[cfe-commits] [PATCH] Fix for PR7218

Jordy Rose jediknil at belkadan.com
Mon May 31 13:19:20 PDT 2010


On Sun, 30 May 2010 18:26:05 -0700, Ted Kremenek <kremenek at apple.com>
wrote:
> This code was added to support addressing within a larger (non-array,
> non-struct) object, e.g.:
> 
>   int x = 10;
>   char *y = ((char*) &x) + 1;
>   return *y;
> 
> The code was too aggressive in the cases it handled.  I've now checked
in
> a refinement which causes the test case for PR 7218 to now pass:
> 
>  http://llvm.org/viewvc/llvm-project?view=rev&revision=105195
> 
> Zhongxing: Please review this patch and let me know what you think.
> 
> Jordy: Please review it as well.  Once we're satisfied, let's move on to
> talking about your changes to MallocChecker.

Mm. Yeah, it's not /so/ tidy, but it does limit it to the one example
that's there, and it's unlikely to disturb anything else. And of course I
have no better idea.



More information about the cfe-commits mailing list