On Sat, Jul 13, 2013 at 02:27:03AM +0200, Dario Faggioli wrote:
> Starting from Xen 4.2, libxl has all the bits and pieces in place
> for retrieving an adequate amount of information about the host
> NUMA topology. It is therefore possible, after a bit of shuffling,
> to arrange those information in the way libvirt wants to present
> them to the outside world.
>
> Therefore, with this patch, the <topology> section of the host
> capabilities is properly populated, when running on Xen, so that
> we can figure out whether or not we're running on a NUMA host,
> and what its characteristics are.
>
> [raistlin@Zhaman ~]$ sudo virsh --connect xen:/// capabilities
> <capabilities>
> <host>
> <cpu>
> ....
> <topology>
> <cells num='2'>
> <cell id='0'>
> <memory unit='KiB'>6291456</memory>
> <cpus num='8'>
> <cpu id='0' socket_id='1' core_id='0' siblings='0-1'/>
> <cpu id='1' socket_id='1' core_id='0' siblings='0-1'/>
> <cpu id='2' socket_id='1' core_id='1' siblings='2-3'/>
> <cpu id='3' socket_id='1' core_id='1' siblings='2-3'/>
> <cpu id='4' socket_id='1' core_id='9' siblings='4-5'/>
> <cpu id='5' socket_id='1' core_id='9' siblings='4-5'/>
> <cpu id='6' socket_id='1' core_id='10' siblings='6-7'/>
> <cpu id='7' socket_id='1' core_id='10' siblings='6-7'/>
> </cpus>
> </cell>
> <cell id='1'>
> <memory unit='KiB'>6881280</memory>
> <cpus num='8'>
> <cpu id='8' socket_id='0' core_id='0' siblings='8-9'/>
> <cpu id='9' socket_id='0' core_id='0' siblings='8-9'/>
> <cpu id='10' socket_id='0' core_id='1' siblings='10-11'/>
> <cpu id='11' socket_id='0' core_id='1' siblings='10-11'/>
> <cpu id='12' socket_id='0' core_id='9' siblings='12-13'/>
> <cpu id='13' socket_id='0' core_id='9' siblings='12-13'/>
> <cpu id='14' socket_id='0' core_id='10' siblings='14-15'/>
> <cpu id='15' socket_id='0' core_id='10' siblings='14-15'/>
> </cpus>
> </cell>
> </cells>
> </topology>
> </host>
> ....
>
> Signed-off-by: Dario Faggioli <dario.faggioli@xxxxxxxxxx>
> ---
> Changes from v2:
> * iterators turned from int to size_t;
> * fixed wrong sibling maps if on same node but different socket;
> * code motion and error handling, as requested during review.
>
> Changes from v1:
> * fixed a typo in the commit message, as requested during review;
> * fixed coding style (one function parameters per line and no spaces
> between variable definitions), as requested during review;
> * avoid zero-filling memory after VIR_ALLOC_N(), since it does that
> already, as requested during review;
> * improved out of memory error reporting, as requested during review;
> * libxlMakeNumaCapabilities() created, accommodating all the NUMA
> related additions, instead of having them within
> libxlMakeCapabilitiesInternal(), as suggested during review.
> ---
> src/libxl/libxl_conf.c | 148 ++++++++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 147 insertions(+), 1 deletion(-)
>
> diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
> index 4a0fba9..c097d1e 100644
> --- a/src/libxl/libxl_conf.c
> +++ b/src/libxl/libxl_conf.c
> @@ -161,6 +161,117 @@ libxlBuildCapabilities(virArch hostarch,
> }
>
> static virCapsPtr
> +libxlMakeNumaCapabilities(libxl_numainfo *numa_info,
> + int nr_nodes,
> + libxl_cputopology *cpu_topo,
> + int nr_cpus,
> + virCapsPtr caps)
> +{
> + virCapsHostNUMACellCPUPtr *cpus = NULL;
> + int *nr_cpus_node = NULL;
> + bool numa_failed = false;
> + size_t i;
> +
> + if (VIR_ALLOC_N(cpus, nr_nodes)) {
> + virReportOOMError();
> + return caps;
> + }
> +
> + if (VIR_ALLOC_N(nr_cpus_node, nr_nodes)) {
> + VIR_FREE(cpus);
> + virReportOOMError();
> + return caps;
> + }
Afraid some more changes have landed in GIT that affect you.
All the VIR_*ALLOC* functions, except those suffixed _QUIET
will now call virReportOOMError() for you. So you can remove
virReportOOMError here and all other places related to these
alloc fnuctions.
Also, you should be checking 'VIR_ALLOC_N() < 0', not just
for a non-zero value.
