hugetlbfs: take read_lock on i_mmap for PMD sharing
authorWaiman Long <longman@redhat.com>
Sun, 1 Dec 2019 01:56:49 +0000 (17:56 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sun, 1 Dec 2019 20:59:08 +0000 (12:59 -0800)
commit930668c34408ba983049322e04f13f03b6f1fafa
tree70b658341e9392125e05254ec17f8c956f2d17bd
parent1ab5b82f540b31852fbf4a3c975f3c16e0e76b9f
hugetlbfs: take read_lock on i_mmap for PMD sharing

A customer with large SMP systems (up to 16 sockets) with application
that uses large amount of static hugepages (~500-1500GB) are
experiencing random multisecond delays.  These delays were caused by the
long time it took to scan the VMA interval tree with mmap_sem held.

The sharing of huge PMD does not require changes to the i_mmap at all.
Therefore, we can just take the read lock and let other threads
searching for the right VMA share it in parallel.  Once the right VMA is
found, either the PMD lock (2M huge page for x86-64) or the
mm->page_table_lock will be acquired to perform the actual PMD sharing.

Lock contention, if present, will happen in the spinlock.  That is much
better than contention in the rwsem where the time needed to scan the
the interval tree is indeterminate.

With this patch applied, the customer is seeing significant performance
improvement over the unpatched kernel.

Link: http://lkml.kernel.org/r/20191107211809.9539-1-longman@redhat.com
Signed-off-by: Waiman Long <longman@redhat.com>
Suggested-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/hugetlb.c