diff options
| author | Peter Varga <pvarga@inf.u-szeged.hu> | 2013-06-25 15:53:03 +0200 |
|---|---|---|
| committer | The Qt Project <gerrit-noreply@qt-project.org> | 2013-06-25 19:19:52 +0200 |
| commit | 91f936a862644c7ca7a1762ad05cc91d263b808e (patch) | |
| tree | a2979ec732eb9b896333d960b0a8cf1f9c87756a | |
| parent | 6f2625e0e0e2818af6fdeb69528ada18556daef2 (diff) | |
| download | qtjsbackend-91f936a862644c7ca7a1762ad05cc91d263b808e.tar.gz | |
[V8] Fix stack alignment corruption for MinGW32 build
BUG: https://bugreports.qt-project.org/browse/QTBUG-31396
Change-Id: Ibf815ab7c1f679b01b347f7141d269a220657968
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@digia.com>
| -rw-r--r-- | src/3rdparty/v8/src/platform-win32.cc | 7 |
1 files changed, 5 insertions, 2 deletions
diff --git a/src/3rdparty/v8/src/platform-win32.cc b/src/3rdparty/v8/src/platform-win32.cc index 76e35f5..ae9ab2a 100644 --- a/src/3rdparty/v8/src/platform-win32.cc +++ b/src/3rdparty/v8/src/platform-win32.cc @@ -1523,9 +1523,12 @@ double OS::nan_value() { int OS::ActivationFrameAlignment() { #ifdef _WIN64 return 16; // Windows 64-bit ABI requires the stack to be 16-byte aligned. -#else - return 8; // Floating-point math runs faster with 8-byte alignment. +#elif defined(__MINGW32__) + // With gcc 4.4 the tree vectorization optimizer can generate code + // that requires 16 byte alignment such as movdqa on x86. + return 16; #endif + return 8; // Floating-point math runs faster with 8-byte alignment. } |