From: Jens Axboe <axboe@fb.com> To: Bart Van Assche <Bart.VanAssche@sandisk.com>, "hch@lst.de" <hch@lst.de> Cc: "linux-block@vger.kernel.org" <linux-block@vger.kernel.org>, "linux-scsi@vger.kernel.org" <linux-scsi@vger.kernel.org>, "snitzer@redhat.com" <snitzer@redhat.com>, "linux-raid@vger.kernel.org" <linux-raid@vger.kernel.org>, "dm-devel@redhat.com" <dm-devel@redhat.com>, "j-nomura@ce.jp.nec.com" <j-nomura@ce.jp.nec.com> Subject: Re: split scsi passthrough fields out of struct request V2 Date: Thu, 26 Jan 2017 14:51:52 -0700 [thread overview] Message-ID: <e1d5f4e6-c63f-545d-710f-d4d10fa8d2e0@fb.com> (raw) In-Reply-To: <1485467235.2540.14.camel@sandisk.com> On 01/26/2017 02:47 PM, Bart Van Assche wrote: > (gdb) list *(blk_mq_sched_get_request+0x310) > 0xffffffff8132dcf0 is in blk_mq_sched_get_request (block/blk-mq-sched.c:136). > 131 rq->rq_flags |= RQF_QUEUED; > 132 } else > 133 rq = __blk_mq_alloc_request(data, op); > 134 } else { > 135 rq = __blk_mq_alloc_request(data, op); > 136 data->hctx->tags->rqs[rq->tag] = rq; > 137 } > 138 > 139 if (rq) { > 140 if (!op_is_flush(op)) { > > (gdb) disas blk_mq_sched_get_request > [ ... ] > 0xffffffff8132dce3 <+771>: callq 0xffffffff81324ab0 <__blk_mq_alloc_request> > 0xffffffff8132dce8 <+776>: mov %rax,%rcx > 0xffffffff8132dceb <+779>: mov 0x18(%r12),%rax > 0xffffffff8132dcf0 <+784>: movslq 0x5c(%rcx),%rdx > [ ... ] > (gdb) print &((struct request *)0)->tag > $1 = (int *) 0x5c <irq_stack_union+92> > > I think this means that rq == NULL and that a test for rq is missing after the > __blk_mq_alloc_request() call? That is exactly what it means, looks like that one path doesn't handle that. You'd have to exhaust the pool with atomic allocs for this to trigger, we don't do that at all in the normal IO path. So good catch, must be the dm part that enables this since it does NOWAIT allocations. diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c index 3136696f4991..c27613de80c5 100644 --- a/block/blk-mq-sched.c +++ b/block/blk-mq-sched.c @@ -134,7 +134,8 @@ struct request *blk_mq_sched_get_request(struct request_queue *q, rq = __blk_mq_alloc_request(data, op); } else { rq = __blk_mq_alloc_request(data, op); - data->hctx->tags->rqs[rq->tag] = rq; + if (rq) + data->hctx->tags->rqs[rq->tag] = rq; } if (rq) { -- Jens Axboe
WARNING: multiple messages have this Message-ID (diff)
From: Jens Axboe <axboe@fb.com> To: Bart Van Assche <Bart.VanAssche@sandisk.com>, "hch@lst.de" <hch@lst.de> Cc: "linux-scsi@vger.kernel.org" <linux-scsi@vger.kernel.org>, "linux-raid@vger.kernel.org" <linux-raid@vger.kernel.org>, "dm-devel@redhat.com" <dm-devel@redhat.com>, "linux-block@vger.kernel.org" <linux-block@vger.kernel.org>, "snitzer@redhat.com" <snitzer@redhat.com>, "j-nomura@ce.jp.nec.com" <j-nomura@ce.jp.nec.com> Subject: Re: [dm-devel] split scsi passthrough fields out of struct request V2 Date: Thu, 26 Jan 2017 14:51:52 -0700 [thread overview] Message-ID: <e1d5f4e6-c63f-545d-710f-d4d10fa8d2e0@fb.com> (raw) In-Reply-To: <1485467235.2540.14.camel@sandisk.com> On 01/26/2017 02:47 PM, Bart Van Assche wrote: > (gdb) list *(blk_mq_sched_get_request+0x310) > 0xffffffff8132dcf0 is in blk_mq_sched_get_request (block/blk-mq-sched.c:136). > 131 rq->rq_flags |= RQF_QUEUED; > 132 } else > 133 rq = __blk_mq_alloc_request(data, op); > 134 } else { > 135 rq = __blk_mq_alloc_request(data, op); > 136 data->hctx->tags->rqs[rq->tag] = rq; > 137 } > 138 > 139 if (rq) { > 140 if (!op_is_flush(op)) { > > (gdb) disas blk_mq_sched_get_request > [ ... ] > 0xffffffff8132dce3 <+771>: callq 0xffffffff81324ab0 <__blk_mq_alloc_request> > 0xffffffff8132dce8 <+776>: mov %rax,%rcx > 0xffffffff8132dceb <+779>: mov 0x18(%r12),%rax > 0xffffffff8132dcf0 <+784>: movslq 0x5c(%rcx),%rdx > [ ... ] > (gdb) print &((struct request *)0)->tag > $1 = (int *) 0x5c <irq_stack_union+92> > > I think this means that rq == NULL and that a test for rq is missing after the > __blk_mq_alloc_request() call? That is exactly what it means, looks like that one path doesn't handle that. You'd have to exhaust the pool with atomic allocs for this to trigger, we don't do that at all in the normal IO path. So good catch, must be the dm part that enables this since it does NOWAIT allocations. diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c index 3136696f4991..c27613de80c5 100644 --- a/block/blk-mq-sched.c +++ b/block/blk-mq-sched.c @@ -134,7 +134,8 @@ struct request *blk_mq_sched_get_request(struct request_queue *q, rq = __blk_mq_alloc_request(data, op); } else { rq = __blk_mq_alloc_request(data, op); - data->hctx->tags->rqs[rq->tag] = rq; + if (rq) + data->hctx->tags->rqs[rq->tag] = rq; } if (rq) { -- Jens Axboe
next prev parent reply other threads:[~2017-01-26 21:51 UTC|newest] Thread overview: 172+ messages / expand[flat|nested] mbox.gz Atom feed top 2017-01-25 17:25 split scsi passthrough fields out of struct request V2 Christoph Hellwig 2017-01-25 17:25 ` [PATCH 01/18] block: add a op_is_flush helper Christoph Hellwig 2017-01-26 2:58 ` Martin K. Petersen 2017-01-26 22:38 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 02/18] md: cleanup bio op / flags handling in raid1_write_request Christoph Hellwig 2017-01-26 2:59 ` Martin K. Petersen 2017-01-26 23:18 ` Bart Van Assche 2017-01-26 23:18 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 03/18] block: fix elevator init check Christoph Hellwig 2017-01-26 3:01 ` Martin K. Petersen 2017-01-26 23:21 ` Bart Van Assche 2017-01-26 23:21 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 04/18] block: simplify blk_init_allocated_queue Christoph Hellwig 2017-01-26 3:02 ` Martin K. Petersen 2017-01-26 23:27 ` Bart Van Assche 2017-01-26 23:27 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 05/18] block: allow specifying size for extra command data Christoph Hellwig 2017-01-26 3:15 ` Martin K. Petersen 2017-01-27 16:12 ` Christoph Hellwig 2017-01-27 16:12 ` Christoph Hellwig 2017-01-27 17:21 ` Bart Van Assche 2017-01-27 17:21 ` Bart Van Assche 2017-01-27 17:26 ` Jens Axboe 2017-01-27 17:26 ` Jens Axboe 2017-01-27 17:30 ` Bart Van Assche 2017-01-27 17:30 ` Bart Van Assche 2017-01-27 17:33 ` Jens Axboe 2017-01-25 17:25 ` [PATCH 06/18] dm: remove incomple BLOCK_PC