04])
1: be 1f
andcc %o3, 3, %o3
EX(ld [%o0 + 0x00], %g2, add %o3, 4)
add %o1, 4, %o1
addcc %g2, %g7, %g7
EX2(st %g2, [%o1 - 0x04])
addx %g0, %g7, %g7
andcc %o3, 3, %g0
add %o0, 4, %o0
1: be 1f
addcc %o3, -1, %g0
bne 2f
subcc %o3, 2, %o3
b 4f
or %g0, %g0, %o4
2: EX(lduh [%o0 + 0x00], %o4, add %o3, 2)
add %o0, 2, %o0
EX2(sth %o4, [%o1 + 0x00])
be 6f
add %o1, 2, %o1
sll %o4, 16, %o4
4: EX(ldub [%o0 + 0x00], %o5, add %g0, 1)
EX2(stb %o5, [%o1 + 0x00])
sll %o5, 8, %o5
or %o5, %o4, %o4
6: addcc %o4, %g7, %g7
1: retl
addx %g0, %g7, %o0
/* Also, handle the alignment code out of band. */
cc_dword_align:
cmp %g1, 16
bge 1f
srl %g1, 1, %o3
2: cmp %o3, 0
be,a ccte
andcc %g1, 0xf, %o3
andcc %o3, %o0, %g0 ! Check %o0 only (%o1 has the same last 2 bits)
be,a 2b
srl %o3, 1, %o3
1: andcc %o0, 0x1, %g0
bne ccslow
andcc %o0, 0x2, %g0
be 1f
andcc %o0, 0x4, %g0
EX(lduh [%o0 + 0x00], %g4, add %g1, 0)
sub %g1, 2, %g1
EX2(sth %g4, [%o1 + 0x00])
add %o0, 2, %o0
sll %g4, 16, %g4
addcc %g4, %g7, %g7
add %o1, 2, %o1
srl %g7, 16, %g3
addx %g0, %g3, %g4
sll %g7, 16, %g7
sll %g4, 16, %g3
srl %g7, 16, %g7
andcc %o0, 0x4, %g0
or %g3, %g7, %g7
1: be 3f
andcc %g1, 0xffffff80, %g0
EX(ld [%o0 + 0x00], %g4, add %g1, 0)
sub %g1, 4, %g1
EX2(st %g4, [%o1 + 0x00])
add %o0, 4, %o0
addcc %g4, %g7, %g7
add %o1, 4, %o1
addx %g0, %g7, %g7
b 3f
andcc %g1, 0xffffff80, %g0
/* Sun, you just can't beat me, you just can't. Stop trying,
* give up. I'm serious, I am going to kick the living shit
* out of you, game over, lights out.
*/
.align 8
.globl __csum_partial_copy_sparc_generic
EXPORT_SYMBOL(__csum_partial_copy_sparc_generic)
__csum_partial_copy_sparc_generic:
/* %o0=src, %o1=dest, %g1=len, %g7=sum */
xor %o0, %o1, %o4 ! get changing bits
andcc %o4, 3, %g0 ! check for mismatched alignment
bne ccslow ! better this than unaligned/fixups
andcc %o0, 7, %g0 ! need to align things?
bne cc_dword_align ! yes, we check for short lengths there
andcc %g1, 0xffffff80, %g0 ! can we use unrolled loop?
3: be 3f ! nope, less than one loop remains
andcc %o1, 4, %g0 ! dest aligned on 4 or 8 byte boundary?
be ccdbl + 4 ! 8 byte aligned, kick ass
5: CSUMCOPY_BIGCHUNK(%o0,%o1,%g7,0x00,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
CSUMCOPY_BIGCHUNK(%o0,%o1,%g7,0x20,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
CSUMCOPY_BIGCHUNK(%o0,%o1,%g7,0x40,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
CSUMCOPY_BIGCHUNK(%o0,%o1,%g7,0x60,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
10: EXT(5b, 10b, 20f) ! note for exception handling
sub %g1, 128, %g1 ! detract from length
addx %g0, %g7, %g7 ! add in last carry bit
andcc %g1, 0xffffff80, %g0 ! more to csum?
