1 From 5b1058c9eafcafe88a8bd4c664fc904a44d21aca Mon Sep 17 00:00:00 2001
2 From: Mike Marciniszyn <mike.marciniszyn@intel.com>
3 Date: Mon, 24 Jun 2019 12:14:29 -0400
4 Subject: IB/hfi1: Avoid hardlockup with flushlist_lock
6 commit cf131a81967583ae737df6383a0893b9fee75b4e upstream.
8 Heavy contention of the sde flushlist_lock can cause hard lockups at
9 extreme scale when the flushing logic is under stress.
11 Mitigate by replacing the item at a time copy to the local list with
12 an O(1) list_splice_init() and using the high priority work queue to
15 Ported to linux-4.14.y.
17 Fixes: 7724105686e7 ("IB/hfi1: add driver files")
18 Cc: <stable@vger.kernel.org>
19 Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
20 Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
21 Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
22 Signed-off-by: Doug Ledford <dledford@redhat.com>
23 Signed-off-by: Sasha Levin <sashal@kernel.org>
25 drivers/infiniband/hw/hfi1/sdma.c | 9 +++------
26 1 file changed, 3 insertions(+), 6 deletions(-)
28 diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c
29 index 88e326d6cc49..d648a4167832 100644
30 --- a/drivers/infiniband/hw/hfi1/sdma.c
31 +++ b/drivers/infiniband/hw/hfi1/sdma.c
32 @@ -410,10 +410,7 @@ static void sdma_flush(struct sdma_engine *sde)
33 sdma_flush_descq(sde);
34 spin_lock_irqsave(&sde->flushlist_lock, flags);
36 - list_for_each_entry_safe(txp, txp_next, &sde->flushlist, list) {
37 - list_del_init(&txp->list);
38 - list_add_tail(&txp->list, &flushlist);
40 + list_splice_init(&sde->flushlist, &flushlist);
41 spin_unlock_irqrestore(&sde->flushlist_lock, flags);
42 /* flush from flush list */
43 list_for_each_entry_safe(txp, txp_next, &flushlist, list)
44 @@ -2426,7 +2423,7 @@ int sdma_send_txreq(struct sdma_engine *sde,
46 wait->count += tx->num_desc;
48 - schedule_work(&sde->flush_worker);
49 + queue_work_on(sde->cpu, system_highpri_wq, &sde->flush_worker);
53 @@ -2526,7 +2523,7 @@ int sdma_send_txlist(struct sdma_engine *sde, struct iowait *wait,
56 spin_unlock(&sde->flushlist_lock);
57 - schedule_work(&sde->flush_worker);
58 + queue_work_on(sde->cpu, system_highpri_wq, &sde->flush_worker);