https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #47 from Paul Thomas ---
Author: pault
Date: Fri Dec 21 20:58:59 2018
New Revision: 267342
URL: https://gcc.gnu.org/viewcvs?rev=267342&root=gcc&view=rev
Log:
2018-12-21 Paul Thomas
PR fortran/87359
* trans-array.c
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
Jürgen Reuter changed:
What|Removed |Added
Status|REOPENED|RESOLVED
Resolution|---
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #45 from Paul Thomas ---
Author: pault
Date: Sun Sep 30 13:52:55 2018
New Revision: 264725
URL: https://gcc.gnu.org/viewcvs?rev=264725&root=gcc&view=rev
Log:
2018-09-30 Paul Thomas
PR fortran/87359
* trans-array.c
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #44 from paul.richard.thomas at gmail dot com ---
Hi Jeurgen,
Thanks for the confirmation. I will take care of a composite fix over
the weeknd. (I get home tomorrow lunchtime.).
Cheers
Paul
On Fri, 28 Sep 2018 at 11:13, juergen.re
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #43 from Jürgen Reuter ---
I just checked that Paul's fix actually solves all problems that our code had
with the current trunk of gcc/gfortran as of r264501.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #42 from Paul Thomas ---
Created attachment 44746
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44746&action=edit
Patch for the PR
Many thanks Juergen for the reduced test. This is now DEJA-GNUified (see
below).
The patch reg
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #41 from Paul Thomas ---
(In reply to Jürgen Reuter from comment #40)
> (In reply to Dominique d'Humieres from comment #39)
> > The culprit for the second problem is r264358, exposed only once the
> > original problem is fixed, i.e. w
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #40 from Jürgen Reuter ---
(In reply to Dominique d'Humieres from comment #39)
> The culprit for the second problem is r264358, exposed only once the
> original problem is fixed, i.e. with the patch of r264485 (tested on both
> reprod
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #39 from Dominique d'Humieres ---
The culprit for the second problem is r264358, exposed only once the original
problem is fixed, i.e. with the patch of r264485 (tested on both reproducers).
It would have been better to open a new PR
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #38 from Jürgen Reuter ---
Created attachment 44745
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44745&action=edit
Small reproducer for the second problem
And an 89 line small reproducer for the second problem.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #37 from Jürgen Reuter ---
Paul, here is a simple reproducer of 89 lines, this should now make it
relatively
easy to debug, I am using gcc trunk revision r264501.
Here is the code (I will also attach it), it contains also a workaround
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #36 from Dominique d'Humieres ---
> The test succeeds with r264348+patch of r264485, but not with r264349+patch.
I make a mistake in my bissection: r264349+patch is OK as well as
r264357+patch, but not r264358+patch.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #35 from Jürgen Reuter ---
(In reply to Dominique d'Humieres from comment #34)
> > Created attachment 44739 [details]
> > Reproducer for the second problem.
>
> The test succeeds with r264348+patch of r264485, but not with r264349+pa
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #34 from Dominique d'Humieres ---
> Created attachment 44739 [details]
> Reproducer for the second problem.
The test succeeds with r264348+patch of r264485, but not with r264349+patch.
IMO it would be better to open a new PR for it
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #33 from Jürgen Reuter ---
Created attachment 44739
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44739&action=edit
Reproducer for the ICE.
Paul, here is a first (still massive) reproducer of the second problem. Though
it is s
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #32 from Jürgen Reuter ---
Paul, I found a workaround: in lines 530-533 in the file
src/transforms/evt_nlo.f90 there is an assignment of an allocatable array of
different DT components which apparently doesn't work any more. Changing
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #31 from Jürgen Reuter ---
That is really strange, I never used fink, I use macports. But not for gcc, I
always compile them by myself on MACOSX. Clearly the C/Fortran I/O doesn't work
properly, the string parsing doesn't work properl
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #30 from Dominique d'Humieres ---
> This is really strange, is that after Paul's fix?
It is with gfortran 8.2 (from fink).
> This seems to be a hiccup with the I/O. Did you use gcc or clang
> as the underlying C compiler?
