[PATCH] Fix the broken aha.c test

Nick Lewycky nicholas at mxc.ca
Mon Sep 16 22:29:57 PDT 2013

Jim Stichnoth wrote:
> Long-overdue ping on this.
> The aha test as it stands is an invalid C program since it accesses an
> array outside its bounds, though that data is never used due to a
> subsequent bounds check.  If we don't want to fix the test (given its
> third-party origin), should we at least add a comment documenting the
> invalid access?
> Admittedly, there's almost no system that will crash on this code.
>   Previously, it was causing a failure in PNaCl-generated code, but a
> later PNaCl change (as a result of this investigation) fixed the failure.

Invalid code is always bad and should be fixed. It could also cause the 
test to start failing upon any innocent optimizer change too.

Thanks for the patch! Committed in r190848!


> Jim
> On Wed, Jul 3, 2013 at 10:55 AM, Jim Stichnoth <stichnot at google.com
> <mailto:stichnot at google.com>> wrote:
>     On Wed, Jul 3, 2013 at 10:39 AM, Tim Northover
>     <t.p.northover at gmail.com <mailto:t.p.northover at gmail.com>> wrote:
>         Hi Jim,
>          > The attached patch moves this clause to the beginning of the
>          > condition, giving the same result without the illegal reference.
>         I think you forgot the patch.
>     D'oh!  Thanks, attached now.
>     Jim
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

More information about the llvm-commits mailing list