-
Notifications
You must be signed in to change notification settings - Fork 51
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Empty constructor compilation error with PGI 14.4 #15
Comments
Even when commented default constructor test I get this error: $ make tests pgf95 datetime_tests.f90 -g -O0 -o tests datetime.o datetime_tests.f90: pgf95-Fatal-/sb/software/CentOS-6/compilers/pgi144/linux86-64/14.4/bin/pgf901 TERMINATED by signal 11 Arguments to /sb/software/CentOS-6/compilers/pgi144/linux86-64/14.4/bin/pgf901 /sb/software/CentOS-6/compilers/pgi144/linux86-64/14.4/bin/pgf901 datetime_tests.f90 -debug -x 120 0x200 -opt 0 -terse 1 -inform warn -nohpf -nostatic -x 19 0x400000 -quad -x 59 4 -x 15 2 -x 49 0x400004 -x 51 0x20 -x 57 0x4c -x 58 0x10000 -x 124 0x1000 -tp sandybridge -x 57 0xfb0000 -x 58 0x78031040 -x 48 4608 -x 49 0x100 -x 120 0x200 -stdinc /sb/software/CentOS-6/compilers/pgi144/linux86-64/14.4/include-gcc44:/sb/software/CentOS-6/compilers/pgi144/linux86-64/14.4/include:/usr/local/include:/usr/lib/gcc/x86_64-redhat-linux/4.4.4/include:/usr/include -def unix -def __unix -def __unix__ -def linux -def __linux -def __linux__ -def __NO_MATH_INLINES -def __x86_64 -def __x86_64__ -def __LONG_MAX__=9223372036854775807L -def '__SIZE_TYPE__=unsigned long int' -def '__PTRDIFF_TYPE__=long int' -def __THROW= -def __extension__= -def __amd_64__amd64__ -def __k8 -def __k8__ -def __SSE__ -def __MMX__ -def __SSE2__ -def __SSE3__ -def __SSSE3__ -freeform -vect 48 -y 54 1 -x 70 0x40000000 -x 49 0x1000 -modexport /tmp/huziy/25235/pgf95ovfhQk0jW0Gh.cmod -modindex /tmp/huziy/25235/pgf95_vfh6FvxM08e.cmdx -output /tmp/huziy/25235/pgf95EvfhApIoyfAI.ilm make: *** [tests] Error 127 |
Hi guziy, Yes, this was fixed by Portland Group in version 14.7 a few months ago as per my request (Tim Hilton of UC Merced pointed this problem out to me). I haven't had chance to try this version though. With 13.10-0 on my system, datetime.f90 compiles but the tests don't, with the same error that you get:
This is a signal for segmentation fault, and is likely to be compiler problem. I will leave this issue open for future reference if somebody comes up with a newer PGF version. |
Note that at that time (a few months ago), I modified all empty constructors (
|
@guziy Do you know if this problem still persists with the latest version of datetime_fortran? |
Hi Milan: I am trying to compile the module from github to answer your question, but autoconf is complaining, automake as well, do you have a description of how Cheers 2016-01-29 17:26 GMT-05:00 Izaak Beekman [email protected]:
Sasha |
You should be able to do the following after cloning a fresh copy, or pulling in the latest changes:
I'm not 100% certain that this will ensure that the correct compiler is used. Also, it looks like #37 needs to be merged before using |
Thanks @zbeekman: Here is what I get after
Cheers |
When using a newer automake (
pkg-config version is
Cheers |
@guziy I'm guessing this is due to outdated pkg-config. I think you need >= 0.27. Another tact you could take is downloading the latest release from the releases page which is packaged with a @blippy if this is true, is there a way to say that a certain version of pkg-config is required? ref: sardemff7/j4status#2 |
You've forgotten the "datetime-fortran" part in your link to the latest release...
Cheers |
@guziy , @zbeekman I have the same problem on my cubietruck, which is running Ubuntu LTS. A new LTS is due in less than 3 months time. Do you intend to upgrade? We can go down two routes: either make pkg-config a requirement, or make it an optional extra. You won't get the benefits of pkg-config if it can't be included, of course, but at least it shouldn't fail. I think I'll opt for the latter, when I work out how it can be done. If @milancurcic were to release a new version containing my recent patch (#37), you will be able to set both FC and FCFLAGS as required. |
@guziy What OS + version are you using? My pkg-config is at version 0.26, and I'm using an LTS. |
@guziy Until we merge @blippy's PR #37 (probably later today), try with his master branch which should be able to handle FCFLAGS. |
I am on CentOS, where I cannot really install stuff... Thought to just test pgi, since I do not have it locally..
