Fix stack alignment corruption for MinGW32 build

Fix stack alignment corruption for MinGW32 build

Contributed by Peter Varga <pvarga@inf.u-szeged.hu>;

BUG=
TEST=cctest/test-assembler-ia32/StackAlignmentForSSE2,cctest/test-assembler-x64/StackAlignmentForSSE2,cctest/test-platform/StackAlignment
R=jkummerow@chromium.org

Committed: http://code.google.com/p/v8/source/detail?r=15502

[Jakob Kummerow, 2013-07-02 15:24:18]

Mostly LGTM, I'll land this for you if you address my comments. Most of the
comments on test-assembler-ia32.cc apply to x64 as well.

https://codereview.chromium.org/18300003/diff/1/src/platform-win32.cc
File src/platform-win32.cc (right):

https://codereview.chromium.org/18300003/diff/1/src/platform-win32.cc#newcode...
src/platform-win32.cc:1486: #endif
Let's use an #else here, and keep the #endif at the end of the functions. Some
compilers don't like unreachable code after unconditional early returns.

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
File test/cctest/test-assembler-ia32.cc (right):

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:490: if (CpuFeatures::IsSupported(SSE2)) {
I think you should just ASSERT(CpuFeatures::IsSupported(SSE2)) at the beginning
of this function to make it clear that it must not be called when SSE2 is not
supported.

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:493: // Remove return address from the stack
for fix stack frame alignment
nit: All comments should end with appropriate punctuation, e.g. a full stop.
(This applies to pretty much all comments in this patch.)

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:498: __
push(Immediate(vec->Get(i)->Int32Value()));
nit: {} please

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:525: Code::cast(code)->Print();
debugging leftover?

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:528: ::printf("f() = %d\n", res);
debugging leftover?

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:545: "function foo(vec) {"
nit: indentation (4 spaces when you break inside a call's arguments list)

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:546: " return do_sse2(vec);"
nit: indentation (2 spaces inside function body, i.e.: "  return ...)

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:551:
v8::Local<v8::Function>::Cast(global_object->Get(v8_str("foo")));
nit: indentation (4 spaces when you break after '=')

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:556: v8_vec->Set(i, v8_num(vec[i]));
nit: {} please

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-assembler-ia...
test/cctest/test-assembler-ia32.cc:568: 
nit: two empty lines is enough

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-platform.cc
File test/cctest/test-platform.cc (right):

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-platform.cc#...
test/cctest/test-platform.cc:84: "function foo() {"
nit: indentation

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-platform.cc#...
test/cctest/test-platform.cc:85: " return get_stack_pointer();"
nit: indentation of 'return'

https://codereview.chromium.org/18300003/diff/1/test/cctest/test-platform.cc#...
test/cctest/test-platform.cc:90:
v8::Local<v8::Function>::Cast(global_object->Get(v8_str("foo")));
nit: indentation

[Peter Varga, 2013-07-03 08:23:31]

I've fixed the style issues that you pointed out. Some of them come from
copy-pasting code from other part of the V8 source. If I missed or misunderstood
something feel free to notice me, I will fix it ASAP.

[Jakob Kummerow, 2013-07-03 15:02:42]

Thanks, LGTM. I'll land it for you.

[Jakob Kummerow, 2013-07-03 15:34:56]

Committed patchset #2 manually as r15480 (presubmit successful).

[Peter Varga, 2013-07-04 07:31:38]

I suppose the compilation of this change fails with MSVC due to GNU specific
snippets in test-platform.cc (https://codereview.chromium.org/18649002/). I've
guarded it in the new patch set.

[Jakob Kummerow, 2013-07-05 08:18:29]

OK, let's give it another shot.

[Jakob Kummerow, 2013-07-05 08:34:23]

Committed patchset #3 manually as r15502 (presubmit successful).