sched/numa: use down_read_trylock for mmap_sem
PATCH 1/1 sched/numa: use down_read_trylock for mmap_sem
A customer has reported a soft-lockup when running a proprietary intensive memory stress test, where the trace on multiple CPU's looks like this:
RIP: 0010:[<ffffffff810c53fe>] [<ffffffff810c53fe>] native_queued_spin_lock_slowpath+0x10e/0x190 ... Call Trace: [<ffffffff81182d07>] queued_spin_lock_slowpath+0x7/0xa [<ffffffff811bc331>] change_protection_range+0x3b1/0x930 [<ffffffff811d4be8>] change_prot_numa+0x18/0x30 [<ffffffff810adefe>] task_numa_work+0x1fe/0x310 [<ffffffff81098322>] task_work_run+0x72/0x90
Further investigation showed that the lock contention here is pmd_lock().
The task_numa_work() function makes sure that only one thread is let to perform the work in a single scan period (via cmpxchg), but if there's a thread with mmap_sem locked for writing for several periods, multiple threads in task_numa_work() can build up a convoy waiting for mmap_sem for read and then all get unblocked at once.
This patch changes the down_read() to the trylock version, which prevents the build up. For a workload experiencing mmap_sem contention, it's probably better to postpone the NUMA balancing work anyway. This seems to have fixed the soft lockups involving pmd_lock(), which is in line with the convoy theory.
--- kernel/sched/fair.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
kernel/sched/fair.c +2 -1
@@ -2475,7 +2475,8 @@ void task_numa_work(struct callback_head *work)return ; -down_read (&mm->mmap_sem); +if (!down_read_trylock (&mm->mmap_sem)) +return ; vma =find_vma (mm, start);if (!vma) {reset_ptenuma_scan (p);
Signature · 1 line
2.12.2
Quoted 36 lines · Vlastimil Babka
A customer has reported a soft-lockup when running a proprietary intensive
memory stress test, where the trace on multiple CPU's looks like this:RIP: 0010:[<ffffffff810c53fe>] [<ffffffff810c53fe>] native_queued_spin_lock_slowpath+0x10e/0x190 ... Call Trace: [<ffffffff81182d07>] queued_spin_lock_slowpath+0x7/0xa [<ffffffff811bc331>] change_protection_range+0x3b1/0x930 [<ffffffff811d4be8>] change_prot_numa+0x18/0x30 [<ffffffff810adefe>] task_numa_work+0x1fe/0x310 [<ffffffff81098322>] task_work_run+0x72/0x90Further investigation showed that the lock contention here is
pmd_lock().The
task_numa_work()function makes sure that only one thread is let to perform
the work in a single scan period (via cmpxchg), but if there's a thread withmmap_semlocked for writing for several periods, multiple threads intask_numa_work()can build up a convoy waiting formmap_semfor read and then
all get unblocked at once.This patch changes the
down_read()to the trylock version, which prevents the
build up. For a workload experiencingmmap_semcontention, it's probably better
to postpone the NUMA balancing work anyway. This seems to have fixed the soft
lockups involvingpmd_lock(), which is in line with the convoy theory.
Quoted 1 line
Signed-off-by Vlastimil Babka <[email protected]>
Quoted 21 lines · Vlastimil Babka
A customer has reported a soft-lockup when running a proprietary intensive memory stress test, where the trace on multiple CPU's looks like this:
RIP: 0010:[<ffffffff810c53fe>] [<ffffffff810c53fe>] native_queued_spin_lock_slowpath+0x10e/0x190 ... Call Trace: [<ffffffff81182d07>] queued_spin_lock_slowpath+0x7/0xa [<ffffffff811bc331>] change_protection_range+0x3b1/0x930 [<ffffffff811d4be8>] change_prot_numa+0x18/0x30 [<ffffffff810adefe>] task_numa_work+0x1fe/0x310 [<ffffffff81098322>] task_work_run+0x72/0x90Further investigation showed that the lock contention here is
pmd_lock().The
task_numa_work()function makes sure that only one thread is let to perform the work in a single scan period (via cmpxchg), but if there's a thread withmmap_semlocked for writing for several periods, multiple threads intask_numa_work()can build up a convoy waiting formmap_semfor read and then all get unblocked at once.
