From 54b2b50c20a61b51199bedb6e5d2f8ec2568fb43 Mon Sep 17 00:00:00 2001 From: "Martin K. Petersen" Date: Wed, 23 Oct 2013 06:25:40 -0400 Subject: [SCSI] Disable WRITE SAME for RAID and virtual host adapter drivers Some host adapters do not pass commands through to the target disk directly. Instead they provide an emulated target which may or may not accurately report its capabilities. In some cases the physical device characteristics are reported even when the host adapter is processing commands on the device's behalf. This can lead to adapter firmware hangs or excessive I/O errors. This patch disables WRITE SAME for devices connected to host adapters that provide an emulated target. Driver writers can disable WRITE SAME by setting the no_write_same flag in the host adapter template. [jejb: fix up rejections due to eh_deadline patch] Signed-off-by: Martin K. Petersen Cc: stable@kernel.org Signed-off-by: James Bottomley --- drivers/ata/libata-scsi.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/ata/libata-scsi.c') diff --git a/drivers/ata/libata-scsi.c b/drivers/ata/libata-scsi.c index db6dfcfa3e2..ab58556d347 100644 --- a/drivers/ata/libata-scsi.c +++ b/drivers/ata/libata-scsi.c @@ -3625,6 +3625,7 @@ int ata_scsi_add_hosts(struct ata_host *host, struct scsi_host_template *sht) shost->max_lun = 1; shost->max_channel = 1; shost->max_cmd_len = 16; + shost->no_write_same = 1; /* Schedule policy is determined by ->qc_defer() * callback and it needs to see every deferred qc. -- cgit v1.2.3-70-g09d2 From 85fbd722ad0f5d64d1ad15888cd1eb2188bfb557 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 18 Dec 2013 07:07:32 -0500 Subject: libata, freezer: avoid block device removal while system is frozen MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Freezable kthreads and workqueues are fundamentally problematic in that they effectively introduce a big kernel lock widely used in the kernel and have already been the culprit of several deadlock scenarios. This is the latest occurrence. During resume, libata rescans all the ports and revalidates all pre-existing devices. If it determines that a device has gone missing, the device is removed from the system which involves invalidating block device and flushing bdi while holding driver core layer locks. Unfortunately, this can race with the rest of device resume. Because freezable kthreads and workqueues are thawed after device resume is complete and block device removal depends on freezable workqueues and kthreads (e.g. bdi_wq, jbd2) to make progress, this can lead to deadlock - block device removal can't proceed because kthreads are frozen and kthreads can't be thawed because device resume is blocked behind block device removal. 839a8e8660b6 ("writeback: replace custom worker pool implementation with unbound workqueue") made this particular deadlock scenario more visible but the underlying problem has always been there - the original forker task and jbd2 are freezable too. In fact, this is highly likely just one of many possible deadlock scenarios given that freezer behaves as a big kernel lock and we don't have any debug mechanism around it. I believe the right thing to do is getting rid of freezable kthreads and workqueues. This is something fundamentally broken. For now, implement a funny workaround in libata - just avoid doing block device hot[un]plug while the system is frozen. Kernel engineering at its finest. :( v2: Add EXPORT_SYMBOL_GPL(pm_freezing) for cases where libata is built as a module. v3: Comment updated and polling interval changed to 10ms as suggested by Rafael. v4: Add #ifdef CONFIG_FREEZER around the hack as pm_freezing is not defined when FREEZER is not configured thus breaking build. Reported by kbuild test robot. Signed-off-by: Tejun Heo Reported-by: Tomaž Šolc Reviewed-by: "Rafael J. Wysocki" Link: https://bugzilla.kernel.org/show_bug.cgi?id=62801 Link: http://lkml.kernel.org/r/20131213174932.GA27070@htj.dyndns.org Cc: Greg Kroah-Hartman Cc: Len Brown Cc: Oleg Nesterov Cc: stable@vger.kernel.org Cc: kbuild test robot --- drivers/ata/libata-scsi.c | 21 +++++++++++++++++++++ kernel/freezer.c | 6 ++++++ 2 files changed, 27 insertions(+) (limited to 'drivers/ata/libata-scsi.c') diff --git a/drivers/ata/libata-scsi.c b/drivers/ata/libata-scsi.c index db6dfcfa3e2..176f62950e3 100644 --- a/drivers/ata/libata-scsi.c +++ b/drivers/ata/libata-scsi.c @@ -3871,6 +3871,27 @@ void ata_scsi_hotplug(struct work_struct *work) return; } + /* + * XXX - UGLY HACK + * + * The block layer suspend/resume path is fundamentally broken due + * to freezable kthreads and workqueue and may deadlock if a block + * device gets removed while resume is in progress. I don't know + * what the solution is short of removing freezable kthreads and + * workqueues altogether. + * + * The following is an ugly hack to avoid kicking off device + * removal while freezer is active. This is a joke but does avoid + * this particular deadlock scenario. + * + * https://bugzilla.kernel.org/show_bug.cgi?id=62801 + * http://marc.info/?l=linux-kernel&m=138695698516487 + */ +#ifdef CONFIG_FREEZER + while (pm_freezing) + msleep(10); +#endif + DPRINTK("ENTER\n"); mutex_lock(&ap->scsi_scan_mutex); diff --git a/kernel/freezer.c b/kernel/freezer.c index b462fa19751..aa6a8aadb91 100644 --- a/kernel/freezer.c +++ b/kernel/freezer.c @@ -19,6 +19,12 @@ EXPORT_SYMBOL(system_freezing_cnt); bool pm_freezing; bool pm_nosig_freezing; +/* + * Temporary export for the deadlock workaround in ata_scsi_hotplug(). + * Remove once the hack becomes unnecessary. + */ +EXPORT_SYMBOL_GPL(pm_freezing); + /* protects freezing and frozen transitions */ static DEFINE_SPINLOCK(freezer_lock); -- cgit v1.2.3-70-g09d2