blk-mq: fix schedule-while-atomic with scheduler attached

We must have dropped the ctx before we call
blk_mq_sched_insert_request() with can_block=true, otherwise we risk
that a flush request can block on insertion if we are currently out of
tags.

[   47.667190] BUG: scheduling while atomic: jbd2/sda2-8/2089/0x00000002
[   47.674493] Modules linked in: x86_pkg_temp_thermal btrfs xor zlib_deflate raid6_pq sr_mod cdre
[   47.690572] Preemption disabled at:
[   47.690584] [<ffffffff81326c7c>] blk_mq_sched_get_request+0x6c/0x280
[   47.701764] CPU: 1 PID: 2089 Comm: jbd2/sda2-8 Not tainted 4.11.0-rc7+ #271
[   47.709630] Hardware name: Dell Inc. PowerEdge T630/0NT78X, BIOS 2.3.4 11/09/2016
[   47.718081] Call Trace:
[   47.720903]  dump_stack+0x4f/0x73
[   47.724694]  ? blk_mq_sched_get_request+0x6c/0x280
[   47.730137]  __schedule_bug+0x6c/0xc0
[   47.734314]  __schedule+0x559/0x780
[   47.738302]  schedule+0x3b/0x90
[   47.741899]  io_schedule+0x11/0x40
[   47.745788]  blk_mq_get_tag+0x167/0x2a0
[   47.750162]  ? remove_wait_queue+0x70/0x70
[   47.754901]  blk_mq_get_driver_tag+0x92/0xf0
[   47.759758]  blk_mq_sched_insert_request+0x134/0x170
[   47.765398]  ? blk_account_io_start+0xd0/0x270
[   47.770679]  blk_mq_make_request+0x1b2/0x850
[   47.775766]  generic_make_request+0xf7/0x2d0
[   47.780860]  submit_bio+0x5f/0x120
[   47.784979]  ? submit_bio+0x5f/0x120
[   47.789631]  submit_bh_wbc.isra.46+0x10d/0x130
[   47.794902]  submit_bh+0xb/0x10
[   47.798719]  journal_submit_commit_record+0x190/0x210
[   47.804686]  ? _raw_spin_unlock+0x13/0x30
[   47.809480]  jbd2_journal_commit_transaction+0x180a/0x1d00
[   47.815925]  kjournald2+0xb6/0x250
[   47.820022]  ? kjournald2+0xb6/0x250
[   47.824328]  ? remove_wait_queue+0x70/0x70
[   47.829223]  kthread+0x10e/0x140
[   47.833147]  ? commit_timeout+0x10/0x10
[   47.837742]  ? kthread_create_on_node+0x40/0x40
[   47.843122]  ret_from_fork+0x29/0x40

Fixes: a4d907b6a33b ("blk-mq: streamline blk_mq_make_request")
Reviewed-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Jens Axboe <axboe@fb.com>

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 9d7645f24b05c7ddd56513a91267afc97e149ed0..47b81063872929f8e549295807fb08ce5e83edb7 100644 (file)
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1571,6 +1571,7 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
 
        plug = current->plug;
        if (unlikely(is_flush_fua)) {
+               blk_mq_put_ctx(data.ctx);
                blk_mq_bio_to_request(rq, bio);
                if (q->elevator) {
                        blk_mq_sched_insert_request(rq, false, true, true,
@@ -1582,6 +1583,7 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
        } else if (plug && q->nr_hw_queues == 1) {
                struct request *last = NULL;
 
+               blk_mq_put_ctx(data.ctx);
                blk_mq_bio_to_request(rq, bio);
 
                /*
@@ -1626,20 +1628,19 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
                if (same_queue_rq)
                        blk_mq_try_issue_directly(data.hctx, same_queue_rq,
                                        &cookie);
-
-               return cookie;
        } else if (q->nr_hw_queues > 1 && is_sync) {
                blk_mq_put_ctx(data.ctx);
                blk_mq_bio_to_request(rq, bio);
                blk_mq_try_issue_directly(data.hctx, rq, &cookie);
-               return cookie;
        } else if (q->elevator) {
+               blk_mq_put_ctx(data.ctx);
                blk_mq_bio_to_request(rq, bio);
                blk_mq_sched_insert_request(rq, false, true, true, true);
-       } else if (!blk_mq_merge_queue_io(data.hctx, data.ctx, rq, bio))
+       } else if (!blk_mq_merge_queue_io(data.hctx, data.ctx, rq, bio)) {
+               blk_mq_put_ctx(data.ctx);
                blk_mq_run_hw_queue(data.hctx, true);
+       }
 
-       blk_mq_put_ctx(data.ctx);
        return cookie;
 }