From e7ac6c6666bec0a354758a1298d3231e4a635362 Mon Sep 17 00:00:00 2001 From: "Seymour, Shane M" Date: Thu, 2 Jul 2015 12:01:10 +0000 Subject: st: null pointer dereference panic caused by use after kref_put by st_open MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Two SLES11 SP3 servers encountered similar crashes simultaneously following some kind of SAN/tape target issue: ... qla2xxx [0000:81:00.0]-801c:3: Abort command issued nexus=3:0:2 -- 1 2002. qla2xxx [0000:81:00.0]-801c:3: Abort command issued nexus=3:0:2 -- 1 2002. qla2xxx [0000:81:00.0]-8009:3: DEVICE RESET ISSUED nexus=3:0:2 cmd=ffff882f89c2c7c0. qla2xxx [0000:81:00.0]-800c:3: do_reset failed for cmd=ffff882f89c2c7c0. qla2xxx [0000:81:00.0]-800f:3: DEVICE RESET FAILED: Task management failed nexus=3:0:2 cmd=ffff882f89c2c7c0. qla2xxx [0000:81:00.0]-8009:3: TARGET RESET ISSUED nexus=3:0:2 cmd=ffff882f89c2c7c0. qla2xxx [0000:81:00.0]-800c:3: do_reset failed for cmd=ffff882f89c2c7c0. qla2xxx [0000:81:00.0]-800f:3: TARGET RESET FAILED: Task management failed nexus=3:0:2 cmd=ffff882f89c2c7c0. qla2xxx [0000:81:00.0]-8012:3: BUS RESET ISSUED nexus=3:0:2. qla2xxx [0000:81:00.0]-802b:3: BUS RESET SUCCEEDED nexus=3:0:2. qla2xxx [0000:81:00.0]-505f:3: Link is operational (8 Gbps). qla2xxx [0000:81:00.0]-8018:3: ADAPTER RESET ISSUED nexus=3:0:2. qla2xxx [0000:81:00.0]-00af:3: Performing ISP error recovery - ha=ffff88bf04d18000. rport-3:0-0: blocked FC remote port time out: removing target and saving binding qla2xxx [0000:81:00.0]-505f:3: Link is operational (8 Gbps). qla2xxx [0000:81:00.0]-8017:3: ADAPTER RESET SUCCEEDED nexus=3:0:2. rport-2:0-0: blocked FC remote port time out: removing target and saving binding sg_rq_end_io: device detached BUG: unable to handle kernel NULL pointer dereference at 00000000000002a8 IP: [] __pm_runtime_idle+0x28/0x90 PGD 7e6586f067 PUD 7e5af06067 PMD 0 [1739975.390354] Oops: 0002 [#1] SMP CPU 0 ... Supported: No, Proprietary modules are loaded [1739975.390463] Pid: 27965, comm: ABCD Tainted: PF X 3.0.101-0.29-default #1 HP ProLiant DL580 Gen8 RIP: 0010:[] [] __pm_runtime_idle+0x28/0x90 RSP: 0018:ffff8839dc1e7c68 EFLAGS: 00010202 RAX: 0000000000000000 RBX: ffff883f0592fc00 RCX: 0000000000000090 RDX: 0000000000000000 RSI: 0000000000000004 RDI: 0000000000000138 RBP: 0000000000000138 R08: 0000000000000010 R09: ffffffff81bd39d0 R10: 00000000000009c0 R11: ffffffff81025790 R12: 0000000000000001 R13: ffff883022212b80 R14: 0000000000000004 R15: ffff883022212b80 FS: 00007f8e54560720(0000) GS:ffff88407f800000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 00000000000002a8 CR3: 0000007e6ced6000 CR4: 00000000001407f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process ABCD (pid: 27965, threadinfo ffff8839dc1e6000, task ffff883592e0c640) Stack: ffff883f0592fc00 00000000fffffffa 0000000000000001 ffff883022212b80 ffff883eff772400 ffffffffa03fa309 0000000000000000 0000000000000000 ffffffffa04003a0 ffff883f063196c0 ffff887f0379a930 ffffffff8115ea1e Call Trace: [] st_open+0x129/0x240 [st] [] chrdev_open+0x13e/0x200 [] __dentry_open+0x198/0x310 [] do_last+0x1f4/0x800 [] path_openat+0xd9/0x420 [] do_filp_open+0x4c/0xc0 [] do_sys_open+0x17f/0x250 [] system_call_fastpath+0x16/0x1b [<00007f8e4f617fd0>] 0x7f8e4f617fcf Code: eb d3 90 48 83 ec 28 40 f6 c6 04 48 89 6c 24 08 4c 89 74 24 20 48 89 fd 48 89 1c 24 4c 89 64 24 10 41 89 f6 4c 89 6c 24 18 74 11 ff 8f 70 01 00 00 0f 94 c0 45 31 ed 84 c0 74 2b 4c 8d a5 a0 RIP [] __pm_runtime_idle+0x28/0x90 RSP CR2: 00000000000002a8 Analysis reveals the cause of the crash to be due to STp->device being NULL. The pointer was NULLed via scsi_tape_put(STp) when it calls scsi_tape_release(). In st_open() we jump to err_out after scsi_block_when_processing_errors() completes and returns the device as offline (sdev_state was SDEV_DEL): 1180 /* Open the device. Needs to take the BKL only because of incrementing the SCSI host 1181 module count. */ 1182 static int st_open(struct inode *inode, struct file *filp) 1183 { 1184 int i, retval = (-EIO); 1185 int resumed = 0; 1186 struct scsi_tape *STp; 1187 struct st_partstat *STps; 1188 int dev = TAPE_NR(inode); 1189 char *name; ... 