[Bug 15388] New: gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

Summary: gdi32: dib.c fails to build with gcc 4.2, -Werror -O3
           Product: Wine
           Version: 1.1.5
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: source
          Severity: enhancement
          Priority: P2
         Component: gdi32
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

ccache /usr/bin/gcc-4.2 -O3 -c -I. -I. -I../../include -I../../include
-I/usr/include/freetype2  -D__WINESRC__ -D_GDI32_ -D_REENTRANT -fPIC -Wall
-pipe -fno-strength-reduce -fno-strict-aliasing -Wdeclaration-after-statement
-Wwrite-strings -Wpointer-arith  -Werror  -o dib.o dib.c
cc1: warnings being treated as errors
dib.c: In function ‘CreateDIBSection16’:
dib.c:1162: warning: ‘size’ may be used uninitialized in this function
dib.c:1162: warning: ‘compr’ may be used uninitialized in this function
dib.c:1161: warning: ‘bpp’ may be used uninitialized in this function
dib.c:1160: warning: ‘height’ may be used uninitialized in this function
dib.c:1160: warning: ‘width’ may be used uninitialized in this function
make: *** [dib.o] Error 1

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

--- Comment #1 from Austin English   2008-09-24 14:13:47 ---
Created an attachment (id=16257)
 --> (http://bugs.winehq.org/attachment.cgi?id=16257)
patch to initialize the variables

Haven't submitted yet, comments appreciated.

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

Nikolay Sivov  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #2 from Nikolay Sivov   2008-09-25 04:02:45 ---
I think it's better to check return value for DIB_GetBitmapInfo and exit on -1.
Since DIB_GetBitmapInfo is internal function which always fills this variables
on success. So with properly initialized header there shouldn't be any problem.

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

Andrey Turkin  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #3 from Andrey Turkin   2008-09-25 04:14:37 ---
The compiler wisely warn us that DIB_GetBitmapInfo would not always fill the
variables. I'd follow way it is done in CreateDIBitmap (get info early, before
any other operations, and bail out if DIB_GetBitmapInfo fails)

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

--- Comment #4 from Nikolay Sivov   2008-09-25 04:27:39 ---
(In reply to comment #3)
> I'd follow way it is done in CreateDIBitmap (get info early, before
> any other operations, and bail out if DIB_GetBitmapInfo fails)

Exactly, that's what I meant.

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #16257|0                           |1
        is obsolete|                            |

--- Comment #5 from Austin English   2008-09-29 21:16:59 ---
Created an attachment (id=16377)
 --> (http://bugs.winehq.org/attachment.cgi?id=16377)
new patch

This what you had in mind?

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

--- Comment #6 from Nikolay Sivov   2008-09-29 23:11:38 ---
(In reply to comment #5)
> Created an attachment (id=16377)

--> (http://bugs.winehq.org/attachment.cgi?id=16377) [details]
> new patch
>
> This what you had in mind?

Yes, but why are you duplicating a call? You could just use a result of
existing one:

DIB_GetBitmapInfo(bi, &width, &height, &planes, &bpp, &compr, &size);

+ if (DIB_GetBitmapInfo( bi, &width, &height, &planes, &bpp, &compr, &size ) ==
-1) return 0;

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

--- Comment #7 from Andrey Turkin   2008-09-30 01:17:02 ---
Uhm... After re-reading sources this error actually looks like false positive
as we already know DIB_GetBitmapInfo will work if we get to its call (because
of succeed CreateDIBSection). Probably too complicated for compiler's little
brain :)
However, in any other case this patch would be wrong because of introduced
resource leaks (DIB section was already created). I can see two ways - either
silence compiler like your original patch did, or pretend this is not false
positive and move DIB_GetBitmapInfo call up to very function beginning.

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

--- Comment #8 from Austin English   2008-12-08 03:47:41 ---
Still present in 1.1.10.

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

André H.  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |[e-mail]
         Resolution|                            |FIXED

--- Comment #9 from André H.   2010-01-20 06:56:44 ---
cannot confirm this, seems fixed
Wine-1.1.36

[Bug 15388] gdi32: dib.c fails to build with gcc 4.2, -Werror -O3

http://bugs.winehq.org/show_bug.cgi?id=15388

Alexandre Julliard  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #10 from Alexandre Julliard   2010-01-22 11:01:51 ---
Closing bugs fixed in 1.1.37.