darkkirb/old-cross-binutils
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Fix AMDFAM10 POPCNT instruction

Browse source
This commit is contained in:
Michael Meissner 2006-10-23 22:53:29 +00:00
parent a3ffa599fe
commit 7918206c55
9 changed files with 63 additions and 31 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
gas/ChangeLog
View file

@ -1,3 +1,7 @@
2006-10-23 Dwarakanath Rajagopal <dwarak.rajagopal@amd.com>
* doc/c-i386.texi : Document amdfam10,.sse4a and .abm in cpu_type.
2006-10-23 Alan Modra <amodra@bigpond.net.au>
* config/tc-m68hc11.c (md_assemble): Quiet warning.

5
gas/doc/c-i386.texi
View file

@ -753,10 +753,11 @@ supported on the CPU specified. The choices for @var{cpu_type} are:
@item @samp{i486} @tab @samp{i586} @tab @samp{i686} @tab @samp{pentium}
@item @samp{pentiumpro} @tab @samp{pentiumii} @tab @samp{pentiumiii} @tab @samp{pentium4}
@item @samp{prescott} @tab @samp{nocona} @tab @samp{core} @tab @samp{core2}
@item @samp{amdfam10}
@item @samp{k6} @tab @samp{athlon} @tab @samp{sledgehammer} @tab @samp{k8}
@item @samp{.mmx} @tab @samp{.sse} @tab @samp{.sse2} @tab @samp{.sse3}
@item @samp{.3dnow} @tab @samp{.3dnowa} @tab @samp{.padlock} @tab @samp{.pacifica}
@item @samp{.svme}
@item @samp{.sse4a} @tab @samp{.3dnow} @tab @samp{.3dnowa} @tab @samp{.padlock}
@item @samp{.pacifica} @tab @samp{.svme} @tab @samp{.abm}
@end multitable
Apart from the warning, there are only two other effects on

6
gas/testsuite/ChangeLog
View file

@ -1,3 +1,9 @@
2006-10-23 Dwarakanath Rajagopal <dwarak.rajagopal@amd.com>
* gas/i386/amdfam10.d : Modify to support for the change in POPCNT
opcode in amdfam10 architecture.
* gas/i386/x86-64-amdfam10.d : Ditto.
2006-10-21 Kaz Kojima <kkojima@rr.iij4u.or.jp>
* gas/sh/sh64/syntax-1.d: Update.

23
gas/testsuite/gas/i386/amdfam10.d
View file

@ -10,15 +10,14 @@ Disassembly of section .text:
4: f3 66 0f bd 19[ ]+lzcnt \(%ecx\),%bx
9: f3 0f bd d9[ ]+lzcnt %ecx,%ebx
d: f3 66 0f bd d9[ ]+lzcnt %cx,%bx
12: 0f b8 19[ ]+popcnt \(%ecx\),%ebx
15: 66 0f b8 19[ ]+popcnt \(%ecx\),%bx
19: 0f b8 d9[ ]+popcnt %ecx,%ebx
1c: 66 0f b8 d9[ ]+popcnt %cx,%bx
20: 66 0f 79 ca[ ]+extrq %xmm2,%xmm1
24: 66 0f 78 c1 02 04[ ]*extrq \$0x4,\$0x2,%xmm1
2a: f2 0f 79 ca[ ]+insertq %xmm2,%xmm1
2e: f2 0f 78 ca 02 04[ ]*insertq \$0x4,\$0x2,%xmm2,%xmm1
34: f2 0f 2b 09[ ]+movntsd %xmm1,\(%ecx\)
38: f3 0f 2b 09[ ]+movntss %xmm1,\(%ecx\)
3c: 00 00 [ ]+add %al,\(%eax\)
...
12: f3 0f b8 19[ ]+popcnt \(%ecx\),%ebx
16: f3 66 0f b8 19[ ]+popcnt \(%ecx\),%bx
1b: f3 0f b8 d9[ ]+popcnt %ecx,%ebx
1f: f3 66 0f b8 d9[ ]+popcnt %cx,%bx
24: 66 0f 79 ca[ ]+extrq %xmm2,%xmm1
28: 66 0f 78 c1 02 04[ ]*extrq \$0x4,\$0x2,%xmm1
2e: f2 0f 79 ca[ ]+insertq %xmm2,%xmm1
32: f2 0f 78 ca 02 04[ ]*insertq \$0x4,\$0x2,%xmm2,%xmm1
38: f2 0f 2b 09[ ]+movntsd %xmm1,\(%ecx\)
3c: f3 0f 2b 09[ ]+movntss %xmm1,\(%ecx\)

