[Bug target/86753] New: gcc.target/aarch64/sve/vcond_[45.c

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

[Bug target/86753] New: gcc.target/aarch64/sve/vcond_[45.c

kargl at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86753

            Bug ID: 86753
           Summary: gcc.target/aarch64/sve/vcond_[45.c
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rsandifo at gcc dot gnu.org
  Target Milestone: ---
Reply | Threaded
Open this post in threaded view
|

[Bug target/86753] gcc.target/aarch64/sve/vcond_[45].c fail after recent combine patch

kargl at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86753

rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|                            |aarch64-linux-gnu
            Summary|gcc.target/aarch64/sve/vcon |gcc.target/aarch64/sve/vcon
                   |d_[45.c                     |d_[45].c fail after recent
                   |                            |combine patch

--- Comment #1 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> ---
gcc.target/aarch64/sve/vcond_4.c and
gcc.target/aarch64/sve/vcond_5.c fail after r263067.
The change is an improvement in the sense that it replaces
two consecutive instructions (a comparison and an AND)
with two independent instructions (two comparisons),
so it's not a problem with the patch.

What we really want here is a single predicated comparison
that's used for both the load and the select.  However,
that would need support for conditional comparisons
(perhaps along the lines of the IFN_COND_* arithmetic
functions) and would also need us to ditch the nested
comparison in VEC_COND_EXPRs.
Reply | Threaded
Open this post in threaded view
|

[Bug target/86753] [9 Regression] gcc.target/aarch64/sve/vcond_[45].c fail after recent combine patch

kargl at gcc dot gnu.org
In reply to this post by kargl at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86753

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|unknown                     |9.0
   Target Milestone|---                         |9.0
            Summary|gcc.target/aarch64/sve/vcon |[9 Regression]
                   |d_[45].c fail after recent  |gcc.target/aarch64/sve/vcon
                   |combine patch               |d_[45].c fail after recent
                   |                            |combine patch
Reply | Threaded
Open this post in threaded view
|

[Bug target/86753] [9 Regression] gcc.target/aarch64/sve/vcond_[45].c fail after recent combine patch

kargl at gcc dot gnu.org
In reply to this post by kargl at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86753

--- Comment #2 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> ---
Author: rsandifo
Date: Wed Aug  1 16:00:05 2018
New Revision: 263227

URL: https://gcc.gnu.org/viewcvs?rev=263227&root=gcc&view=rev
Log:
[AArch64] XFAIL sve/vcond_[45].c tests

See PR 86753 for details.

2018-08-01  Richard Sandiford  <[hidden email]>

gcc/testsuite/
        PR target/86753
        * gcc.target/aarch64/sve/vcond_4.c: XFAIL positive tests.
        * gcc.target/aarch64/sve/vcond_5.c: Likewise.

Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.target/aarch64/sve/vcond_4.c
    trunk/gcc/testsuite/gcc.target/aarch64/sve/vcond_5.c
Reply | Threaded
Open this post in threaded view
|

[Bug target/86753] [9 Regression] gcc.target/aarch64/sve/vcond_[45].c fail after recent combine patch

kargl at gcc dot gnu.org
In reply to this post by kargl at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86753

huangzao <zao.huang at intel dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |zao.huang at intel dot com

--- Comment #3 from huangzao <zao.huang at intel dot com> ---
r263067 also causes failures of gcc.target/i386/avx-cvt-2.c on haswell.
Reply | Threaded
Open this post in threaded view
|

[Bug target/86753] [9 Regression] gcc.target/aarch64/sve/vcond_[45].c fail after recent combine patch

kargl at gcc dot gnu.org
In reply to this post by kargl at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86753

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |law at redhat dot com
         Resolution|---                         |FIXED

--- Comment #4 from Jeffrey A. Law <law at redhat dot com> ---
Per Richard's c#2, the ultimate result is actually an improvement and Richard
has twiddled the tests.

WRT c#3, reports of problems on Haswell, if this is still an issue, please
report it as a separate bug.
Reply | Threaded
Open this post in threaded view
|

[Bug target/86753] [9 Regression] gcc.target/aarch64/sve/vcond_[45].c fail after recent combine patch

kargl at gcc dot gnu.org
In reply to this post by kargl at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86753

rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |NEW
   Last reconfirmed|                            |2018-12-07
         Resolution|FIXED                       |---
     Ever confirmed|0                           |1

--- Comment #5 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> ---
The PR is for tracking why the tests needed to be XFAILed.  What the tests are
looking for is still the preferred outcome, so I think we should keep the PR
open until the XFAILs can be removed.

The point of #c1 was that the combine patch that made the tests regress wasn't
really at fault.  We need more gimple stuff to do this properly.
Reply | Threaded
Open this post in threaded view
|

[Bug target/86753] [9 Regression] gcc.target/aarch64/sve/vcond_[45].c fail after recent combine patch

kargl at gcc dot gnu.org
In reply to this post by kargl at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86753

--- Comment #6 from Jeffrey A. Law <law at redhat dot com> ---
Sorry, my misunderstanding.  I thought you had indicated the resulting code was
better.