[Mvapich-discuss] ibmpifort.so: undefined reference to `CFI_is_contiguous'

Lieber, Matt lieber.31 at osu.edu
Fri Sep 29 15:30:10 EDT 2023


This issue has been resolved.  The solution is to use the patch from above and then run autogen.sh even with the new tars users may need to run autogen or else it will fail.


-Matt
________________________________
From: Korzennik, Sylvain <skorzennik at cfa.harvard.edu>
Sent: Friday, September 22, 2023 4:15 PM
To: Lieber, Matt <lieber.31 at osu.edu>
Cc: Announcement about MVAPICH2 (MPI over InfiniBand, RoCE, Omni-Path, iWARP and EFA) Libraries developed at NBCL/OSU <mvapich-discuss at lists.osu.edu>
Subject: Re: [Mvapich-discuss] ibmpifort.so: undefined reference to `CFI_is_contiguous'

With this version, I get to src/pm/mpirun/src/hostfile/parser. y: In function ‘commit’: src/pm/mpirun/src/hostfile/parser. y: 130: 15: warning: incompatible implicit declaration of built-in function ‘realloc’ [enabled by default] ptr = realloc(rank,
ZjQcmQRYFpfptBannerStart
This Message Is From an External Sender
This message came from outside your organization.
<https://us-phishalarm-ewt.proofpoint.com/EWT/v1/KGKeukY!vYQd06kpI8bsxtU60GXIFwsiQDotPAssirlBvAQ2SQJXQFgAR3wTsO8W773WLInWCm-vMd8XWNBtGaqqg4ANJk1abwzCv2cgW5SBTcnej8QNlbC3iLbw1PVrZ4HDMg$>
Report Suspicious

ZjQcmQRYFpfptBannerEnd
With this version, I get to
src/pm/mpirun/src/hostfile/parser.y: In function ‘commit’:
src/pm/mpirun/src/hostfile/parser.y:130:15: warning: incompatible implicit declaration of built-in function ‘realloc’ [enabled by default]
         ptr = realloc(rank, sizeof(struct rank_s) * n_alloc);
               ^
src/pm/mpirun/src/hostfile/parser.y: In function ‘read_hostfile’:
src/pm/mpirun/src/hostfile/parser.y:197:13: warning: incompatible implicit declaration of built-in function ‘exit’ [enabled by default]
             exit(EXIT_FAILURE);
             ^
src/pm/mpirun/src/hostfile/parser.y:197:18: error: ‘EXIT_FAILURE’ undeclared (first use in this function)
             exit(EXIT_FAILURE);
                  ^
src/pm/mpirun/src/hostfile/parser.y:197:18: note: each undeclared identifier is reported only once for each function it appears in
src/pm/mpirun/src/hostfile/parser.y:212:2: warning: incompatible implicit declaration of built-in function ‘exit’ [enabled by default]
  exit(EXIT_FAILURE);
  ^
src/pm/mpirun/src/hostfile/parser.y:222:9: warning: incompatible implicit declaration of built-in function ‘exit’ [enabled by default]
         exit(EXIT_FAILURE);
         ^
src/pm/mpirun/src/hostfile/parser.y:230:9: warning: incompatible implicit declaration of built-in function ‘exit’ [enabled by default]
         exit(EXIT_FAILURE);
         ^
make[2]: *** [src/pm/mpirun/src/hostfile/src_pm_mpirun_src_hostfile_libhostfile_a-parser.o] Error 1
make[2]: Leaving directory `/local/hpc/nvidia/23.7/mvapich2-2.3.7-1'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/local/hpc/nvidia/23.7/mvapich2-2.3.7-1'
make: *** [all] Error 2

I'll poke around to see where EXIT_FAILURE is defined, etc... but the patch isn't clean. BTW, I used
./configure  --disable-f08  -prefix=/...

    Cheers,
      Sylvain
--



On Fri, Sep 22, 2023 at 11:34 AM Lieber, Matt <lieber.31 at osu.edu<mailto:lieber.31 at osu.edu>> wrote:
Hi Sylvain,
Sorry to hear you still have issues.  I think this is because autogen was not run which is required when changes to the configure.ac<https://urldefense.com/v3/__http://configure.ac__;!!KGKeukY!zavLw3zYlg7mz-SuhFcdW0-4R9sb5_Txlxo4iYmMiWBWyACvsoNpgPrAEPdftSUDGfHVYx9ai2YzeJCUrB5oufVJ0XhH$> are made.  Here is a tar that I have made please try these and let me know if there are still any issues.

[https://res-h3.public.cdn.office.net/assets/mail/file-icon/png/generic_16x16.png]mvapich2-2.3.7-1.tar.gz<https://urldefense.com/v3/__https://buckeyemailosu-my.sharepoint.com/:u:/g/personal/lieber_31_osu_edu/Edlz4H8TpXlAvUU9GUbgG70BTf20B55777B5EdBaht2iSA?e=wCqN3V__;!!KGKeukY!zavLw3zYlg7mz-SuhFcdW0-4R9sb5_Txlxo4iYmMiWBWyACvsoNpgPrAEPdftSUDGfHVYx9ai2YzeJCUrB5ouQ0z718W$>