add %o0, 128, %o0 ! advance src ptr
bne 5b ! we did not go negative, continue looping
add %o1, 128, %o1 ! advance dest ptr
3: andcc %g1, 0x70, %o2 ! can use table?
ccmerge:be ccte ! nope, go and check for end cruft
andcc %g1, 0xf, %o3 ! get low bits of length (clears carry btw)
srl %o2, 1, %o4 ! begin negative offset computation
sethi %hi(12f), %o5 ! set up table ptr end
add %o0, %o2, %o0 ! advance src ptr
sub %o5, %o4, %o5 ! continue table calculation
sll %o2, 1, %g2 ! constant multiplies are fun...
sub %o5, %g2, %o5 ! some more adjustments
jmp %o5 + %lo(12f) ! jump into it, duff style, wheee...
add %o1, %o2, %o1 ! advance dest ptr (carry is clear btw)
cctbl: CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x68,%g2,%g3,%g4,%g5)
CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x58,%g2,%g3,%g4,%g5)
CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x48,%g2,%g3,%g4,%g5)
CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x38,%g2,%g3,%g4,%g5)
CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x28,%g2,%g3,%g4,%g5)
CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x18,%g2,%g3,%g4,%g5)
CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x08,%g2,%g3,%g4,%g5)
12: EXT(cctbl, 12b, 22f) ! note for exception table handling
addx %g0, %g7, %g7
andcc %o3, 0xf, %g0 ! check for low bits set
ccte: bne cc_end_cruft ! something left, handle it out of band
andcc %o3, 8, %g0 ! begin checks for that code
retl ! return
mov %g7, %o0 ! give em the computed checksum
ccdbl: CSUMCOPY_BIGCHUNK_ALIGNED(%o0,%o1,%g7,0x00,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
CSUMCOPY_BIGCHUNK_ALIGNED(%o0,%o1,%g7,0x20,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
CSUMCOPY_BIGCHUNK_ALIGNED(%o0,%o1,%g7,0x40,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
CSUMCOPY_BIGCHUNK_ALIGNED(%o0,%o1,%g7,0x60,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
11: EXT(ccdbl, 11b, 21f) ! note for exception table handling
sub %g1, 128, %g1 ! detract from length
addx %g0, %g7, %g7 ! add in last carry bit
andcc %g1, 0xffffff80, %g0 ! more to csum?
add %o0, 128, %o0 ! advance src ptr
bne ccdbl ! we did not go negative, continue looping
add %o1, 128, %o1 ! advance dest ptr
b ccmerge ! finish it off, above
andcc %g1, 0x70, %o2 ! can use table? (clears carry btw)
ccslow: cmp %g1, 0
mov 0, %g5
bleu 4f
andcc %o0, 1, %o5
be,a 1f
srl %g1, 1, %g4
sub %g1, 1, %g1
EX(ldub [%o0], %g5, add %g1, 1)
add %o0, 1, %o0
EX2(stb %g5, [%o1])
srl %g1, 1, %g4
add %o1, 1, %o1
1: cmp %g4, 0
be,a 3f
andcc %g1, 1, %g0
andcc %o0, 2, %g0
be,a 1f
srl %g4, 1, %g4
EX(lduh [%o0], %o4, add %g1, 0)
sub %g1, 2, %g1
srl %o4, 8, %g2
sub %g4, 1, %g4
EX2(stb %g2, [%o1])
add %o4, %g5, %g5
EX2(stb %o4, [%o1 + 1])
add %o0, 2, %o0
srl %g4, 1, %g4
add %o1, 2, %o1
1: cmp %g4, 0
