From: Thomas Schwinge <thomas@codesourcery.com>
To: Alexandre Oliva <oliva@adacore.com>
Cc: <ebotcazou@adacore.com>, <gcc-patches@gcc.gnu.org>,
<joseph@codesourcery.com>, Richard Biener <rguenther@suse.de>,
Jakub Jelinek <jakub@redhat.com>
Subject: Re: drop -aux{dir,base}, revamp -dump{dir,base}
Date: Tue, 9 Jun 2020 15:08:01 +0200 [thread overview]
Message-ID: <874krkqte6.fsf@euler.schwinge.homeip.net> (raw)
In-Reply-To: <or7dwzb2bn.fsf@livre.home>
Hi!
On 2020-05-26T04:08:44-0300, Alexandre Oliva <oliva@adacore.com> wrote:
> Thanks, here's the combined patch I'm checking in.
>
> revamp dump and aux output names
For libgomp offloading testing, I'm seeing a number of failures like:
UNSUPPORTED: libgomp.oacc-c/../libgomp.oacc-c-c++-common/nvptx-merged-loop.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O0
PASS: libgomp.oacc-c/../libgomp.oacc-c-c++-common/nvptx-merged-loop.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 (test for excess errors)
PASS: libgomp.oacc-c/../libgomp.oacc-c-c++-common/nvptx-merged-loop.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 execution test
[-PASS:-]{+UNRESOLVED:+} libgomp.oacc-c/../libgomp.oacc-c-c++-common/nvptx-merged-loop.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 scan-offload-rtl-dump mach "Merging loop .* into "
[...]
UNSUPPORTED: libgomp.oacc-c/../libgomp.oacc-c-c++-common/pr85381-2.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O0
PASS: libgomp.oacc-c/../libgomp.oacc-c-c++-common/pr85381-2.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 (test for excess errors)
PASS: libgomp.oacc-c/../libgomp.oacc-c-c++-common/pr85381-2.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 execution test
[-PASS:-]{+FAIL:+} libgomp.oacc-c/../libgomp.oacc-c-c++-common/pr85381-2.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 scan-assembler-times bar.sync 2
[...]
PASS: libgomp.oacc-c/../libgomp.oacc-c-c++-common/vector-length-128-1.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O0 (test for excess errors)
PASS: libgomp.oacc-c/../libgomp.oacc-c-c++-common/vector-length-128-1.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O0 execution test
PASS: libgomp.oacc-c/../libgomp.oacc-c-c++-common/vector-length-128-1.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O0 output pattern test
[-PASS:-]{+UNRESOLVED:+} libgomp.oacc-c/../libgomp.oacc-c-c++-common/vector-length-128-1.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O0 scan-offload-tree-dump oaccdevlow "__attribute__\\(\\(oacc function \\(1, 1, 128\\)"
[...]
UNSUPPORTED: libgomp.oacc-c/vec.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O0
PASS: libgomp.oacc-c/vec.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 (test for excess errors)
PASS: libgomp.oacc-c/vec.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 execution test
[-PASS:-]{+UNRESOLVED:+} libgomp.oacc-c/vec.c -DACC_DEVICE_TYPE_nvidia=1 -DACC_MEM_SHARED=0 -foffload=nvptx-none -O2 scan-offload-tree-dump slp1 "vector\\(2\\) long long unsigned int"
[...]
(A number of similar ones skipped.)
A word of warning: some of that testing indeed is fragile/ugly ;-) -- but
we have to restore it anyway. That means, we shall gladly accept
suggestions how to do it better.
Consider 'libgomp.oacc-c-c++-common/nvptx-merged-loop.c':
/* { dg-options "-foffload=-fdump-rtl-mach" } */
/* { dg-final { scan-offload-rtl-dump "Merging loop .* into " "mach" } } */
Previously, for '-foffload=nvptx-none -foffload=-fdump-rtl-mach
-save-temps -o ./nvptx-merged-loop.exe', GCC produced the expected
'nvptx-merged-loop.o.307r.mach'.
Now, I find the file at '/tmp/cc6nlTN9.mkoffload-cc4ebUn4.o.307r.mach'
(The '-save-temps' comes from
'libgomp/testsuite/lib/libgomp-dg.exp:libgomp-dg-test', by the way.)
(Yes, that's all a bit ugly, how the 'mkoffload's etc. handle their
files...)
Similar for 'libgomp.oacc-c-c++-common/vector-length-128-1.c':
/* { dg-additional-options "-foffload=-fdump-tree-oaccdevlow" } */
/* { dg-final { scan-offload-tree-dump "__attribute__\\(\\(oacc function \\(1, 1, 128\\)" "oaccdevlow" } } */
Similar for 'libgomp.oacc-c/vec.c':
/* { dg-additional-options "-std=c99 -ftree-slp-vectorize -foffload=-fdump-tree-slp1" } */
/* { dg-final { scan-offload-tree-dump "vector\\(2\\) long long unsigned int" "slp1" } } */
(Thus not analyed these any further.)
