[libvirt] [PATCH] The numa node numbers can be non-sequential.
On PowerNV platform the numa node numbers can be non-sequential, #numactl --hardware node distances: node 0 1 16 17 0: 10 40 40 40 1: 40 10 40 40 16: 40 40 10 40 17: 40 40 40 10 The numa node number is used as index and vice versa which can cause the libvirt to crash. Signed-off-by: Shivaprasad G Bhat <sbhat@linux.vnet.ibm.com> Signed-off-by: Pradipta Kumar Banerjee <bpradip@in.ibm.com> --- src/conf/capabilities.c | 12 ++++++++++-- src/qemu/qemu_driver.c | 5 +++-- src/qemu/qemu_process.c | 2 +- 3 files changed, 14 insertions(+), 5 deletions(-) diff --git a/src/conf/capabilities.c b/src/conf/capabilities.c index 726ec3f..78f65cb 100644 --- a/src/conf/capabilities.c +++ b/src/conf/capabilities.c @@ -1000,10 +1000,18 @@ virCapabilitiesGetCpusForNode(virCapsPtr caps, size_t node, virBitmapPtr cpumask) { - virCapsHostNUMACellPtr cell = caps->host.numaCell[node]; + virCapsHostNUMACellPtr cell = NULL; size_t cpu; + size_t index; + /* The numa node numbers can be non-contiguous. Ex: 0,1,16,17. */ + for (index = 0; index < caps->host.nnumaCell; index++) { + if (caps->host.numaCell[index]->num == node) { + cell = caps->host.numaCell[index]; + break; + } + } - for (cpu = 0; cpu < cell->ncpus; cpu++) { + for (cpu = 0; cell && cpu < cell->ncpus; cpu++) { if (virBitmapSetBit(cpumask, cell->cpus[cpu].id) < 0) { virReportError(VIR_ERR_INTERNAL_ERROR, _("Cpu '%u' in node '%zu' is out of range " diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 6b3825a..ae36fbe 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -8532,12 +8532,13 @@ qemuDomainSetNumaParamsLive(virDomainObjPtr vm, for (i = 0; i < caps->host.nnumaCell; i++) { bool result; - if (virBitmapGetBit(nodeset, i, &result) < 0) { + virCapsHostNUMACellPtr cell = caps->host.numaCell[i]; + if (virBitmapGetBit(nodeset, cell->num, &result) < 0) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Failed to get cpuset bit values")); goto cleanup; } - if (result && (virBitmapSetBit(temp_nodeset, i) < 0)) { + if (result && (virBitmapSetBit(temp_nodeset, cell->num) < 0)) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Failed to set temporary cpuset bit values")); goto cleanup; diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 8bcd98e..c28f84d 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -2002,7 +2002,7 @@ qemuPrepareCpumap(virQEMUDriverPtr driver, size_t j; int cur_ncpus = caps->host.numaCell[i]->ncpus; bool result; - if (virBitmapGetBit(nodemask, i, &result) < 0) { + if (virBitmapGetBit(nodemask, caps->host.numaCell[i]->num, &result) < 0) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Failed to convert nodeset to cpuset")); virBitmapFree(cpumap);
On Tue, Feb 04, 2014 at 05:57:22AM -0500, Shivaprasad G Bhat wrote:
On PowerNV platform the numa node numbers can be non-sequential, #numactl --hardware node distances: node 0 1 16 17 0: 10 40 40 40 1: 40 10 40 40 16: 40 40 10 40 17: 40 40 40 10
The numa node number is used as index and vice versa which can cause the libvirt to crash.
Signed-off-by: Shivaprasad G Bhat <sbhat@linux.vnet.ibm.com> Signed-off-by: Pradipta Kumar Banerjee <bpradip@in.ibm.com> --- src/conf/capabilities.c | 12 ++++++++++-- src/qemu/qemu_driver.c | 5 +++-- src/qemu/qemu_process.c | 2 +- 3 files changed, 14 insertions(+), 5 deletions(-)
diff --git a/src/conf/capabilities.c b/src/conf/capabilities.c index 726ec3f..78f65cb 100644 --- a/src/conf/capabilities.c +++ b/src/conf/capabilities.c @@ -1000,10 +1000,18 @@ virCapabilitiesGetCpusForNode(virCapsPtr caps, size_t node, virBitmapPtr cpumask) { - virCapsHostNUMACellPtr cell = caps->host.numaCell[node]; + virCapsHostNUMACellPtr cell = NULL; size_t cpu; + size_t index; + /* The numa node numbers can be non-contiguous. Ex: 0,1,16,17. */ + for (index = 0; index < caps->host.nnumaCell; index++) { + if (caps->host.numaCell[index]->num == node) { + cell = caps->host.numaCell[index]; + break; + } + }
- for (cpu = 0; cpu < cell->ncpus; cpu++) { + for (cpu = 0; cell && cpu < cell->ncpus; cpu++) { if (virBitmapSetBit(cpumask, cell->cpus[cpu].id) < 0) { virReportError(VIR_ERR_INTERNAL_ERROR, _("Cpu '%u' in node '%zu' is out of range "
At least for this code, I think I'd like to see a test case written. We don't have any existing test case which is targetting the src/conf/capabilities.c APIs directly, so just start a new test case tests/vircapstest.c and test this one API in it. Ideally the test should crash, unless this code fix is applied, so we can demonstrate the fix is working.
diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 6b3825a..ae36fbe 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -8532,12 +8532,13 @@ qemuDomainSetNumaParamsLive(virDomainObjPtr vm,
for (i = 0; i < caps->host.nnumaCell; i++) { bool result; - if (virBitmapGetBit(nodeset, i, &result) < 0) { + virCapsHostNUMACellPtr cell = caps->host.numaCell[i]; + if (virBitmapGetBit(nodeset, cell->num, &result) < 0) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Failed to get cpuset bit values")); goto cleanup; } - if (result && (virBitmapSetBit(temp_nodeset, i) < 0)) { + if (result && (virBitmapSetBit(temp_nodeset, cell->num) < 0)) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Failed to set temporary cpuset bit values")); goto cleanup; diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 8bcd98e..c28f84d 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -2002,7 +2002,7 @@ qemuPrepareCpumap(virQEMUDriverPtr driver, size_t j; int cur_ncpus = caps->host.numaCell[i]->ncpus; bool result; - if (virBitmapGetBit(nodemask, i, &result) < 0) { + if (virBitmapGetBit(nodemask, caps->host.numaCell[i]->num, &result) < 0) { virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Failed to convert nodeset to cpuset")); virBitmapFree(cpumap);
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 :|
participants (2)
-
Daniel P. Berrange -
Shivaprasad G Bhat