Bug 118275 – [WTF]: Fix out-of-bounds subscript gcc 4.8 warning in dtoa/double-conversion.cc

• Status: NEW
• Component: New Bugs

[Sergio Correia (qrwteyrutiyoup)]

[WTF]: Fix subscript out-of-bound gcc 4.8 warning in dtoa/double-conversion.cc

[Sergio Correia (qrwteyrutiyoup)]

Created attachment 205858 [details]
Patch

[Sergio Correia (qrwteyrutiyoup)]

Warning:

/home/sergio/projects/webkitnix/Source/WTF/wtf/dtoa/double-conversion.cc: In static member function 'static double WTF::double_conversion::StringToDoubleConverter::StringToDouble(const char*, size_t, size_t*)':
/home/sergio/projects/webkitnix/Source/WTF/wtf/dtoa/double-conversion.cc:597:26: error: array subscript is above array bounds [-Werror=array-bounds]

[Darin Adler]

Comment on attachment 205858 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=205858&action=review

> Source/WTF/wtf/dtoa/double-conversion.cc:449
> -        char buffer[kBufferSize];  // NOLINT: size is known at compile time.
> +        char buffer[kBufferSize + 1];  // NOLINT: size is known at compile time.
> +        buffer[kBufferSize] = '\0';

I don’t understand this. The code below always null-character-terminates the buffer, and the buffer is always large enough. This seems like a useless change driven by an incorrect compiler warning.

Am I wrong? If so, tell me why.

[Sergio Correia (qrwteyrutiyoup)]

(In reply to comment #3)
> (From update of attachment 205858 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=205858&action=review
> 
> > Source/WTF/wtf/dtoa/double-conversion.cc:449
> > -        char buffer[kBufferSize];  // NOLINT: size is known at compile time.
> > +        char buffer[kBufferSize + 1];  // NOLINT: size is known at compile time.
> > +        buffer[kBufferSize] = '\0';
> 
> I don’t understand this. The code below always null-character-terminates the buffer, and the buffer is always large enough. This seems like a useless change driven by an incorrect compiler warning.
> 
> Am I wrong? If so, tell me why.

You are right, the problem seems to be the warning itself, in this case. Just checked out and there are a few reports of bogus warnings for the array bounds checking in gcc's bugzilla. And yeah, this boils down to a useless change to silence the compiler, which was needed because it's treating warnings as errors.

[Build Bot]

Comment on attachment 205858 [details]
Patch

Attachment 205858 [details] did not pass win-ews (win):
Output: http://webkit-queues.appspot.com/results/938890

New failing tests:
fast/forms/select/popup-closes-on-blur.html

[Build Bot]

Created attachment 205951 [details]
Archive of layout-test-results from APPLE-EWS-3 for win-future

The attached test failures were seen while running run-webkit-tests on the win-ews.
Bot: APPLE-EWS-3  Port: win-future  Platform: CYGWIN_NT-6.1-WOW64-1.7.20-0.266-5-3-i686-32bit