Title: [160063] trunk/Source
- Revision
- 160063
- Author
- mark....@apple.com
- Date
- 2013-12-03 18:19:09 -0800 (Tue, 03 Dec 2013)
Log Message
testapi test crashes on Windows in WTF::Vector<wchar_t,64,WTF::UnsafeVectorOverflow>::size().
https://bugs.webkit.org/show_bug.cgi?id=121972.
Reviewed by Brent Fulgham.
Source/_javascript_Core:
* interpreter/JSStack.cpp:
(JSC::JSStack::~JSStack):
- Reverting the change from r160004 since it's better to fix OSAllocatorWin
to be consistent with OSAllocatorPosix.
Source/WTF:
* wtf/OSAllocatorWin.cpp:
(WTF::OSAllocator::decommit):
(WTF::OSAllocator::releaseDecommitted):
- Added a check to ensure that the bytes to decommit / release is not 0.
On Windows, a 0 length passed to VirtualFree() has a special meaning,
and it's not "decommit / release nothing" as one would expect. Adding
this check makes OSAllocatorWin consistent with OSAllocatorPosix for
these 2 functions.
Modified Paths
Diff
Modified: trunk/Source/_javascript_Core/ChangeLog (160062 => 160063)
--- trunk/Source/_javascript_Core/ChangeLog 2013-12-04 01:18:19 UTC (rev 160062)
+++ trunk/Source/_javascript_Core/ChangeLog 2013-12-04 02:19:09 UTC (rev 160063)
@@ -1,5 +1,17 @@
2013-12-03 Mark Lam <mark....@apple.com>
+ testapi test crashes on Windows in WTF::Vector<wchar_t,64,WTF::UnsafeVectorOverflow>::size().
+ https://bugs.webkit.org/show_bug.cgi?id=121972.
+
+ Reviewed by Brent Fulgham.
+
+ * interpreter/JSStack.cpp:
+ (JSC::JSStack::~JSStack):
+ - Reverting the change from r160004 since it's better to fix OSAllocatorWin
+ to be consistent with OSAllocatorPosix.
+
+2013-12-03 Mark Lam <mark....@apple.com>
+
Fix LLINT_C_LOOP build for Win64.
https://bugs.webkit.org/show_bug.cgi?id=125186.
Modified: trunk/Source/_javascript_Core/interpreter/JSStack.cpp (160062 => 160063)
--- trunk/Source/_javascript_Core/interpreter/JSStack.cpp 2013-12-04 01:18:19 UTC (rev 160062)
+++ trunk/Source/_javascript_Core/interpreter/JSStack.cpp 2013-12-04 02:19:09 UTC (rev 160063)
@@ -63,10 +63,8 @@
JSStack::~JSStack()
{
void* highAddress = reinterpret_cast<void*>(static_cast<char*>(m_reservation.base()) + m_reservation.size());
- if (highAddress > m_commitEnd) {
- m_reservation.decommit(reinterpret_cast<void*>(m_commitEnd), reinterpret_cast<intptr_t>(highAddress) - reinterpret_cast<intptr_t>(m_commitEnd));
- addToCommittedByteCount(-(reinterpret_cast<intptr_t>(highAddress) - reinterpret_cast<intptr_t>(m_commitEnd)));
- }
+ m_reservation.decommit(reinterpret_cast<void*>(m_commitEnd), reinterpret_cast<intptr_t>(highAddress) - reinterpret_cast<intptr_t>(m_commitEnd));
+ addToCommittedByteCount(-(reinterpret_cast<intptr_t>(highAddress) - reinterpret_cast<intptr_t>(m_commitEnd)));
m_reservation.deallocate();
}
Modified: trunk/Source/WTF/ChangeLog (160062 => 160063)
--- trunk/Source/WTF/ChangeLog 2013-12-04 01:18:19 UTC (rev 160062)
+++ trunk/Source/WTF/ChangeLog 2013-12-04 02:19:09 UTC (rev 160063)
@@ -1,3 +1,19 @@
+2013-12-03 Mark Lam <mark....@apple.com>
+
+ testapi test crashes on Windows in WTF::Vector<wchar_t,64,WTF::UnsafeVectorOverflow>::size().
+ https://bugs.webkit.org/show_bug.cgi?id=121972.
+
+ Reviewed by Brent Fulgham.
+
+ * wtf/OSAllocatorWin.cpp:
+ (WTF::OSAllocator::decommit):
+ (WTF::OSAllocator::releaseDecommitted):
+ - Added a check to ensure that the bytes to decommit / release is not 0.
+ On Windows, a 0 length passed to VirtualFree() has a special meaning,
+ and it's not "decommit / release nothing" as one would expect. Adding
+ this check makes OSAllocatorWin consistent with OSAllocatorPosix for
+ these 2 functions.
+
2013-12-02 Mark Lam <mark....@apple.com>
Build failure when disabling JIT, YARR_JIT, and ASSEMBLER.
Modified: trunk/Source/WTF/wtf/OSAllocatorWin.cpp (160062 => 160063)
--- trunk/Source/WTF/wtf/OSAllocatorWin.cpp 2013-12-04 01:18:19 UTC (rev 160062)
+++ trunk/Source/WTF/wtf/OSAllocatorWin.cpp 2013-12-04 02:19:09 UTC (rev 160063)
@@ -65,6 +65,14 @@
void OSAllocator::decommit(void* address, size_t bytes)
{
+ // According to http://msdn.microsoft.com/en-us/library/aa366892(VS.85).aspx,
+ // bytes (i.e. dwSize) being 0 when dwFreeType is MEM_DECOMMIT means that we'll
+ // decommit the entire region allocated by VirtualAlloc() instead of decommitting
+ // nothing as we would expect. Hence, we should check if bytes is 0 and handle it
+ // appropriately before calling VirtualFree().
+ // See: https://bugs.webkit.org/show_bug.cgi?id=121972.
+ if (!bytes)
+ return;
bool result = VirtualFree(address, bytes, MEM_DECOMMIT);
if (!result)
CRASH();
@@ -72,6 +80,10 @@
void OSAllocator::releaseDecommitted(void* address, size_t bytes)
{
+ // See comment in OSAllocator::decommit(). Similarly, when bytes is 0, we
+ // don't want to release anything. So, don't call VirtualFree() below.
+ if (!bytes)
+ return;
// According to http://msdn.microsoft.com/en-us/library/aa366892(VS.85).aspx,
// dwSize must be 0 if dwFreeType is MEM_RELEASE.
bool result = VirtualFree(address, 0, MEM_RELEASE);
_______________________________________________
webkit-changes mailing list
webkit-changes@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-changes
