Message ID | 20250124120145.410066-1-elver@google.com (mailing list archive) |
---|---|
State | New |
Headers | show |
Series | kfence: skip __GFP_THISNODE allocations on NUMA systems | expand |
On 1/24/25 13:01, Marco Elver wrote: > On NUMA systems, __GFP_THISNODE indicates that an allocation _must_ be > on a particular node, and failure to allocate on the desired node will > result in a failed allocation. > > Skip __GFP_THISNODE allocations if we are running on a NUMA system, > since KFENCE can't guarantee which node its pool pages are allocated on. > > Reported-by: Vlastimil Babka <vbabka@suse.cz> > Cc: Christoph Lameter <cl@linux.com> > Fixes: 236e9f153852 ("kfence: skip all GFP_ZONEMASK allocations") > Signed-off-by: Marco Elver <elver@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Thanks. > --- > mm/kfence/core.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/mm/kfence/core.c b/mm/kfence/core.c > index 67fc321db79b..102048821c22 100644 > --- a/mm/kfence/core.c > +++ b/mm/kfence/core.c > @@ -21,6 +21,7 @@ > #include <linux/log2.h> > #include <linux/memblock.h> > #include <linux/moduleparam.h> > +#include <linux/nodemask.h> > #include <linux/notifier.h> > #include <linux/panic_notifier.h> > #include <linux/random.h> > @@ -1084,6 +1085,7 @@ void *__kfence_alloc(struct kmem_cache *s, size_t size, gfp_t flags) > * properties (e.g. reside in DMAable memory). > */ > if ((flags & GFP_ZONEMASK) || > + ((flags & __GFP_THISNODE) && num_online_nodes() > 1) || > (s->flags & (SLAB_CACHE_DMA | SLAB_CACHE_DMA32))) { > atomic_long_inc(&counters[KFENCE_COUNTER_SKIP_INCOMPAT]); > return NULL;
diff --git a/mm/kfence/core.c b/mm/kfence/core.c index 67fc321db79b..102048821c22 100644 --- a/mm/kfence/core.c +++ b/mm/kfence/core.c @@ -21,6 +21,7 @@ #include <linux/log2.h> #include <linux/memblock.h> #include <linux/moduleparam.h> +#include <linux/nodemask.h> #include <linux/notifier.h> #include <linux/panic_notifier.h> #include <linux/random.h> @@ -1084,6 +1085,7 @@ void *__kfence_alloc(struct kmem_cache *s, size_t size, gfp_t flags) * properties (e.g. reside in DMAable memory). */ if ((flags & GFP_ZONEMASK) || + ((flags & __GFP_THISNODE) && num_online_nodes() > 1) || (s->flags & (SLAB_CACHE_DMA | SLAB_CACHE_DMA32))) { atomic_long_inc(&counters[KFENCE_COUNTER_SKIP_INCOMPAT]); return NULL;
On NUMA systems, __GFP_THISNODE indicates that an allocation _must_ be on a particular node, and failure to allocate on the desired node will result in a failed allocation. Skip __GFP_THISNODE allocations if we are running on a NUMA system, since KFENCE can't guarantee which node its pool pages are allocated on. Reported-by: Vlastimil Babka <vbabka@suse.cz> Cc: Christoph Lameter <cl@linux.com> Fixes: 236e9f153852 ("kfence: skip all GFP_ZONEMASK allocations") Signed-off-by: Marco Elver <elver@google.com> --- mm/kfence/core.c | 2 ++ 1 file changed, 2 insertions(+)