Open MPI logo

Hardware Locality Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Hardware Locality Development mailing list

Subject: Re: [hwloc-devel] hwloc 1.1 rc2 make check fails on SLES10SP1 on PPC64
From: Brice Goglin (Brice.Goglin_at_[hidden])
Date: 2010-11-21 13:00:06

Le 21/11/2010 18:28, Christopher Samuel a écrit :
> #### use cpuinfo or sysfs?
> #### using hwloc_accessat
> #### hwloc_accessat failed No such file or directory
> #### using cpuinfo

This is caused by /sys/devices/system/cpu/cpu0/topology not existing in
the tarball. So this is actually expected. Can you check whether this
directory exists on the machine and whether it's empty or not?

Your former "good.txt" used sysfs+devicetree. Your "bad.txt" and
"bad2.txt" used cpuinfo without devicetree. It says:
"No cores and L2 cache were found in /proc/device-tree/cpus, exiting"

So two things:
1) it should use always cpuinfo instead of sysfs so something looks
wrong in the way we detect whether /sys/devices/system/cpu/cpu0/topology
exists. Where/how did you run your "good.txt" ? Is it a lstopo on the
actual machine without any options/envvariable?
2) it should find things in the devicetree, but we don't have a lot of
debug info in there unfortunately. let's look at (1) first