Consider 'libgomp.oacc-c-c++-common/pr85381-2.c':
/* { dg-additional-options "-save-temps" } */
/* { dg-final { scan-assembler-times "bar.sync" 2 } } */
This expects to scan the PTX offloading compilation assembler code (not
host code!), expecting that nvptx offloading code assembly is produced
after the host code, and thus overwrites the latter file. (Yes, that's
certainly ugly/fragile...)
Previously, for '-foffload=nvptx-none -save-temps -o ./pr85381-2.exe',
GCC left the '-foffload=nvptx-none' assembly in the expected
'pr85381-2.s'.
Now, I find the file at '/tmp/cc8JsG4H.mkoffload-pr85381-2.s'.
Are you able to easily create/suggest patches for these? (You're
probably not set up for offloading compilation...) Can you suggest
how/where to adjust: producer-side (GCC driver, 'mkoffload's?), or
consumer-side (testsuite: offload tree scanning machinery etc., or have
to put some '-dumpbase' or similar into all such test case files?)?
Grüße
Thomas
-----------------
Mentor Graphics (Deutschland) GmbH, Arnulfstraße 201, 80634 München / Germany
Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, Alexander Walter
next prev parent reply other threads:[~2020-06-09 13:08 UTC|newest]
Thread overview: 117+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-26 5:30 introduce -fcallgraph-info option Alexandre Oliva
2019-10-26 8:12 ` Richard Biener
2019-10-26 11:21 ` Alexandre Oliva
2019-10-27 9:08 ` Alexandre Oliva
2019-10-28 8:38 ` Richard Biener
2019-10-30 10:12 ` Alexandre Oliva
2019-10-30 11:28 ` Richard Biener
2019-11-02 19:31 ` Alexandre Oliva
2019-11-04 8:28 ` Richard Biener
2019-11-06 10:53 ` Alexandre Oliva
2019-11-20 4:35 ` Alexandre Oliva
2019-11-20 9:16 ` Richard Biener
2019-11-06 15:38 ` Alexandre Oliva
2019-11-07 7:49 ` Richard Biener
2019-11-07 10:50 ` Alexandre Oliva
2019-11-07 11:48 ` Richard Biener
2019-11-07 12:30 ` Alexandre Oliva
2019-11-07 14:02 ` Richard Biener
2019-11-07 22:39 ` Alexandre Oliva
2019-11-08 8:28 ` Richard Biener
2019-11-08 23:12 ` Eric Gallager
2019-11-15 1:34 ` Alexandre Oliva
2019-11-15 1:01 ` Alexandre Oliva
2019-11-15 7:30 ` Alexandre Oliva
2019-11-15 8:08 ` Richard Biener
2019-11-15 22:50 ` Alexandre Oliva
2019-12-03 22:44 ` Alexandre Oliva
2019-12-09 9:41 ` Richard Biener
2019-12-12 1:14 ` Alexandre Oliva
2019-12-25 10:05 ` Alexandre Oliva
2019-12-26 19:00 ` drop -aux{dir,base}, revamp -dump{dir,base} (was: Re: introduce -fcallgraph-info option) Alexandre Oliva
2020-01-09 13:33 ` Richard Biener
2020-01-09 19:09 ` drop -aux{dir,base}, revamp -dump{dir,base} Alexandre Oliva
2020-01-16 11:06 ` Alexandre Oliva
2020-01-16 11:15 ` Alexandre Oliva
2020-01-17 2:14 ` Joseph Myers
2020-01-20 10:38 ` Richard Biener
2020-01-22 1:25 ` Alexandre Oliva
2020-01-22 8:15 ` Richard Biener
2020-01-23 20:06 ` Alexandre Oliva
2020-01-24 1:50 ` Alexandre Oliva
2020-05-19 8:51 ` Alexandre Oliva
2020-05-19 8:59 ` Alexandre Oliva
2020-05-19 9:29 ` Richard Biener
2020-05-19 8:59 ` Alexandre Oliva
2020-05-19 9:30 ` Richard Biener
2020-05-19 9:00 ` Alexandre Oliva
2020-05-19 9:04 ` Richard Biener
2020-05-22 0:32 ` Alexandre Oliva
2020-05-22 6:05 ` Richard Biener
2020-05-26 7:08 ` Alexandre Oliva
2020-05-26 8:52 ` Richard Biener
2020-05-26 9:02 ` Martin Liška
2020-05-26 10:00 ` Alexandre Oliva
2020-05-26 12:14 ` Alexandre Oliva
2020-05-26 13:52 ` Alexandre Oliva
2020-05-26 13:56 ` Richard Biener
