[knem-devel] Test failures

Brice Goglin Brice.Goglin at inria.fr
Thu Feb 3 14:24:52 CET 2011


Hello,

I don't see any problem on my machine with 2.6.37. What's your hardware
and distribution?
Did you run the same test in the past without failure? I don't see any
recent commit that could explain such a problem.
Can you attach a gdb or valgrind and see what it says?

Brice



Le 03/02/2011 14:02, ryuuta a écrit :
> Hi,
>
> I have just checked out knem from git repository, successfully built,
> and installed it.
> Then I ran tests to get 2 failures:
>
> ********************************************
> Checking loopback performance with all flags
> DMA support not available, ignoring it
>      1024: 3.516 us 291.27 MB/s 277.78 MiB/s
>      2048: 9.484 us 215.93 MB/s 205.93 MiB/s
>      4096: 17.395 us 235.48 MB/s 224.57 MiB/s
>      8192: 20.215 us 405.25 MB/s 386.47 MiB/s
> *** glibc detected *** ../knem_loopback: free(): invalid pointer:
> 0x08568010 ***
> ======= Backtrace: =========
> /lib/libc.so.6(+0x6b661)[0x400a8661]
> /lib/libc.so.6(+0x6cf7b)[0x400a9f7b]
> /lib/libc.so.6(cfree+0x6d)[0x400ad11d]
> ../knem_loopback[0x80490e9]
> /lib/libc.so.6(__libc_start_main+0xe6)[0x40053db6]
> ../knem_loopback[0x8048641]
> ======= Memory map: ========
> 08048000-0804a000 r-xp 00000000 08:14 2242611  
>  /home/ryuta/devel/knem/src/build/tools/knem_loopback
> 0804a000-0804b000 rw-p 00001000 08:14 2242611  
>  /home/ryuta/devel/knem/src/build/tools/knem_loopback
> 08566000-08587000 rw-p 00000000 00:00 0          [heap]
> 40000000-4001c000 r-xp 00000000 08:13 748        /lib/ld-2.13.so
> <http://ld-2.13.so>
> 4001c000-4001d000 r--p 0001b000 08:13 748        /lib/ld-2.13.so
> <http://ld-2.13.so>
> 4001d000-4001e000 rw-p 0001c000 08:13 748        /lib/ld-2.13.so
> <http://ld-2.13.so>
> 4001e000-4001f000 r-xp 00000000 00:00 0          [vdso]
> 4001f000-40020000 rw-p 00000000 00:00 0 
> 40020000-40021000 rw-s 00000000 00:05 137123     /dev/knem
> 40021000-40022000 rw-p 00000000 00:00 0 
> 4003d000-40182000 r-xp 00000000 08:13 725        /lib/libc-2.13.so
> <http://libc-2.13.so>
> 40182000-40183000 ---p 00145000 08:13 725        /lib/libc-2.13.so
> <http://libc-2.13.so>
> 40183000-40185000 r--p 00145000 08:13 725        /lib/libc-2.13.so
> <http://libc-2.13.so>
> 40185000-40186000 rw-p 00147000 08:13 725        /lib/libc-2.13.so
> <http://libc-2.13.so>
> 40186000-4018a000 rw-p 00000000 00:00 0 
> 401a7000-401c2000 r-xp 00000000 08:13 62492      /usr/lib/libgcc_s.so.1
> 401c2000-401c3000 rw-p 0001a000 08:13 62492      /usr/lib/libgcc_s.so.1
> 40200000-40221000 rw-p 00000000 00:00 0 
> 40221000-40300000 ---p 00000000 00:00 0 
> bfc8c000-bfcad000 rw-p 00000000 00:00 0          [stack]
> ./loopback2: line 45: 27363 Aborted                 ../knem_loopback -F 63
> FAIL: loopback2
> *****************************
>
> Checking master-initiated scatter performance
> aggregate throughput moving buffer chunks from master (rank 0) to
> slaves (ranks 1-1)
> processing data transfers in master process
>      1024 to/among 1 procs: 4.562 us 224.44 MB/s 214.04 MiB/s
>      2048 to/among 1 procs: 4.574 us 447.73 MB/s 426.99 MiB/s
>      4096 to/among 1 procs: 11.918 us 343.68 MB/s 327.76 MiB/s
>      8192 to/among 1 procs: 5.531 us 1481.04 MB/s 1412.43 MiB/s
>     16384 to/among 1 procs: 17.977 us 911.41 MB/s 869.19 MiB/s
> ./collectives3: line 45: 27418 Killed                
>  ../knem_collectives -D -m -s
> FAIL: collectives3
>
> My kernel is 2.6.37.
>
> Thanks,
>
> Ryuta
>
>
> _______________________________________________
> knem-devel mailing list
> knem-devel at lists.gforge.inria.fr
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/knem-devel
>   

-------------- section suivante --------------
Une pi?ce jointe HTML a ?t? nettoy?e...
URL: <http://lists.gforge.inria.fr/pipermail/knem-devel/attachments/20110203/1bec44a3/attachment.htm>


More information about the knem-devel mailing list