Thanks,
Matt
________________________________
From: Korzennik, Sylvain <skorzennik at cfa.harvard.edu<mailto:skorzennik at cfa.harvard.edu>>
Sent: Thursday, September 21, 2023 5:33 PM
To: Lieber, Matt <lieber.31 at osu.edu<mailto:lieber.31 at osu.edu>>
Cc: Announcement about MVAPICH2 (MPI over InfiniBand, RoCE, Omni-Path, iWARP and EFA) Libraries developed at NBCL/OSU <mvapich-discuss at lists.osu.edu<mailto:mvapich-discuss at lists.osu.edu>>
Subject: Re: [Mvapich-discuss] ibmpifort.so: undefined reference to `CFI_is_contiguous'

Thx Matt, but 1- I had to execute /bin/sh /local/hpc/nvidia/23. 7/mvapich2-2. 3. 7-1/confdb/missing aclocal -I confdb /bin/sh /local/hpc/nvidia/23. 7/mvapich2-2. 3. 7-1/confdb/missing automake --foreign b/c the make was trying to execute /bin/sh
Thx Matt, but

1- I had to execute
 /bin/sh /local/hpc/nvidia/23.7/mvapich2-2.3.7-1/confdb/missing aclocal -I  confdb
 /bin/sh /local/hpc/nvidia/23.7/mvapich2-2.3.7-1/confdb/missing automake  --foreign
b/c the make was trying to execute
 /bin/sh /local/hpc/nvidia/23.7/mvapich2-2.3.7-1/confdb/missing aclocal-1.15 -I  confdb
 /bin/sh /local/hpc/nvidia/23.7/mvapich2-2.3.7-1/confdb/missing automake-1.15  --foreign
b/c the configure.ac<https://urldefense.com/v3/__http://configure.ac__;!!KGKeukY!2msSyVSCwsxQmNkdBFxbFYZi7st28k_-41G55m9ua21Gx6frh-dQaJ8e80ElBTMtWC8IYCWbjqPCnl3y9x0HDygfO0Se$> was modified by the patching, and looking for aclocal-1.15 automake-1.15 fail.

 After that the make reran the config (--recheck) and then croaks on

make  all-recursive
make[1]: Entering directory `/local/hpc/nvidia/23.7/mvapich2-2.3.7-1'
make[1]: execvp: /bin/sh: Argument list too long
make[1]: *** [all-recursive] Error 127
make[1]: Leaving directory `/local/hpc/nvidia/23.7/mvapich2-2.3.7-1'
make: *** [all] Error 2

What am I missing???

    Cheers,
      Sylvain
--




On Thu, Sep 21, 2023 at 1:32 PM Lieber, Matt <lieber.31 at osu.edu<mailto:lieber.31 at osu.edu>> wrote:
Hi Sylvain,
Thank you for the report, we think we have a patch that can fix this issue.  Please use the attached patch with --disable-f08 and let us know if you have any other questions.


-Matt
________________________________
From: Mvapich-discuss <mvapich-discuss-bounces at lists.osu.edu<mailto:mvapich-discuss-bounces at lists.osu.edu>> on behalf of Korzennik, Sylvain via Mvapich-discuss <mvapich-discuss at lists.osu.edu<mailto:mvapich-discuss at lists.osu.edu>>
Sent: Tuesday, September 19, 2023 6:54 PM
To: Announcement about MVAPICH2 (MPI over InfiniBand, RoCE, Omni-Path, iWARP and EFA) Libraries developed at NBCL/OSU <mvapich-discuss at lists.osu.edu<mailto:mvapich-discuss at lists.osu.edu>>
Subject: [Mvapich-discuss] ibmpifort.so: undefined reference to `CFI_is_contiguous'

I was able to build mvapich2-2. 3. 7-1 w/ nvidia's compiler ver 23. 5 or 23. 7 under CO7 (w/ same tweaks as when using ver 23. 3), but I can't compile a trivial MPI test program w/ either version: % mpif90 -o ring0f ring0f. o /share/apps/nvidia/Linux_x86_64/23. 5/mvapich2-2. 3. 7-1/lib/libmpifort. so: 
I was able to build mvapich2-2.3.7-1 w/ nvidia's compiler ver 23.5 or 23.7 under CO7 (w/ same tweaks as when using ver 23.3), but I can't compile a trivial MPI test program w/ either version:

% mpif90 -o ring0f ring0f.o
/share/apps/nvidia/Linux_x86_64/23.5/mvapich2-2.3.7-1/lib/libmpifort.so: undefined reference to `CFI_is_contiguous'

or

% mpif90 -o ring0f ring0f.o
/share/apps/nvidia/Linux_x86_64/23.7/mvapich2-2.3.7-1/lib/libmpifort.so: undefined reference to `CFI_is_contiguous'

Anyone ran into this and has a solution? Rebuilding w/ --disable-f08 in the config step does not help (based on https://github.com/pmodels/mpich/issues/6505<https://urldefense.com/v3/__https://github.com/pmodels/mpich/issues/6505__;!!KGKeukY!2u0ZW6taxuRSfFe8Dk9FewQ4ND7Y-Ju0qBVz8Px7kDWV693LGxXFKlcoAd77SHwXNULF6xYZ9Lli75GOIA1Lc3yIM33ujPVzm2KqxA$> for MPICH)

    Cheers,
      Sylvain
--

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osu.edu/pipermail/mvapich-discuss/attachments/20230929/aa30417b/attachment-0002.html>


More information about the Mvapich-discuss mailing list