24
gas/testsuite/gas/i386/x86-64-amdfam10.d
View file

@ -12,16 +12,16 @@ Disassembly of section .text:
e: f3 48 0f bd d9[ ]+lzcnt %rcx,%rbx
13: f3 0f bd d9[ ]+lzcnt %ecx,%ebx
17: f3 66 0f bd d9[ ]+lzcnt %cx,%bx
1c: 48 0f b8 19[ ]+popcnt \(%rcx\),%rbx
20: 0f b8 19[ ]+popcnt \(%rcx\),%ebx
23: 66 0f b8 19[ ]+popcnt \(%rcx\),%bx
27: 48 0f b8 d9[ ]+popcnt %rcx,%rbx
2b: 0f b8 d9[ ]+popcnt %ecx,%ebx
2e: 66 0f b8 d9[ ]+popcnt %cx,%bx
32: 66 0f 79 ca[ ]+extrq %xmm2,%xmm1
36: 66 0f 78 c1 02 04[ ]+extrq \$0x4,\$0x2,%xmm1
3c: f2 0f 79 ca[ ]+insertq %xmm2,%xmm1
40: f2 0f 78 ca 02 04[ ]+insertq \$0x4,\$0x2,%xmm2,%xmm1
46: f2 0f 2b 09[ ]+movntsd %xmm1,\(%rcx\)
4a: f3 0f 2b 09[ ]+movntss %xmm1,\(%rcx\)
1c: f3 48 0f b8 19[ ]+popcnt \(%rcx\),%rbx
21: f3 0f b8 19[ ]+popcnt \(%rcx\),%ebx
25: f3 66 0f b8 19[ ]+popcnt \(%rcx\),%bx
2a: f3 48 0f b8 d9[ ]+popcnt %rcx,%rbx
2f: f3 0f b8 d9[ ]+popcnt %ecx,%ebx
33: f3 66 0f b8 d9[ ]+popcnt %cx,%bx
38: 66 0f 79 ca[ ]+extrq %xmm2,%xmm1
3c: 66 0f 78 c1 02 04[ ]+extrq \$0x4,\$0x2,%xmm1
42: f2 0f 79 ca[ ]+insertq %xmm2,%xmm1
46: f2 0f 78 ca 02 04[ ]+insertq \$0x4,\$0x2,%xmm2,%xmm1
4c: f2 0f 2b 09[ ]+movntsd %xmm1,\(%rcx\)
50: f3 0f 2b 09[ ]+movntss %xmm1,\(%rcx\)
...

5
include/opcode/ChangeLog
View file

@ -1,3 +1,8 @@
2006-10-23 Dwarakanath Rajagopal <dwarak.rajagopal@amd.com>
* i386.h : Modify opcode to support for the change in POPCNT opcode
in amdfam10 architecture.
2006-09-28 H.J. Lu <hongjiu.lu@intel.com>
* i386.h: Replace CpuMNI with CpuSSSE3.