1217 if (scsi_autopm_get_device(STp->device) < 0) { 1218 retval = -EIO; 1219 goto err_out; 1220 } 1221 resumed = 1; 1222 if (!scsi_block_when_processing_errors(STp->device)) { 1223 retval = (-ENXIO); 1224 goto err_out; 1225 } ... 1264 err_out: 1265 normalize_buffer(STp->buffer); 1266 spin_lock(&st_use_lock); 1267 STp->in_use = 0; 1268 spin_unlock(&st_use_lock); 1269 scsi_tape_put(STp); <-- STp->device = 0 after this 1270 if (resumed) 1271 scsi_autopm_put_device(STp->device); 1272 return retval; The ref count for the struct scsi_tape had already been reduced to 1 when the .remove method of the st module had been called. The kref_put() in scsi_tape_put() caused scsi_tape_release() to be called: 0266 static void scsi_tape_put(struct scsi_tape *STp) 0267 { 0268 struct scsi_device *sdev = STp->device; 0269 0270 mutex_lock(&st_ref_mutex); 0271 kref_put(&STp->kref, scsi_tape_release); <-- calls this 0272 scsi_device_put(sdev); 0273 mutex_unlock(&st_ref_mutex); 0274 } In scsi_tape_release() the struct scsi_device in the struct scsi_tape gets set to NULL: 4273 static void scsi_tape_release(struct kref *kref) 4274 { 4275 struct scsi_tape *tpnt = to_scsi_tape(kref); 4276 struct gendisk *disk = tpnt->disk; 4277 4278 tpnt->device = NULL; <<<---- where the dev is nulled 4279 4280 if (tpnt->buffer) { 4281 normalize_buffer(tpnt->buffer); 4282 kfree(tpnt->buffer->reserved_pages); 4283 kfree(tpnt->buffer); 4284 } 4285 4286 disk->private_data = NULL; 4287 put_disk(disk); 4288 kfree(tpnt); 4289 return; 4290 } Although the problem was reported on SLES11.3 the problem appears in linux-next as well. The crash is fixed by reordering the code so we no longer access the struct scsi_tape after the kref_put() is done on it in st_open(). Signed-off-by: Shane Seymour Signed-off-by: Darren Lavender Reviewed-by: Johannes Thumshirn Acked-by: Kai Mäkisara Cc: stable@vger.kernel.org Signed-off-by: James Bottomley --- drivers/scsi/st.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/scsi/st.c b/drivers/scsi/st.c index 3f25b8fa921d..871f3553987d 100644 --- a/drivers/scsi/st.c +++ b/drivers/scsi/st.c @@ -1329,9 +1329,9 @@ static int st_open(struct inode *inode, struct file *filp) spin_lock(&st_use_lock); STp->in_use = 0; spin_unlock(&st_use_lock); - scsi_tape_put(STp); if (resumed) scsi_autopm_put_device(STp->device); + scsi_tape_put(STp); return retval; } -- cgit v1.2.3 From 1278dd6809b11dc298e19d81ac0916275f7318c1 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Mon, 13 Jul 2015 08:24:39 -0600 Subject: scsi: fix host max depth checking for the 'queue_depth' sysfs interface Commit 1e6f2416044c0 changed the scsi sysfs 'queue_depth' code to rejects depths higher than the scsi host template setting. But lots of hosts set this to 1, and update the settings in the scsi host when the controller/devices probing happens. This breaks (at least) mpt2sas and mpt3sas runtime setting of queue depth, returning EINVAL for all settings but '1'. And once it's set to 1, there's no way to go back up. Cc: stable@vger.kernel.org Fixes: 1e6f2416044c0 "scsi: don't allow setting of queue_depth bigger than can_queue" Signed-off-by: Jens Axboe Reviewed-by: Martin K. Petersen Reviewed-by: Christoph Hellwig Signed-off-by: James Bottomley --- drivers/scsi/scsi_sysfs.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/scsi/scsi_sysfs.c b/drivers/scsi/scsi_sysfs.c index 1ac38e73df7e..9ad41168d26d 100644 --- a/drivers/scsi/scsi_sysfs.c +++ b/drivers/scsi/scsi_sysfs.c @@ -859,7 +859,7 @@ sdev_store_queue_depth(struct device *dev, struct device_attribute *attr, depth = simple_strtoul(buf, NULL, 0); - if (depth < 1 || depth > sht->can_queue) + if (depth < 1 || depth > sdev->host->can_queue) return -EINVAL; retval = sht->change_queue_depth(sdev, depth); -- cgit v1.2.3