On Mon 26-06-17 14:00:27, Vlastimil Babka wrote: > On 06/23/2017 10:53 AM, Michal Hocko wrote: > > From: Michal Hocko <[email protected]> > > > > Now that __GFP_RETRY_MAYFAIL has a reasonable semantic regardless of the > > request size we can drop the hackish implementation for !costly orders. > > __GFP_RETRY_MAYFAIL retries as long as the reclaim makes a forward > > progress and backs of when we are out of memory for the requested size. > > Therefore we do not need to enforce__GFP_NORETRY for !costly orders just > > to silent the oom killer anymore. > > > > Signed-off-by: Michal Hocko <[email protected]> > > The flag is now supported, but not for the embedded page table > allocations, so OOM is still theoretically possible, right? > That should be rare, though. Worth mentioning anywhere?
Yes that is true. Not sure I would make it more complicated than necessary. I can add a note in there if you insist but to me it sounds like something that will only confuse people. > Other than that. > Acked-by: Vlastimil Babka <[email protected]> Thanks! > > --- > > mm/util.c | 14 ++++---------- > > 1 file changed, 4 insertions(+), 10 deletions(-) > > > > diff --git a/mm/util.c b/mm/util.c > > index 6520f2d4a226..ee250e2cde34 100644 > > --- a/mm/util.c > > +++ b/mm/util.c > > @@ -339,9 +339,9 @@ EXPORT_SYMBOL(vm_mmap); > > * Uses kmalloc to get the memory but if the allocation fails then falls > > back > > * to the vmalloc allocator. Use kvfree for freeing the memory. > > * > > - * Reclaim modifiers - __GFP_NORETRY and __GFP_NOFAIL are not supported. > > __GFP_RETRY_MAYFAIL > > - * is supported only for large (>32kB) allocations, and it should be used > > only if > > - * kmalloc is preferable to the vmalloc fallback, due to visible > > performance drawbacks. > > + * Reclaim modifiers - __GFP_NORETRY and __GFP_NOFAIL are not supported. > > + * __GFP_RETRY_MAYFAIL is supported, and it should be used only if kmalloc > > is > > + * preferable to the vmalloc fallback, due to visible performance > > drawbacks. > > * > > * Any use of gfp flags outside of GFP_KERNEL should be consulted with mm > > people. > > */ > > @@ -366,13 +366,7 @@ void *kvmalloc_node(size_t size, gfp_t flags, int node) > > if (size > PAGE_SIZE) { > > kmalloc_flags |= __GFP_NOWARN; > > > > - /* > > - * We have to override __GFP_RETRY_MAYFAIL by __GFP_NORETRY for > > !costly > > - * requests because there is no other way to tell the allocator > > - * that we want to fail rather than retry endlessly. > > - */ > > - if (!(kmalloc_flags & __GFP_RETRY_MAYFAIL) || > > - (size <= PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER)) > > + if (!(kmalloc_flags & __GFP_RETRY_MAYFAIL)) > > kmalloc_flags |= __GFP_NORETRY; > > } > > > > -- Michal Hocko SUSE Labs