Cheers |
@guzly If you don't have admin privileges, there are (legitimate) ways around it. The standard location for per-user files is ~/.local. So binaries would typically be installed in ~/.local/bin, etc. You add those to your path, and hey presto. I see that 6.6 was released in Oct 2014, which is actually not all that long ago. I guess it must take a long time for software to make it into the distro. |
Hmmm, I suspect the problem is with automake. On my Ubuntu LTS, $ automake --version automake (GNU automake) 1.14.1 $ automake Makefile.am:3: error: 'pkgconfig_DATA' is used but 'pkgconfigdir' is undefined whereas on my Arch box $ automake --version automake (GNU automake) 1.15 and it works fine. I'll need to go away and think about that. |
In fact, I'm not sure what the problem is wrt to the older versions of automake. |
Hold on, I might have a solution. |
Ah. Got it, I think. Standby. |
@guziy et al : OK. I think I got it. the latest patch I submitted should fix everyone (!). autoreconf -i You'll need to wait for @milancurcic to merge the patches. Then it should work. Let me know how you get on. |
I've generated the Makefile using your master.
Cheers |
@guziy Thanks. What is your pgf90 version again? It's sad that it breaks as something as trivial as
Note that blippy's PR has been merged into my master so you can use that one as well. |
Is it still v14.4? May be worth reporting. |
Yes, it is still 14.4:
Cheers |
Deleting '::' at 3 places helps to compile the library but the compilation of the tests still fails:
Cheers |
sure, but the PGI compiler still ICE-s. |
Are you still having this problem? |
Hi @issamsaid: Yes, I have the same error with PGI14.4 and PGI14.9 fails with segfault (11)...
PS: I've tried after 'git pull' from the master. Cheers |
PGI: "Why support new language features when you can half-ass support for old ones...." |
I'm so unmotivated about PGI that I haven't even tried their free compiler. :) |
I did and it looks that it is not much more changed from the closed version I stopped to use in the 2008... |
@issamsaid Do you have an easy way to pass this issue to the development team? |
@milancurcic I already did, I personally ran in a similar issue. I will get back to you if I have some fixes. |
I am getting the same errors with: pgf90 15.10-0 64-bit target on x86-64 Linux -tp haswell If I remove '::' at mod_timedelta.f90 line 73, I get the same segfault (11). Any chance you found a solution @issamsaid or @guziy? Thanks in advance! |
Hi Guys please try the 17.7 it is released today. My errors were fixed and
those fixes were included in 17.7. hopefully this will help you as well.
…On Aug 29, 2017 11:54 PM, "fullerrd" ***@***.***> wrote:
I am getting the same errors with:
pgf90 15.10-0 64-bit target on x86-64 Linux -tp haswell
The Portland Group - PGI Compilers and Tools
Copyright (c) 2015, NVIDIA CORPORATION. All rights reserved.
If I remove '::' at mod_timedelta.f90 line 73, I get the same segfault
(11).
Any chance you found a solution @issamsaid <https://github.com/issamsaid>
or @guziy <https://github.com/guziy>? Thanks in advance!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#15 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AActMVNKQOw7NH53wtRyf8babkBHma5tks5sdJaWgaJpZM4Cpfb_>
.
|
Hi, @issamsaid. I upgraded pgf90 to v17.7 and got a fresh copy of the 1.6.0 release (from here: https://github.com/wavebitscientific/datetime-fortran/releases/tag/v1.6.0 ). I'm running Making all in src/lib Was anybody able to compile this with pgf90? Any suggestions? Thanks! |
@fullerrd What happens if you remove |
@milancurcic After removing :: and running pgf90 -c -g mod_constants.f90 |
please file a bug at https://nvbugswb.nvidia.com/ or post a comment on https://devtalk.nvidia.com/ |
Is this still an issue? Or is it resolved on newer versions of PGI? |
I haven't had opportunity to try it and am curious about it as well. |
nvfortran (PGI is now totally obsolete) 23.1 builds fine, but segfaults on test.
|
I know this is might not be an issue for fort or gfortran ( for sure not an issue - tested) but with 14.4 compiling tests fails:
They say they fixed it in 14.7 (Have no idea if it is true).
The text was updated successfully, but these errors were encountered: