Re: [hwloc-devel] hwloc on ARM [Was: hwloc at SC09]
Ashley Pittman, le Tue 24 Nov 2009 17:25:18 +, a écrit : > Do you mean the attached file? Yes. It seems to work with the v0.9 branch so it's all good. > gather-topology.sh failed Yes, after building the .tar.gz it dumps what lstopo would see, thus the same failure. Thanks, Samuel
Re: [hwloc-devel] hwloc on ARM [Was: hwloc at SC09]
Ashley Pittman, le Tue 24 Nov 2009 17:25:18 +, a écrit : > Just to be absolutely clear, head SVN works but the error is running > from a version from two weeks ago. Ah, ok. Samuel
Re: [hwloc-devel] hwloc on ARM [Was: hwloc at SC09]
Just to be absolutely clear, head SVN works but the error is running from a version from two weeks ago. On Tue, 2009-11-24 at 17:27 +0100, Samuel Thibault wrote: > Ashley Pittman, le Tue 24 Nov 2009 16:18:36 +, a écrit : > > * Topology extraction from /proc/cpuinfo * > > > > 0 online processors found, with id max 0 > > That was it. > > > Removing empty objects except numa nodes and PCI devices > > Segmentation fault > > Not so fine, then :) Could you post a backtrace and the output of > gather-topology.sh ? Do you mean the attached file? gather-topology.sh failed (gdb) where full #0 hwloc_obj_snprintf (string=0xbe913b40 "", size=256, topology=, l=0x0, _indexprefix=0x4009d670 "#", verbose=1) at traversal.c:177 type = indexprefix = os_index = "\000\000\000\000\001\000\000\000xx\n@" #1 0x400959ac in print_objects (topology=0x19050, indent=0, obj=0x0) at topology.c:373 line = "\000\000\000\000\001\000\000\000�u\000@L\000\000\000\002\000 \000\000�]\001@(\\\001@\000\000\000\000\000\000\000\000'C\t@L\000\000 \000\000�\001@�]\001@(\\\001@", '\0' , "\001\000\000 \000\000\000\000\000\001\000\000\000\000\000\000\000L\000\000\000X� \001@D:\t@(\\\001@\001\000\000\000D:\t@U\000\000\000$>\t@\\t@ #2 0x400979c8 in hwloc_topology_load (topology=0x4001ce14) at topology.c:1009 No locals. #3 0x93e4 in main (argc=1, argv=0x1) at lstopo.c:206 err = 0 verbose_mode = 1 topology = (hwloc_topology_t) 0x401b8000 filename = 0x0 flags = 1073860116 merge = -109932 ignorecache = 0 callname = 0xfbd0 "pD-�<�\237�\n�\217�;���4 \237�40\237�\0030b�Ca�� \0020\212�p\204�\b\003P��" synthetic = 0x1 fsysroot = 0x9270 "�O-�\001`��" opt = -- Ashley Pittman, Bath, UK. Padb - A parallel job inspection tool for cluster computing http://padb.pittman.org.uk arm.tar.gz Description: application/compressed-tar
Re: [hwloc-devel] hwloc on ARM [Was: hwloc at SC09]
Ashley Pittman, le Tue 24 Nov 2009 16:18:36 +, a écrit : > * Topology extraction from /proc/cpuinfo * > > 0 online processors found, with id max 0 That was it. > Removing empty objects except numa nodes and PCI devices > Segmentation fault Not so fine, then :) Could you post a backtrace and the output of gather-topology.sh ? Samuel
Re: [hwloc-devel] hwloc on ARM [Was: hwloc at SC09]
I did a svn update this morning and the code works fine now. $ lstopo System(29MB) + P#0 $ On Sat, 2009-11-14 at 01:56 +0100, Samuel Thibault wrote: > So at some point the traversal ends up with a NULL pointer, weird. What > would be useful is output after passing --enable-debug to ./configure . For reference here is the offending output after building the old version with --enable-debug $ lstopo No cgroup or cpuset found No cgroup or cpuset found * Topology extraction from /proc/cpuinfo * 0 online processors found, with id max 0 0 * Topology summary * 0 processors (0 max id) 0 sockets 0 cores * CPU cpusets * System(29MB HP=0*0kB ) ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, Computing the system cpuset by ORing all Proc objects Applying the system cpuset to all nodes Removing empty objects except numa nodes and PCI devices Segmentation fault $ /proc/cpuinfo contains: Processor : XScale-IXP42x Family rev 1 (v5l) BogoMIPS: 266.24 Features: swp half fastmult edsp CPU implementer : 0x69 CPU architecture: 5TE CPU variant : 0x0 CPU part: 0x41f CPU revision: 1 Cache type : undefined 5 Cache clean : undefined 5 Cache lockdown : undefined 5 Cache format: Harvard I size : 32768 I assoc : 32 I line length : 32 I sets : 32 D size : 32768 D assoc : 32 D line length : 32 D sets : 32 Hardware: Linksys NSLU2 Revision: Serial : -- Ashley Pittman, Bath, UK. Padb - A parallel job inspection tool for cluster computing http://padb.pittman.org.uk