be,a 2f
andcc %g1, 2, %g0
EX3(ld [%o0], %o4)
5: srl %o4, 24, %g2
srl %o4, 16, %g3
EX2(stb %g2, [%o1])
srl %o4, 8, %g2
EX2(stb %g3, [%o1 + 1])
add %o0, 4, %o0
EX2(stb %g2, [%o1 + 2])
addcc %o4, %g5, %g5
EX2(stb %o4, [%o1 + 3])
addx %g5, %g0, %g5 ! I am now to lazy to optimize this (question it
add %o1, 4, %o1 ! is worthy). Maybe some day - with the sll/srl
subcc %g4, 1, %g4 ! tricks
bne,a 5b
EX3(ld [%o0], %o4)
sll %g5, 16, %g2
srl %g5, 16, %g5
srl %g2, 16, %g2
andcc %g1, 2, %g0
add %g2, %g5, %g5
2: be,a 3f
andcc %g1, 1, %g0
EX(lduh [%o0], %o4, and %g1, 3)
andcc %g1, 1, %g0
srl %o4, 8, %g2
add %o0, 2, %o0
EX2(stb %g2, [%o1])
add %g5, %o4, %g5
EX2(stb %o4, [%o1 + 1])
add %o1, 2, %o1
3: be,a 1f
sll %g5, 16, %o4
EX(ldub [%o0], %g2, add %g0, 1)
sll %g2, 8, %o4
EX2(stb %g2, [%o1])
add %g5, %o4, %g5
sll %g5, 16, %o4
1: addcc %o4, %g5, %g5
srl %g5, 16, %o4
addx %g0, %o4, %g5
orcc %o5, %g0, %g0
be 4f
srl %g5, 8, %o4
and %g5, 0xff, %g2
and %o4, 0xff, %o4
sll %g2, 8, %g2
or %g2, %o4, %g5
4: addcc %g7, %g5, %g7
retl
addx %g0, %g7, %o0
__csum_partial_copy_end:
/* We do these strange calculations for the csum_*_from_user case only, ie.
* we only bother with faults on loads... */
/* o2 = ((g2%20)&3)*8
* o3 = g1 - (g2/20)*32 - o2 */
20:
cmp %g2, 20
blu,a 1f
and %g2, 3, %o2
sub %g1, 32, %g1
b 20b
sub %g2, 20, %g2
1:
sll %o2, 3, %o2
b 31f
sub %g1, %o2, %o3
/* o2 = (!(g2 & 15) ? 0 : (((g2 & 15) + 1) & ~1)*8)
* o3 = g1 - (g2/16)*32 - o2 */
21:
andcc %g2, 15, %o3
srl %g2, 4, %g2
be,a 1f
clr %o2
add %o3, 1, %o3
and %o3, 14, %o3
sll %o3, 3, %o2
1:
sll %g2, 5, %g2
sub %g1, %g2, %o3
b 31f
sub %o3, %o2, %o3
/* o0 += (g2/10)*16 - 0x70
* 01 += (g2/10)*16 - 0x70
* o2 = (g2 % 10) ? 8 : 0
* o3 += 0x70 - (g2/10)*16 - o2 */
22:
cmp %g2, 10
blu,a 1f
sub %o0, 0x70, %o0
add %o0, 16, %o0
add %o1, 16, %o1
sub %o3, 16, %o3
b 22b
sub %g2, 10, %g2
1:
sub %o1, 0x70, %o1
add %o3, 0x70, %o3
clr %o2
tst %g2
bne,a 1f
mov 8, %o2
1:
b 31f
sub %o3, %o2, %o3
96:
and %g1, 3, %g1
sll %g4, 2, %g4
add %g1, %g4, %o3
30:
/* %o1 is dst
* %o3 is # bytes to zero out
* %o4 is faulting address
* %o5 is %pc where fault occurred */
clr %o2
31:
/* %o0 is src
* %o1 is dst
* %o2 is # of bytes to copy from src to dst
* %o3 is # bytes to zero out
* %o4 is faulting address
* %o5 is %pc where fault occurred */
save %sp, -104, %sp
mov %i5, %o0
mov %i7, %o1
mov %i4, %o2
call lookup_fault
mov %g7, %i4
cmp %o0, 2
bne 1f
add %g0, -EFAULT, %i5
tst %i2
be 2f
mov %i0, %o1
mov %i1, %o0
5:
call memcpy
mov %i2, %o2
tst %o0
bne,a 2f
add %i3, %i2, %i3
add %i1, %i2, %i1
2:
mov %i1, %o0
6:
call __bzero
mov %i3, %o1
1:
ld [%sp + 168], %o2 ! struct_ptr of parent
st %i5, [%o2]
ret
restore
.section __ex_table,#alloc
.align 4
.word 5b,2
.word 6b,2
Reference in New Issue View Git Blame Copy Permalink