#191 Problems building make lapack_test_al_pt

Developer
closed-fixed
5
2012-01-01
2011-12-15
Anonymous
No

After reading one of the bug reports I found about some LAPACK tests. I issued "make lapack_test_al_pt" from the build directory. It starts building, but errors out with the following message:
LAS/build/lib/libatlas.a -o ../xlintsts
/home/melrobin/octave/ATLAS/build/lib/libatlas.a(ATL_thread_start.o): In function `ATL_thread_start':
ATL_thread_start.c:(.text+0xc2): undefined reference to `pthread_attr_setaffinity_np'
ATL_thread_start.c:(.text+0xe0): undefined reference to `pthread_create'
/home/melrobin/octave/ATLAS/build/lib/libatlas.a(ATL_thread_join.o): In function `ATL_thread_join':
ATL_thread_join.c:(.text+0xa): undefined reference to `pthread_join'
collect2: ld returned 1 exit status
make[3]: *** [../xlintsts] Error 1
make[3]: Leaving directory `/home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST/LIN'
make[2]: *** [ATL_all] Error 2
make[2]: Leaving directory `/home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST'
mv /home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST/*.out /home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST/LAOUT_AL_PT/.
mv: cannot stat `/home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST/*.out': No such file or directory
make[1]: *** [lapack_test_al_pt] Error 1
make[1]: Leaving directory `/home/melrobin/octave/ATLAS/build'
make: *** [lapack_test_al_pt] Error 2

I performed this as root. Trying to perform it as a normal user led to a permission error. There seems to some problem addressing my pthread library, but "make ptcheck" works. Is there something that I can do to fix these errors?

Discussion

  • Comment has been marked as spam. 
    Undo

    You can see all pending comments posted by this user  here

    Anonymous - 2011-12-15
    • milestone: --> 148063
     
  • Comment has been marked as spam. 
    Undo

    You can see all pending comments posted by this user  here

    Anonymous - 2011-12-27

    I tried this with the 3.9.59 and it still fails, but no longer gives me the permission errors. I looked through the Makefile and found some other targets and some of them work, but others fail.

     
  • R. Clint Whaley

    R. Clint Whaley - 2011-12-29
    • assigned_to: nobody --> rwhaley
     
  • Comment has been marked as spam. 
    Undo

    You can see all pending comments posted by this user  here

    Anonymous - 2011-12-29

    Error report for make lapack tests

     
  • Comment has been marked as spam. 
    Undo

    You can see all pending comments posted by this user  here

    Anonymous - 2011-12-29

    Please let me know if this is what you need. I'm trying to learn and have had some problems in doing this in the recent past.

    I'm also willing to rebuild and try anything else to troubleshoot.

     
  • R. Clint Whaley

    R. Clint Whaley - 2011-12-29
    • milestone: 148063 -->
    • labels: 360153 -->
    • status: open --> open-accepted
     
  • R. Clint Whaley

    R. Clint Whaley - 2011-12-29

    This is a confirmed bug in the ATLAS build mechanism

     
  • R. Clint Whaley

    R. Clint Whaley - 2011-12-29

    To fix, change line 95 of BLDdir/Make.top from:
    BLASLIB="$(PTF77BLASlib) $(PTCBLASlib) $(ATLASlib)" \ to:
    BLASLIB="$(PTF77BLASlib) $(PTCBLASlib) $(ATLASlib) $(LIBS)" \

    Pretty much all the testing targets need to add $(LIBS) to the BLASLIB macro.

     
  • R. Clint Whaley

    R. Clint Whaley - 2011-12-29

    Fixed in basefiles for 3.9.60.

     
  • R. Clint Whaley

    R. Clint Whaley - 2011-12-30

    Confirm the dynamic linking problem fixed in 3.9.60, but there is a seg fault that I believe comes from the just discovered SGER bug.

     
  • Comment has been marked as spam. 
    Undo

    You can see all pending comments posted by this user  here

    Anonymous - 2011-12-31

    What's the number of the SGER bug?

     
  • Comment has been marked as spam. 
    Undo

    You can see all pending comments posted by this user  here

    Anonymous - 2011-12-31

    I just rebuilt and tested and it make does complete, so we're good there. It does segfault as you have mentioned. Here is the output:
    make[3]: Entering directory `/home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST'
    Testing REAL LAPACK linear equation routines
    ./xlintsts < stest.in > stest.out 2>&1
    /bin/sh: line 1: 5388 Segmentation fault ./xlintsts < stest.in > stest.out 2>&1
    make[3]: *** [stest.out] Error 139
    make[3]: Leaving directory `/home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST'
    make[2]: *** [ATL_all] Error 2
    make[2]: Leaving directory `/home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST'
    mv /home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST/*.out /home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST/LAOUT_AL_PT/.
    make -f Make.top scope_lapack_test_al_pt | tee \ /home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST/SUMMARY_al_pt.txt 2>&1
    make[2]: Entering directory `/home/melrobin/octave/ATLAS/build'
    cd /home/melrobin/octave/ATLAS/build/bin/LAPACK_TEST/LAOUT_AL_PT ; \ fgrep -e fault -e FAULT -e error -e ERROR -e fail -e FAIL *.out \ | fgrep -v "error exits"
    make[2]: *** [scope_lapack_test_al_pt] Error 1
    make[2]: Leaving directory `/home/melrobin/octave/ATLAS/build'
    make[1]: Leaving directory `/home/melrobin/octave/ATLAS/build'
    [melrobin@melrobin build]$

    This is as is expected, right?

     
  • R. Clint Whaley

    R. Clint Whaley - 2012-01-01
    • milestone: --> Developer
    • labels: --> Incorrect answer
    • status: open-accepted --> closed-fixed
     
  • R. Clint Whaley

    R. Clint Whaley - 2012-01-01

    Mel (or is that Robin),

    Sorry, I forgot to come back here. I can't be sure our seg faults are the same, but you can see what causes mine at:
    http://www.cs.utsa.edu/~whaley/ATL310/

    Since the QR variants are all passing now (as of 60), I suspect an error on ORMQR or related routine. Will hopefully figure it out soon.

    In the meantime, I'm closing this report with the linking/loader problem solved. If you want to open up a new report on the seg fault, feel free, and I'll try to update it once I fixed my own lapck seg fault!

    Thanks,
    Clint

     

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.





No, thanks