2020-05-27 22:05 ` Alexandre Oliva
2020-05-27 23:01 ` Jeff Law
2020-06-02 11:52 ` Alexandre Oliva
2020-06-02 12:02 ` Richard Biener
2020-05-27 1:04 ` Broken build (was: Re: drop -aux{dir,base}, revamp -dump{dir,base}) Hans-Peter Nilsson
2020-05-27 14:30 ` Broken build Alexandre Oliva
2020-05-27 15:04 ` Hans-Peter Nilsson
2020-05-28 0:53 ` Alexandre Oliva
2020-05-28 0:39 ` Anthony Green
2020-06-02 11:29 ` Alexandre Oliva
2020-06-02 14:07 ` Hans-Peter Nilsson
2020-05-27 9:45 ` drop -aux{dir,base}, revamp -dump{dir,base} Andreas Schwab
2020-05-27 10:28 ` Andreas Schwab
2020-05-27 14:41 ` Alexandre Oliva
2020-05-27 14:59 ` Andreas Schwab
2020-06-09 12:29 ` Thomas Schwinge
2020-06-09 12:42 ` BRIG FE testsuite: Fix all dump-scans (Was: Re: drop -aux{dir, base}, revamp -dump{dir, base}) Martin Jambor
2020-06-09 18:31 ` BRIG FE testsuite: Fix all dump-scans (Was: Re: drop -aux{dir,base}, revamp -dump{dir,base}) Mike Stump
2020-06-11 14:28 ` BRIG FE testsuite: Fix all dump-scans (Was: Re: drop -aux{dir, base}, revamp -dump{dir, base}) Martin Jambor
2020-06-12 20:52 ` BRIG FE testsuite: Fix all dump-scans (Was: Re: drop -aux{dir,base}, revamp -dump{dir,base}) Mike Stump
2020-06-10 21:50 ` BRIG FE testsuite: Fix all dump-scans Alexandre Oliva
2020-06-23 9:44 ` drop -aux{dir,base}, revamp -dump{dir,base} Alexandre Oliva
2020-06-23 11:30 ` Martin Jambor
2020-06-09 13:08 ` Thomas Schwinge [this message]
2020-06-10 22:24 ` Alexandre Oliva
2020-06-17 10:50 ` Tobias Burnus
2020-06-18 2:58 ` Alexandre Oliva
2020-06-18 6:10 ` Alexandre Oliva
2020-06-18 9:41 ` Tobias Burnus
2020-06-18 10:39 ` Alexandre Oliva
2020-06-18 12:06 ` Tobias Burnus
2020-06-30 16:13 ` Thomas Schwinge
2020-07-14 5:46 ` Alexandre Oliva
2020-07-24 6:08 ` Thomas Schwinge
2020-07-24 17:54 ` Alexandre Oliva
2020-07-14 5:49 ` Alexandre Oliva
2020-06-19 9:53 ` Alexandre Oliva
2020-06-19 16:09 ` Tobias Burnus
2020-06-22 6:08 ` Alexandre Oliva
2020-06-22 7:07 ` Tobias Burnus
2020-06-22 14:32 ` Alexandre Oliva
2020-06-23 8:17 ` Richard Biener
2020-06-30 16:35 ` Thomas Schwinge
2020-07-24 12:06 ` Thomas Schwinge
2020-06-30 18:52 ` Thomas Schwinge
2020-07-14 4:48 ` Alexandre Oliva
2020-07-24 6:01 ` Thomas Schwinge
2020-07-24 18:00 ` Alexandre Oliva
2023-11-24 12:46 ` testsuite: Add 'only_for_offload_target' wrapper for 'scan-offload-tree-dump' etc. (was: drop -aux{dir,base}, revamp -dump{dir,base}) Thomas Schwinge
2020-06-23 9:50 ` drop -aux{dir,base}, revamp -dump{dir,base} Alexandre Oliva
2020-06-30 16:07 ` Thomas Schwinge
2020-07-23 21:12 ` [PR95720] protect gluefile and wrap_flags with -Wl too (was: Re: drop -aux{dir,base}, revamp -dump{dir,base}) Alexandre Oliva
2020-07-24 6:48 ` Richard Biener
2020-05-26 8:09 ` [wwwdocs] Re: drop -aux{dir,base}, revamp -dump{dir,base} Alexandre Oliva
2019-10-28 23:56 ` introduce -fcallgraph-info option Joseph Myers
2019-10-30 9:22 ` Alexandre Oliva
2019-10-30 9:43 ` Alexandre Oliva
2019-10-30 18:17 ` Joseph Myers
2019-11-06 21:27 ` Thomas Schwinge
2019-11-07 11:23 ` Alexandre Oliva
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=874krkqte6.fsf@euler.schwinge.homeip.net \
--to=thomas@codesourcery.com \
--cc=ebotcazou@adacore.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jakub@redhat.com \
--cc=joseph@codesourcery.com \
--cc=oliva@adacore.com \
--cc=rguenther@suse.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).