This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: predicate aware uninitialized variable analysis
- From: Manuel López-Ibáñez <lopezibanez at gmail dot com>
- To: Xinliang David Li <davidxl at google dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>, Richard Guenther <rguenther at suse dot de>
- Date: Wed, 14 Apr 2010 13:34:11 +0200
- Subject: Re: predicate aware uninitialized variable analysis
- References: <522e93240907221419y2d8dbdbdta7343148338688cc@mail.gmail.com> <q2r522e93241004131223scbd7cc8fx283be7d728f7ef58@mail.gmail.com>
On 13 April 2010 21:23, Xinliang David Li <davidxl@google.com> wrote:
> The patch did not make into 4.5. Try it again for 4.6.
>
> There are new changes since the last attempt: The candidate list of
> PHIs (which may produce undefined values) are not precomputed. In the
> new patch, the predicate check will also be done on uses by phi
> arguments. Only when the uses ?in the phi arguments are not protected
> properly by a predicate, the phi statement will then be added to the
> worklist (and uses of its definitions will be checked later).
>
> Bootstrap and tested on x86_64/linux, and lots of apps.
>
> Ok for 4.6?
Since you have to generate a new patch anyway from Richard's comments.
I would propose to move all tests that are not C or C++ specific to
testsuite/c-c++-common/
Note that if the difference is the command line parameters (you need
-std=* or some -Wno-* to avoid warnings), you can use
/* { dg-options "options for c" { target c } } */
/* { dg-options "options for c++" { target c++ } } */
The same works with other dg-* commands.
Cheers,
Manuel.