> +
> + /* For each node, prepare a list of CPUs belonging to that node */
> + for (i = 0; i < nr_cpus; i++) {
> + int node = cpu_topo[i].node;
> +
> + if (cpu_topo[i].core == LIBXL_CPUTOPOLOGY_INVALID_ENTRY)
> + continue;
> +
> + nr_cpus_node[node]++;
> +
> + if (nr_cpus_node[node] == 1) {
> + if (VIR_ALLOC(cpus[node]) < 0) {
> + virReportOOMError();
> + numa_failed = true;
> + goto cleanup;
> + }
> + }
> + else {
> + if (VIR_REALLOC_N(cpus[node], nr_cpus_node[node]) < 0) {
> + virReportOOMError();
> + numa_failed = true;
> + goto cleanup;
> + }
> + }
> +
> + /* Mapping between what libxl tells and what libvirt wants */
> + cpus[node][nr_cpus_node[node]-1].id = i;
> + cpus[node][nr_cpus_node[node]-1].socket_id = cpu_topo[i].socket;
> + cpus[node][nr_cpus_node[node]-1].core_id = cpu_topo[i].core;
> + /* Allocate the siblings maps. We will be filling them later */
> + cpus[node][nr_cpus_node[node]-1].siblings = virBitmapNew(nr_cpus);
> + if (!cpus[node][nr_cpus_node[node]-1].siblings) {
> + virReportOOMError();
> + numa_failed = true;
> + goto cleanup;
> + }
> + }
> +
> + /* Let's now populate the siblings bitmaps */
> + for (i = 0; i < nr_cpus; i++) {
> + int node = cpu_topo[i].node;
> + size_t j;
> +
> + if (cpu_topo[i].core == LIBXL_CPUTOPOLOGY_INVALID_ENTRY)
> + continue;
> +
> + for (j = 0; j < nr_cpus_node[node]; j++) {
> + if (cpus[node][j].socket_id == cpu_topo[i].socket &&
> + cpus[node][j].core_id == cpu_topo[i].core)
> + ignore_value(virBitmapSetBit(cpus[node][j].siblings, i));
> + }
> + }
> +
> + for (i = 0; i < nr_nodes; i++) {
> + if (numa_info[i].size == LIBXL_NUMAINFO_INVALID_ENTRY)
> + continue;
> +
> + if (virCapabilitiesAddHostNUMACell(caps, i, nr_cpus_node[i],
> + numa_info[i].size / 1024,
> + cpus[i]) < 0) {
> + virCapabilitiesClearHostNUMACellCPUTopology(cpus[i],
> + nr_cpus_node[i]);
> + numa_failed = true;
> + goto cleanup;
> + }
> +
> + /* This is safe, as the CPU list is now stored in the NUMA cell */
> + cpus[i] = NULL;
> + }
> +
> + cleanup:
> +
> + if (numa_failed) {
> + /* Looks like something went wrong. Well, that's bad, but probably
> + * not enough to break the whole driver, so we log and carry on */
> + for (i = 0; i < nr_nodes; i++) {
> + VIR_FREE(cpus[i]);
> + }
> + VIR_WARN("Failed to retrieve and build host NUMA topology properly,\n"
> + "disabling NUMA capabilities");
> + virCapabilitiesFreeNUMAInfo(caps);
This is wrong. All the cases of error leading to this point are out of
memory errors. It is madness to prevent everything is ok & carry on here.
This should be treated as a proper error.
> @@ -788,9 +903,40 @@ libxlMakeCapabilities(libxl_ctx *ctx)
> return NULL;
> }
>
> - return libxlMakeCapabilitiesInternal(virArchFromHost(),
> + caps = libxlMakeCapabilitiesInternal(virArchFromHost(),
> &phy_info,
> ver_info->capabilities);
> +
> + /* Check if caps is valid. If it is, it must remain so till the end! */
> + if (caps == NULL)
> + goto out;
> +
> + /* Let's try to fetch NUMA info now (not critical in case we fail) */
> + numa_info = libxl_get_numainfo(ctx, &nr_nodes);
> + if (numa_info == NULL)
> + VIR_WARN("libxl_get_numainfo failed to retrieve NUMA data");
Under what scenario can libxl_get_numainfo() return NULL ? Unless this
is an valid expected scenario, we should treat this is an error.
> + else {
> + /* If the above failed, we'd have no NUMa caps anyway! */
> + cpu_topo = libxl_get_cpu_topology(ctx, &nr_cpus);
> + if (cpu_topo == NULL) {
> + VIR_WARN("libxl_get_cpu_topology failed to retrieve topology");
> + libxl_numainfo_list_free(numa_info, nr_nodes);
> + }
Likewise here.
> + else {
> + /* And add topology information to caps */
> + caps = libxlMakeNumaCapabilities(numa_info,
> + nr_nodes,
> + cpu_topo,
> + nr_cpus,
> + caps);
> + }
> + }
> +
> + libxl_cputopology_list_free(cpu_topo, nr_cpus);
> + libxl_numainfo_list_free(numa_info, nr_nodes);
> +
> +out:
s/out/cleanup/ is our preferred label naming
> + return caps;
> }
Regards,
Daniel
--
|: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org -o- http://virt-manager.org :|
|: http://autobuild.org -o- http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list