[Webkit-unassigned] [Bug 226193] Fix more GCC warnings

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=226193

Michael Catanzaro <mcatanzaro at gnome.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mcatanzaro at gnome.org

--- Comment #1 from Michael Catanzaro <mcatanzaro at gnome.org> ---
One of these is really frustrating me. I wonder if anybody else sees a problem with this code, or whether GCC's warning is just bogus here:

[643/1468] Building CXX object Source/WebCore/CMakeFiles/W...vedSources/unified-sources/UnifiedSource-4babe430-49.cpp.o
In file included from WTF/Headers/wtf/text/StringImpl.h:32,
                 from WTF/Headers/wtf/text/WTFString.h:31,
                 from ../../Source/WebCore/dom/Exception.h:30,
                 from ../../Source/WebCore/dom/ExceptionOr.h:29,
                 from ../../Source/WebCore/dom/Event.h:29,
                 from ../../Source/WebCore/Modules/websockets/CloseEvent.h:33,
                 from ../../Source/WebCore/Modules/websockets/CloseEvent.cpp:27,
                 from WebCore/DerivedSources/unified-sources/UnifiedSource-4babe430-49.cpp:3:
In member function ‘T* WTF::PackedAlignedPtr<T, <anonymous> >::get() const [with T = JSC::SharedArrayBufferContents; long unsigned int passedAlignment = 1]’,
    inlined from ‘WTF::PackedAlignedPtr<T, <anonymous> >::operator bool() const [with T = JSC::SharedArrayBufferContents; long unsigned int passedAlignment = 1]’ at WTF/Headers/wtf/Packed.h:194:48,
    inlined from ‘WTF::RefPtr<T, <template-parameter-1-2>, <template-parameter-1-3> >::operator WTF::RefPtr<T, <template-parameter-1-2>, <template-parameter-1-3> >::UnspecifiedBoolType() const [with T = JSC::SharedArrayBufferContents; _PtrTraits = WTF::PackedPtrTraits<JSC::SharedArrayBufferContents>; _RefDerefTraits = WTF::DefaultRefDerefTraits<JSC::SharedArrayBufferContents>]’ at WTF/Headers/wtf/RefPtr.h:91:57,
    inlined from ‘bool JSC::ArrayBufferContents::isShared() const’ at JavaScriptCore/PrivateHeaders/JavaScriptCore/ArrayBuffer.h:84:36,
    inlined from ‘bool JSC::ArrayBuffer::isShared() const’ at JavaScriptCore/PrivateHeaders/JavaScriptCore/ArrayBuffer.h:202:31,
    inlined from ‘WTF::RefPtr<JSC::ArrayBuffer> JSC::ArrayBufferView::unsharedBuffer() const’ at JavaScriptCore/PrivateHeaders/JavaScriptCore/ArrayBufferView.h:60:9,
    inlined from ‘WebCore::ExceptionOr<void> WebCore::WebSocket::send(JSC::ArrayBufferView&)’ at ../../Source/WebCore/Modules/websockets/WebSocket.cpp:390:52:
WTF/Headers/wtf/Packed.h:143:15: warning: ‘void* memcpy(void*, const void*, size_t)’ offset [0, 5] is out of the bounds [0, 0] [-Warray-bounds]
  143 |         memcpy(static_cast<void*>(&value), static_cast<void*>(const_cast<uint8_t*>(m_storage.data())), storageSize);
      |         ~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


GCC's -Warray-bounds warning is frankly pretty crappy. It's usually a false-positive, so I normally wouldn't worry too much about suppressing it. But in this case, suppressing it introduces a -Wstringop-overread warning:

[641/1468] Building CXX object Source/WebCore/CMakeFiles/W...vedSources/unified-sources/UnifiedSource-4babe430-49.cpp.o
In file included from WTF/Headers/wtf/text/StringImpl.h:32,
                 from WTF/Headers/wtf/text/WTFString.h:31,
                 from ../../Source/WebCore/dom/Exception.h:30,
                 from ../../Source/WebCore/dom/ExceptionOr.h:29,
                 from ../../Source/WebCore/dom/Event.h:29,
                 from ../../Source/WebCore/Modules/websockets/CloseEvent.h:33,
                 from ../../Source/WebCore/Modules/websockets/CloseEvent.cpp:27,
                 from WebCore/DerivedSources/unified-sources/UnifiedSource-4babe430-49.cpp:3:
In member function ‘T* WTF::PackedAlignedPtr<T, <anonymous> >::get() const [with T = JSC::SharedArrayBufferContents; long unsigned int passedAlignment = 1]’,
    inlined from ‘WTF::PackedAlignedPtr<T, <anonymous> >::operator bool() const [with T = JSC::SharedArrayBufferContents; long unsigned int passedAlignment = 1]’ at WTF/Headers/wtf/Packed.h:194:48,
    inlined from ‘WTF::RefPtr<T, <template-parameter-1-2>, <template-parameter-1-3> >::operator WTF::RefPtr<T, <template-parameter-1-2>, <template-parameter-1-3> >::UnspecifiedBoolType() const [with T = JSC::SharedArrayBufferContents; _PtrTraits = WTF::PackedPtrTraits<JSC::SharedArrayBufferContents>; _RefDerefTraits = WTF::DefaultRefDerefTraits<JSC::SharedArrayBufferContents>]’ at WTF/Headers/wtf/RefPtr.h:91:57,
    inlined from ‘bool JSC::ArrayBufferContents::isShared() const’ at JavaScriptCore/PrivateHeaders/JavaScriptCore/ArrayBuffer.h:84:36,
    inlined from ‘bool JSC::ArrayBuffer::isShared() const’ at JavaScriptCore/PrivateHeaders/JavaScriptCore/ArrayBuffer.h:202:31,
    inlined from ‘WTF::RefPtr<JSC::ArrayBuffer> JSC::ArrayBufferView::unsharedBuffer() const’ at JavaScriptCore/PrivateHeaders/JavaScriptCore/ArrayBufferView.h:60:9,
    inlined from ‘WebCore::ExceptionOr<void> WebCore::WebSocket::send(JSC::ArrayBufferView&)’ at ../../Source/WebCore/Modules/websockets/WebSocket.cpp:390:52:
WTF/Headers/wtf/Packed.h:143:15: warning: ‘void* memcpy(void*, const void*, size_t)’ reading 6 bytes from a region of size 0 [-Wstringop-overread]
  143 |         memcpy(static_cast<void*>(&value), static_cast<void*>(const_cast<uint8_t*>(m_storage.data())), storageSize);
      |         ~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


And that has got me worried, because -Wstringop-overread is a good warning and I do not remember seeing a false-positive -Wstringop-overread before. But I also do not see any problems with this code. The code is not going crazy here: it seems to be careful to check the size of its buffers. Maybe I am missing something?

Anyway, I assume they are both false-positives, because I don't see anything wrong, but I'm calling them out because they leave me a little nervous.

(There's also a new -Wnonnull in JITCall.cpp complaining that the "this" pointer can be nullptr in JIT::compileOpCall. As usual, GCC shows no evidence for why we should believe this extraordinary claim. If it's possible, it is far from clear bbhow.)

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20210524/3e6ea5c5/attachment-0001.htm>