2
include/opcode/i386.h
View file

@ -1472,7 +1472,7 @@ static const template i386_optab[] =
{"insertq", 4, 0xf20f78, X, CpuSSE4a, NoSuf|IgnoreSize|Modrm, { Imm8, Imm8, RegXMM, RegXMM} },
/* ABM instructions */
{"popcnt", 2, 0x0fb8, X, CpuABM, wlq_Suf|Modrm, { WordReg|WordMem, WordReg, 0} },
{"popcnt", 2, 0xf30fb8, X, CpuABM, wlq_Suf|Modrm, { WordReg|WordMem, WordReg, 0} },
{"lzcnt", 2, 0xf30fbd, X, CpuABM, wlq_Suf|Modrm, { WordReg|WordMem, WordReg, 0} },

7
opcodes/ChangeLog
View file

@ -1,3 +1,10 @@
2006-10-23 Dwarakanath Rajagopal <dwarak.rajagopal@amd.com>
* i386-dis.c (dis386): Add support for the change in POPCNT opcode in
amdfam10 architecture.
(PREGRP37): NEW.
(print_insn): Disallow REP prefix for POPCNT.
2006-10-20 Andrew Stubbs <andrew.stubbs@st.com>
* sh-dis.c (print_insn_sh): Remove 0x from output to prevent GDB

18
opcodes/i386-dis.c
View file

@ -471,6 +471,8 @@ fetch_data (struct disassemble_info *info, bfd_byte *addr)
#define PREGRP34 NULL, NULL, USE_PREFIX_USER_TABLE, NULL, 34, NULL, 0, NULL, 0
#define PREGRP35 NULL, NULL, USE_PREFIX_USER_TABLE, NULL, 35, NULL, 0, NULL, 0
#define PREGRP36 NULL, NULL, USE_PREFIX_USER_TABLE, NULL, 36, NULL, 0, NULL, 0
#define PREGRP37 NULL, NULL, USE_PREFIX_USER_TABLE, NULL, 37, NULL, 0, NULL, 0
#define X86_64_0 NULL, NULL, X86_64_SPECIAL, NULL, 0, NULL, 0, NULL, 0
@ -1028,7 +1030,7 @@ static const struct dis386 dis386_twobyte[] = {
{ "movz{bR|x|bR|x}", Gv, Eb, XX, XX },
{ "movz{wR|x|wR|x}", Gv, Ew, XX, XX }, /* yes, there really is movzww ! */
/* b8 */
{ "popcntS", Gv, Ev, XX, XX },
{ PREGRP37 },
{ "ud2b", XX, XX, XX, XX },
{ GRP8 },
{ "btcS", Ev, Gv, XX, XX },
@ -1820,6 +1822,13 @@ static const struct dis386 prefix_user_table[][4] = {
{ "(bad)", XX, XX, XX, XX },
},
/* PREGRP37 */
{
{ "(bad)", XX, XX, XX, XX },
{ "popcntS",Gv, Ev, XX, XX },
{ "(bad)", XX, XX, XX, XX },
{ "(bad)", XX, XX, XX, XX },
},
};
static const struct dis386 x86_64_table[][2] = {
@ -2827,13 +2836,14 @@ print_insn (bfd_vma pc, disassemble_info *info)
uses_LOCK_prefix = 0;
}
/*"lzcnt"=0xBD is the only non-sse instruction which uses F3 in the opcode without any "rep(z|nz)"*/
if (!uses_SSE_prefix && (prefixes & PREFIX_REPZ) && *codep !=0xBD)
/*"lzcnt"=0xBD and "popcnt"=0xB8 are the only two non-sse
instruction which uses F3 in the opcode without any "rep(z|nz)"*/
if (!uses_SSE_prefix && (prefixes & PREFIX_REPZ) && *codep != 0xBD && *codep != 0xB8)
{
oappend ("repz ");
used_prefixes |= PREFIX_REPZ;
}
if (!uses_SSE_prefix && (prefixes & PREFIX_REPNZ) && *codep !=0xBD)
if (!uses_SSE_prefix && (prefixes & PREFIX_REPNZ) && *codep != 0xBD && *codep != 0xB8)
{
oappend ("repnz ");
used_prefixes |= PREFIX_REPNZ;