support Christoph Hellwig 2017-01-27 17:32 ` Bart Van Assche 2017-01-27 17:32 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 07/18] dm: always defer request allocation to the owner of the request_queue Christoph Hellwig 2017-01-27 16:34 ` Mike Snitzer 2017-01-27 16:34 ` Mike Snitzer 2017-01-27 16:36 ` Christoph Hellwig 2017-01-27 16:36 ` Christoph Hellwig 2017-01-27 16:44 ` Mike Snitzer 2017-01-27 16:44 ` Mike Snitzer 2017-01-25 17:25 ` [PATCH 08/18] scsi_dh_rdac: switch to scsi_execute_req_flags() Christoph Hellwig 2017-01-26 3:18 ` Martin K. Petersen 2017-01-25 17:25 ` [PATCH 09/18] scsi_dh_emc: " Christoph Hellwig 2017-01-26 3:19 ` Martin K. Petersen 2017-01-25 17:25 ` [PATCH 10/18] scsi_dh_hp_sw: " Christoph Hellwig 2017-01-26 3:20 ` Martin K. Petersen 2017-01-26 3:20 ` Martin K. Petersen 2017-01-25 17:25 ` [PATCH 11/18] scsi: remove gfp_flags member in scsi_host_cmd_pool Christoph Hellwig 2017-01-26 3:21 ` Martin K. Petersen 2017-01-27 17:38 ` Bart Van Assche 2017-01-27 17:38 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 12/18] scsi: respect unchecked_isa_dma for blk-mq Christoph Hellwig 2017-01-26 3:23 ` Martin K. Petersen 2017-01-27 17:45 ` Bart Van Assche 2017-01-27 17:45 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 13/18] scsi: remove scsi_cmd_dma_pool Christoph Hellwig 2017-01-26 3:24 ` Martin K. Petersen 2017-01-27 17:51 ` Bart Van Assche 2017-01-27 17:51 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 14/18] scsi: remove __scsi_alloc_queue Christoph Hellwig 2017-01-26 3:25 ` Martin K. Petersen 2017-01-27 17:58 ` Bart Van Assche 2017-01-27 17:58 ` Bart Van Assche 2017-01-28 8:23 ` hch 2017-01-25 17:25 ` [PATCH 15/18] scsi: allocate scsi_cmnd structures as part of struct request Christoph Hellwig 2017-01-26 3:30 ` Martin K. Petersen 2017-01-27 18:39 ` Bart Van Assche 2017-01-27 18:39 ` Bart Van Assche 2017-01-28 8:25 ` hch 2017-01-28 8:25 ` hch 2017-01-25 17:25 ` [PATCH 16/18] block/bsg: move queue creation into bsg_setup_queue Christoph Hellwig 2017-01-27 18:48 ` Bart Van Assche 2017-01-27 18:48 ` Bart Van Assche 2017-01-25 17:25 ` [PATCH 17/18] block: split scsi_request out of struct request Christoph Hellwig 2017-01-25 17:25 ` [PATCH 18/18] block: don't assign cmd_flags in __blk_rq_prep_clone Christoph Hellwig 2017-01-26 3:31 ` Martin K. Petersen 2017-01-26 18:29 ` split scsi passthrough fields out of struct request V2 Bart Van Assche 2017-01-26 18:29 ` Bart Van Assche 2017-01-26 18:44 ` Jens Axboe 2017-01-26 18:52 ` Bart Van Assche 2017-01-26 18:52 ` Bart Van Assche 2017-01-26 18:57 ` Jens Axboe 2017-01-26 18:59 ` hch 2017-01-26 19:01 ` Jens Axboe 2017-01-26 19:01 ` Jens Axboe 2017-01-26 20:47 ` [dm-devel] " Bart Van Assche 2017-01-26 20:54 ` Jens Axboe 2017-01-26 20:54 ` [dm-devel] " Jens Axboe 2017-01-26 21:01 ` Bart Van Assche 2017-01-26 21:01 ` Bart Van Assche 2017-01-26 21:12 ` Jens Axboe 2017-01-26 21:12 ` [dm-devel] " Jens Axboe 2017-01-26 21:47 ` Bart Van Assche 2017-01-26 21:47 ` [dm-devel] " Bart Van Assche 2017-01-26 21:51 ` Jens Axboe [this message] 2017-01-26 21:51 ` Jens Axboe 2017-01-26 23:14 ` Bart Van Assche 2017-01-26 23:14 ` Bart Van Assche 2017-01-26 23:26 ` Jens Axboe 2017-01-26 23:26 ` Jens Axboe 2017-01-26 23:26 ` [dm-devel] " Jens Axboe 2017-01-26 23:47 ` Bart Van Assche 2017-01-26 23:47 ` [dm-devel] " Bart Van Assche 2017-01-26 23:50 ` Jens Axboe 2017-01-27 0:33 ` Jens Axboe 2017-01-27 0:33 ` [dm-devel] " Jens Axboe 2017-01-27 0:38 ` Bart Van Assche 2017-01-27 0:38 ` Bart Van Assche 2017-01-27 0:41 ` Jens Axboe 2017-01-27 1:15 ` Bart Van Assche 2017-01-27 1:15 ` [dm-devel] " Bart Van Assche 2017-01-27 1:22 ` Jens Axboe 2017-01-27 1:22 ` [dm-devel] " Jens Axboe 2017-01-27 6:40 ` Jens Axboe 2017-01-27 8:04 ` Jens Axboe 2017-01-27 8:04 ` [dm-devel] " Jens Axboe 2017-01-27 16:52 ` Bart Van Assche 2017-01-27 16:52 ` [dm-devel] " Bart Van Assche 2017-01-27 16:56 ` Jens Axboe 2017-01-27 16:56 ` [dm-devel] " Jens Axboe 2017-01-27 17:03 ` Bart Van Assche 2017-01-27 17:03 ` [dm-devel] " Bart Van Assche 2017-01-31 1:12 ` Bart Van Assche 2017-01-31 1:12 ` Bart Van Assche 2017-01-31 1:38 ` Jens Axboe 2017-01-31 1:38 ` Jens Axboe 2017-01-31 4:13 ` Jens Axboe 2017-01-31 4:13 ` [dm-devel] " Jens Axboe 2017-01-31 21:35 ` Bart Van Assche 2017-01-31 21:35 ` Bart Van Assche 2017-01-31 21:55 ` Bart Van Assche 2017-01-31 21:55 ` Bart Van Assche 2017-01-31 21:58 ` Jens Axboe 2017-02-01 1:01 ` Bart Van Assche 2017-02-01 6:38 ` Jens Axboe 2017-02-01 16:46 ` Bart Van Assche 2017-02-01 17:13 ` Jens Axboe 2017-02-01 17:28 ` Bart Van Assche 2017-02-01 19:21 ` Bart Van Assche 2017-02-01 22:01 ` Bart Van Assche 2017-02-02 17:27 ` Bart Van Assche 2017-02-02 18:33 ` Mike Snitzer 2017-02-02 18:43 ` Bart Van Assche 2017-02-02 19:13 ` Mike Snitzer 2017-02-02 19:46 ` Bart Van Assche 2017-02-02 21:04 ` Mike Snitzer 2017-02-02 21:10 ` Mike Snitzer 2017-02-03 0:20 ` Bart Van Assche 2017-02-03 0:42 ` Mike Snitzer 2017-02-02 22:38 ` Bart Van Assche 2017-01-27 17:02 ` [dm-devel] " Bart Van Assche 2017-01-27 17:02 ` Bart Van Assche 2017-01-27 16:11 ` Jens Axboe 2017-01-27 16:11 ` Jens Axboe 2017-01-27 16:17 ` Christoph Hellwig 2017-01-27 16:17 ` Christoph Hellwig 2017-01-27 16:21 ` Jens Axboe 2017-01-27 16:21 ` Jens Axboe 2017-01-27 16:23 ` Christoph Hellwig 2017-01-27 16:23 ` Christoph Hellwig 2017-01-27 16:27 ` Jens Axboe 2017-01-27 16:27 ` Jens Axboe 2017-01-27 16:34 ` Christoph Hellwig 2017-01-27 16:34 ` Christoph Hellwig 2017-01-27 16:38 ` Jens Axboe 2017-01-27 16:38 ` Jens Axboe 2017-01-27 16:42 ` Christoph Hellwig 2017-01-27 16:42 ` Christoph Hellwig 2017-01-27 16:58 ` Jens Axboe 2017-01-27 16:58 ` Jens Axboe 2017-01-27 21:27 ` Bart Van Assche 2017-01-27 21:27 ` Bart Van Assche 2017-01-28 8:29 ` hch 2017-01-30 6:58 ` Hannes Reinecke 2017-01-30 6:58 ` Hannes Reinecke
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=e1d5f4e6-c63f-545d-710f-d4d10fa8d2e0@fb.com \ --to=axboe@fb.com \ --cc=Bart.VanAssche@sandisk.com \ --cc=dm-devel@redhat.com \ --cc=hch@lst.de \ --cc=j-nomura@ce.jp.nec.com \ --cc=linux-block@vger.kernel.org \ --cc=linux-raid@vger.kernel.org \ --cc=linux-scsi@vger.kernel.org \ --cc=snitzer@redhat.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.