diff options
| author | Martin Storsjö <martin@martin.st> | 2013-04-23 18:35:53 +0200 |
|---|---|---|
| committer | Niels Möller <nisse@lysator.liu.se> | 2013-04-23 18:40:10 +0200 |
| commit | 05964538d40285e632d848b5f7d1f5a1e481a9d4 (patch) | |
| tree | 04e5c655f1fc98e5ca0714f98e042d1a71337f98 | |
| parent | 508a312a34c9c91c88f37bfbcceac6fd1bb4337e (diff) | |
| download | nettle-05964538d40285e632d848b5f7d1f5a1e481a9d4.tar.gz | |
Use movdqu instead of movdqa for saving xmm registers
The stack is not guaranteed to be 16-byte aligned on win64.
| -rw-r--r-- | ChangeLog | 3 | ||||
| -rw-r--r-- | x86_64/machine.m4 | 40 |
2 files changed, 23 insertions, 20 deletions
@@ -1,6 +1,9 @@ 2013-04-23 Niels Möller <nisse@lysator.liu.se> From Martin Storsjö: + * x86_64/machine.m4 (W64_ENTRY): Use movdqu instead of movdqa for + saving xmm registers, since the stack is not guaranteed to be + 16-byte aligned on win64. * Makefile.in: Consistently use EXEEXT_FOR_BUILD. 2013-04-21 Niels Möller <nisse@lysator.liu.se> diff --git a/x86_64/machine.m4 b/x86_64/machine.m4 index dc23dde8..d5d5b37c 100644 --- a/x86_64/machine.m4 +++ b/x86_64/machine.m4 @@ -71,34 +71,34 @@ define(<W64_ENTRY>, < ifelse(W64_ABI,yes,[ ifelse(eval($2 > 6), 1, [ sub [$]eval(8 + 16*($2 - 6)), %rsp - movdqa %xmm6, 0(%rsp) + movdqu %xmm6, 0(%rsp) ]) ifelse(eval($2 > 7), 1, [ - movdqa %xmm7, 16(%rsp) + movdqu %xmm7, 16(%rsp) ]) ifelse(eval($2 > 8), 1, [ - movdqa %xmm8, 32(%rsp) + movdqu %xmm8, 32(%rsp) ]) ifelse(eval($2 > 9), 1, [ - movdqa %xmm9, 48(%rsp) + movdqu %xmm9, 48(%rsp) ]) ifelse(eval($2 > 10), 1, [ - movdqa %xmm10, 64(%rsp) + movdqu %xmm10, 64(%rsp) ]) ifelse(eval($2 > 11), 1, [ - movdqa %xmm11, 80(%rsp) + movdqu %xmm11, 80(%rsp) ]) ifelse(eval($2 > 12), 1, [ - movdqa %xmm12, 96(%rsp) + movdqu %xmm12, 96(%rsp) ]) ifelse(eval($2 > 13), 1, [ - movdqa %xmm13, 112(%rsp) + movdqu %xmm13, 112(%rsp) ]) ifelse(eval($2 > 14), 1, [ - movdqa %xmm14, 128(%rsp) + movdqu %xmm14, 128(%rsp) ]) ifelse(eval($2 > 15), 1, [ - movdqa %xmm15, 144(%rsp) + movdqu %xmm15, 144(%rsp) ]) ifelse(eval($1 >= 1), 1, [ push %rdi @@ -133,34 +133,34 @@ define(<W64_EXIT>, < pop %rdi ]) ifelse(eval($2 > 15), 1, [ - movdqa 144(%rsp), %xmm15 + movdqu 144(%rsp), %xmm15 ]) ifelse(eval($2 > 14), 1, [ - movdqa 128(%rsp), %xmm14 + movdqu 128(%rsp), %xmm14 ]) ifelse(eval($2 > 13), 1, [ - movdqa 112(%rsp), %xmm13 + movdqu 112(%rsp), %xmm13 ]) ifelse(eval($2 > 12), 1, [ - movdqa 96(%rsp), %xmm12 + movdqu 96(%rsp), %xmm12 ]) ifelse(eval($2 > 11), 1, [ - movdqa 80(%rsp), %xmm11 + movdqu 80(%rsp), %xmm11 ]) ifelse(eval($2 > 10), 1, [ - movdqa 64(%rsp), %xmm10 + movdqu 64(%rsp), %xmm10 ]) ifelse(eval($2 > 9), 1, [ - movdqa 48(%rsp), %xmm9 + movdqu 48(%rsp), %xmm9 ]) ifelse(eval($2 > 8), 1, [ - movdqa 32(%rsp), %xmm8 + movdqu 32(%rsp), %xmm8 ]) ifelse(eval($2 > 7), 1, [ - movdqa 16(%rsp), %xmm7 + movdqu 16(%rsp), %xmm7 ]) ifelse(eval($2 > 6), 1, [ - movdqa 0(%rsp), %xmm6 + movdqu 0(%rsp), %xmm6 add [$]eval(8 + 16*($2 - 6)), %rsp ]) ]) |