Thanks, Tim. (Yes, I'm still here. Just been on vacation since you sent your patches.) I tested your patches with the usual test suite, on an 8-node ia64 and on an 8-node x86_64 (uv). Now available at ftp://oss.sgi.com/www/projects/libnuma/download/ as numactl-2.0.6-rc1.tar.gz -Cliff On Wed, Sep 22, 2010 at 11:03:47AM -0700, Tim Pepper wrote: > In response to a bug report where 'numactl --hardware' was segfaulting > with distro 2.0.3 version I've dug into the current state of > libnuma/numactl after not having looked at it in quite some time. The > following series of patches resulted and hopefully clarifies things a bit > for the next person who comes along and accounts for the possibility of > sparsely numbered, memoryless and cpuless nodes in 'numactl --hardware'. > There are certainly some things in the patches that aren't quite optimal > and other things I saw that I didn't try to tackle (eg: there is a lot > of seeming inconsistency around the different "all" node pointers), > but the updated code works for me and I was trying to limit changes to > not start driving in the direction of a v3 library (or hwloc?). > > The first 4 patches are relatively minor cleanups. > > The bulk is in patch 5 which then deals with better handling sparse node > numbering and fixes parse_numbers() in particular which was experiencing > a buffer overflow in the face of this topology: > > available: 2 nodes (0,3) > node 0 cpus: none > node 0 size: 0 MB > node 0 free: 0 MB > node 3 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 > node 3 size: 15360 MB > node 3 free: 12524 MB > node distances: > node 0 3 > 0: 10 40 > 3: 40 10 > > I've tested the changes on "interesting" topology hardware to which I > have access, but would greatly appreciate if others might do similar. > > -- > Tim Pepper <lnxninja@xxxxxxxxxxxxxxxxxx> > IBM Linux Technology Center > -- > To unsubscribe from this list: send the line "unsubscribe linux-numa" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- Cliff Wickman SGI cpw@xxxxxxx (651) 683-3824 -- To unsubscribe from this list: send the line "unsubscribe linux-numa" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
