This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] new dg-effective-target(s) for >= 32 bits, plus testcase patches
- From: Janis Johnson <janis187 at us dot ibm dot com>
- To: DJ Delorie <dj at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Mon, 10 Apr 2006 16:19:37 -0700
- Subject: Re: [patch] new dg-effective-target(s) for >= 32 bits, plus testcase patches
- References: <200604050237.k352bL2I022741@greed.delorie.com>
On Tue, Apr 04, 2006 at 10:37:21PM -0400, DJ Delorie wrote:
>
> Ok, here's the patch. I added seven new predicates (for completeness,
> not all are used at the moment), which test int, void*, and size_t for
> >=32 bits, or <=16 bits, plus ilp16 for a full-16-bit target.
Sorry for the delay, I've been out of touch for a few days.
Leave out the procs that are never used, it's easy enough to recreate
them later if they are needed.
Add a comment before each of the new procs, instead of comments that
cover groups of procs.
> I also include a set of testcase patches to apply these predicates.
> Tested on m32c and xstormy16. Xstormy16 got 3 fail->pass due to
> cleaned up warning tests, and many fail->xfails.
>
> For some of these test cases, I suppose various constants could be
> changed to make them fix in a 16 bit target, but I don't know which
> cases depend on the actual value, so I chose to not do that.
>
> A few were really m32c-specific xfails; it's got 24 bit pointers, for
> example. Not much point generalizing that ;-)
See the comment below about xfail vs. skip.
> Some testcases, I used skip-if instead of xfail-if, when there were
> multiple failures due indirectly to one root cause (mostly multiple
> excess warning failures).
Do you mean for tests of functionality that ought to be supported for
the target? A test should be skipped if the functionality is not
supported for the target, or xfail'ed if it demonstrates a known bug.
>
> Ok?
OK with the changes requested above, and after checking that each
xfail represents a failure and not something that should be skipped.
Janis