Quoted 6 lines
This patch changes the
down_read()to the trylock version, which prevents the build up. For a workload experiencingmmap_semcontention, it's probably better to postpone the NUMA balancing work anyway. This seems to have fixed the soft lockups involvingpmd_lock(), which is in line with the convoy theory.Signed-off-by Vlastimil Babka <[email protected]>
This skips the entire scan window and defers to the next once. Potentially, with constant contention, it'll never make progress and there could be other disruption. However, I cannot see any way how that's worse than waiting on mmap_sem so
Signature · 2 lines
Mel Gorman SUSE Labs
Peter Zijlstra
maintainerQuoted 21 lines · Vlastimil Babka
A customer has reported a soft-lockup when running a proprietary intensive memory stress test, where the trace on multiple CPU's looks like this:
RIP: 0010:[<ffffffff810c53fe>] [<ffffffff810c53fe>] native_queued_spin_lock_slowpath+0x10e/0x190 ... Call Trace: [<ffffffff81182d07>] queued_spin_lock_slowpath+0x7/0xa [<ffffffff811bc331>] change_protection_range+0x3b1/0x930 [<ffffffff811d4be8>] change_prot_numa+0x18/0x30 [<ffffffff810adefe>] task_numa_work+0x1fe/0x310 [<ffffffff81098322>] task_work_run+0x72/0x90Further investigation showed that the lock contention here is
pmd_lock().The
task_numa_work()function makes sure that only one thread is let to perform the work in a single scan period (via cmpxchg), but if there's a thread withmmap_semlocked for writing for several periods, multiple threads intask_numa_work()can build up a convoy waiting formmap_semfor read and then all get unblocked at once.
Quoted 6 lines
This patch changes the
down_read()to the trylock version, which prevents the build up. For a workload experiencingmmap_semcontention, it's probably better to postpone the NUMA balancing work anyway. This seems to have fixed the soft lockups involvingpmd_lock(), which is in line with the convoy theory.Signed-off-by Vlastimil Babka <[email protected]>
Thanks!
tip-bot for Vlastimil Babka
merge bot[tip:sched/core] sched/numa: Use down_read_trylock() for the mmap_sem
In reply to Vlastimil Babkasched/numa: Use down_read_trylock() for the mmap_sem
Commit-ID: 8655d5497735b288f8a9b458bd22e7d1bf95bb61 Gitweb: http://git.kernel.org/tip/8655d5497735b288f8a9b458bd22e7d1bf95bb61 Author: Vlastimil Babka <[email protected]> AuthorDate: Mon, 15 May 2017 15:13:16 +0200 Committer: Ingo Molnar <[email protected]> CommitDate: Tue, 23 May 2017 10:01:34 +0200
sched/numa: Use down_read_trylock() for the mmap_sem
A customer has reported a soft-lockup when running an intensive memory stress test, where the trace on multiple CPU's looks like this:
RIP: 0010:[<ffffffff810c53fe>] [<ffffffff810c53fe>] native_queued_spin_lock_slowpath+0x10e/0x190 ... Call Trace: [<ffffffff81182d07>] queued_spin_lock_slowpath+0x7/0xa [<ffffffff811bc331>] change_protection_range+0x3b1/0x930 [<ffffffff811d4be8>] change_prot_numa+0x18/0x30 [<ffffffff810adefe>] task_numa_work+0x1fe/0x310 [<ffffffff81098322>] task_work_run+0x72/0x90
Further investigation showed that the lock contention here is pmd_lock().
The task_numa_work() function makes sure that only one thread is let to perform the work in a single scan period (via cmpxchg), but if there's a thread with mmap_sem locked for writing for several periods, multiple threads in task_numa_work() can build up a convoy waiting for mmap_sem for read and then all get unblocked at once.
This patch changes the down_read() to the trylock version, which prevents the build up. For a workload experiencing mmap_sem contention, it's probably better to postpone the NUMA balancing work anyway. This seems to have fixed the soft lockups involving pmd_lock(), which is in line with the convoy theory.
--- kernel/sched/fair.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
kernel/sched/fair.c +2 -1
@@ -2470,7 +2470,8 @@ void task_numa_work(struct callback_head *work)return ; -down_read (&mm->mmap_sem); +if (!down_read_trylock (&mm->mmap_sem)) +return ; vma =find_vma (mm, start);if (!vma) {reset_ptenuma_scan (p);