This patch is to fix a oops from a torn down device. When scsi_run_queue process starved queues, scsi_request_fn can race with scsi_remove_device. In this case, rarely, scsi_request_fn release the last reference and set sdev->request_queue to NULL. It result in NULL-pointer dereference when spin_unlock is tried with (NULL)-> queue_lock. We need to add an extra reference to the device on both sides of the __blk_run_queue to hold reference until scsi_request_fn is finished. [ 8.042972] Unable to handle kernel NULL pointer dereference at virtual address 00000240 [ 8.051061] pgd = 80004000 [ 8.053762] [00000240] *pgd=00000000 [ 8.057342] Internal error: Oops: 17 [#1] PREEMPT SMP ARM [ 8.062736] Modules linked in: [ 8.065793] CPU: 0 Not tainted (3.4.2+ #313) [ 8.070418] PC is at scsi_run_queue+0x19c/0x2b8 [ 8.074947] LR is at scsi_run_queue+0x198/0x2b8 [ 8.079476] pc : [<802569a0>] lr : [<8025699c>] psr: 20000193 [ 8.079481] sp : 9f915f10 ip : 00000f01 fp : 00000001 [ 8.090953] r10: 9f09d424 r9 : 8055294c r8 : 9f915f20 [ 8.096172] r7 : 9f914000 r6 : 00000000 r5 : 9f09e000 r4 : 9f09d400 [ 8.102694] r3 : 00000000 r2 : 80504a14 r1 : 60000193 r0 : 00000043 [ 8.109219] Flags: nzCv IRQs off FIQs on Mode SVC_32 ISA ARM Segment kernel [ 8.116612] Control: 10c53c7d Table: 1f1f804a DAC: 00000015 [ 8.122355] Process kworker/0:1 (pid: 312, stack limit = 0x9f9142f0) [ 8.128705] Stack: (0x9f915f10 to 0x9f916000) [ 8.133059] 5f00: 00000000 00000000 9f1e8000 60000113 [ 8.141236] 5f20: 9f915f20 9f915f20 00000001 9f9be1c0 8095f480 80963100 9f914000 00000000 [ 8.149414] 5f40: 80963105 80258304 9f09da70 80034858 9f9fb6c0 00000000 00000001 9f9be1c0 [ 8.157591] 5f60: 8095f480 8095f488 9f9be1d0 9f914000 804edf80 804fff08 00000009 80037044 [ 8.165767] 5f80: 804edfc0 804edfc0 804edfc0 804edf80 804edf80 804edf80 00000001 9f859ee8 [ 8.173943] 5fa0: 9f915fcc 9f9be1c0 80036ed4 00000000 00000000 00000000 00000000 8003ac64 [ 8.182119] 5fc0: 9f859ee8 00000000 9f9be1c0 00000000 00000000 00000000 9f915fd8 9f915fd8 [ 8.190296] 5fe0: 00000000 9f859ee8 8003abd8 80014c0c 00000013 80014c0c ffffffff ffffffff [ 8.198494] [<802569a0>] (scsi_run_queue+0x19c/0x2b8) from [<80034858>] (process_one_work+0x118/0x39c) [ 8.207812] [<80034858>] (process_one_work+0x118/0x39c) from [<80037044>] (worker_thread+0x170/0x368) [ 8.217047] [<80037044>] (worker_thread+0x170/0x368) from [<8003ac64>] (kthread+0x8c/0x98) [ 8.225337] [<8003ac64>] (kthread+0x8c/0x98) from [<80014c0c>] (kernel_thread_exit+0x0/0x8) [ 8.233692] Code: e59f011c e58dc000 eb05457e e5953004 (e5930240) [ 8.240026] __scsi_remove_device:962 9f1e8000 [ 8.240140] ---[ end trace 1ec4a0217c9f24f3 ]--- Signed-off-by: Chanho Min <chanho.min@xxxxxxx> --- drivers/scsi/scsi_lib.c | 4 ++++ 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c index b583277..1868c35 100644 --- a/drivers/scsi/scsi_lib.c +++ b/drivers/scsi/scsi_lib.c @@ -436,9 +436,13 @@ static void scsi_run_queue(struct request_queue *q) } spin_unlock(shost->host_lock); + /* hold a reference on the device so it doesn't release device */ + get_device(&sdev->sdev_gendev); spin_lock(sdev->request_queue->queue_lock); __blk_run_queue(sdev->request_queue); spin_unlock(sdev->request_queue->queue_lock); + /* ok to remove device now */ + put_device(&sdev->sdev_gendev); spin_lock(shost->host_lock); } /* put any unprocessed entries back */ -- 1.7.0.4 -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