I use gcc
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #29 from Jürgen Reuter ---
(In reply to Dominique d'Humieres from comment #28)
> I am trying to install a clean whizard-2.6.4 on x86_64-apple-darwin17, Xcode
> 9.4.1, SIP disabled, gfortran 8.2, ocaml 4.03.0. I got the following error
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #28 from Dominique d'Humieres ---
I am trying to install a clean whizard-2.6.4 on x86_64-apple-darwin17, Xcode
9.4.1, SIP disabled, gfortran 8.2, ocaml 4.03.0. I got the following errors
with make check:
FAIL: mci_vamp.run
FAIL: inte
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #27 from Jürgen Reuter ---
Interesting: when I run with checking flags, I get the following error:
At line 532 of file evt_nlo.f90
Fortran runtime error: Array bound mismatch for dimension 1 of array
'event_deps' (0/2)
However, I have
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #26 from paul.richard.thomas at gmail dot com ---
Jeurgen,
We are extremely pleased that you do follow developments on trunk. It
really helps to catch regressions early, while the changes are fresh
in mind :-)
Sometime, I would appr
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #25 from Jürgen Reuter ---
This is the part from the test-suite.log for the 4 failures, they are all in
one particular feature of our code, so I am pretty sure that this is only one
remaining open issue:
| Starting simulation for proc
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #24 from Jürgen Reuter ---
Paul, enjoy your time in Wales. Maybe this other issue wasn't caused by r263916
but by something else (though it must have been also in the past 2-3 weeks).
What our functional tests do: they call a code gen
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #23 from Paul Thomas ---
(In reply to Jürgen Reuter from comment #21)
> In our functional test suite, the tests nlo_4, nlo_5, fks_res_1 and another
> test are still failing, they lead to segmentation faults. This will be
> really diff
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #22 from Paul Thomas ---
(In reply to Jürgen Reuter from comment #20)
> Paul, thanks for the fix, our code test suite is still running, most of the
> problems are solved, the unit test suite is completely good now, but there
> are cer
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #21 from Jürgen Reuter ---
In our functional test suite, the tests nlo_4, nlo_5, fks_res_1 and another
test are still failing, they lead to segmentation faults. This will be really
difficult to isolate, but maybe this is a different r
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
Jürgen Reuter changed:
What|Removed |Added
Status|RESOLVED|REOPENED
Resolution|FIXED
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
Paul Thomas changed:
What|Removed |Added
Status|NEW |RESOLVED
Resolution|---
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #18 from Paul Thomas ---
Hi Juergen,
Thanks for doing the reduction of the problem and thanks to Dominique for
testing the patch.
Fixed.
Paul
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #17 from Paul Thomas ---
Author: pault
Date: Fri Sep 21 17:26:23 2018
New Revision: 264485
URL: https://gcc.gnu.org/viewcvs?rev=264485&root=gcc&view=rev
Log:
2018-09-21 Paul Thomas
PR fortran/87359
* trans-stmt.c
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #16 from paul.richard.thomas at gmail dot com ---
Hi Dominique,
Many thanks for coming back so promptly. I will package it up for a
commit this evening.
Best regards
Paul
On 21 September 2018 at 17:12, dominiq at lps dot ens.fr
w
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #15 from Dominique d'Humieres ---
> Could you please test the attached patch?
The patch fixes both the reduced and the original tests.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
Paul Thomas changed:
What|Removed |Added
Assignee|unassigned at gcc dot gnu.org |pault at gcc dot gnu.org
--- Commen
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #13 from Jürgen Reuter ---
Created attachment 44732
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44732&action=edit
Promised shorted reproducer, 93 lines
This is the promised shortened reproducer, 93 lines long. This should ma
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #12 from Dominique d'Humieres ---
The problem is with the file process_mci.f90: if I compile all the other files
with r264428 and process_mci.f90 with r263915, the test succeeds.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #11 from Dominique d'Humieres ---
This is indeed caused by r263916.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #10 from Paul Thomas ---
(In reply to Thomas Koenig from comment #7)
> Well, I can confirm this. Output from valgrind shows as the first error:
>
> Running test: event_transforms_1| Process library 'event_transforms_1_lib':
> initia
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
Dominique d'Humieres changed:
What|Removed |Added
Keywords||wrong-code
Status|UNC
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #8 from Thomas Koenig ---
Created attachment 44728
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44728&action=edit
Makefile for the shortened reproducer
Hi,
if you want to compile the reproducer, you can use this Makefile,
it
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #7 from Thomas Koenig ---
Well, I can confirm this. Output from valgrind shows as the first error:
Running test: event_transforms_1| Process library 'event_transforms_1_lib':
initialized
==23957== Invalid read of size 8
==23957==
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #6 from Paul Thomas ---
(In reply to Jürgen Reuter from comment #5)
> (In reply to Thomas Koenig from comment #4)
> > After generating a lot of module files with -fsyntax-only
> > so make succeeds, and then running "make check" on the
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #5 from Jürgen Reuter ---
(In reply to Thomas Koenig from comment #4)
> After generating a lot of module files with -fsyntax-only
> so make succeeds, and then running "make check" on the
> reproducer, I get
**
> *
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
--- Comment #4 from Thomas Koenig ---
After generating a lot of module files with -fsyntax-only
so make succeeds, and then running "make check" on the
reproducer, I get
|===
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87359
Richard Biener changed:
What|Removed |Added
Priority|P3 |P4
Summary|[9.0 regression]
45 matches
Mail list logo