https://github.com/torvalds/linux
Revision a73914c35b05d80f8ce78288e10056c91090b666 authored by Mark Salyzyn on 22 September 2011, 15:32:23 UTC, committed by James Bottomley on 02 October 2011, 18:28:55 UTC
When a wide port is being utilized to a target, if one disables only one
of the
phys, we get an OS crash:

BUG: unable to handle kernel NULL pointer dereference at
0000000000000238
IP: [<ffffffff814ca9b1>] mutex_lock+0x21/0x50
PGD 4103f5067 PUD 41dba9067 PMD 0
Oops: 0002 [#1] SMP
last sysfs file: /sys/bus/pci/slots/5/address
CPU 0
Modules linked in: pm8001(U) ses enclosure fuse nfsd exportfs autofs4
ipmi_devintf ipmi_si ipmi_msghandler nfs lockd fscache nfs_acl
auth_rpcgss 8021q fcoe libfcoe garp libfc scsi_transport_fc stp scsi_tgt
llc sunrpc cpufreq_ondemand acpi_cpufreq freq_table ipv6 sr_mod cdrom
dm_mirror dm_region_hash dm_log uinput sg i2c_i801 i2c_core iTCO_wdt
iTCO_vendor_support e1000e mlx4_ib ib_mad ib_core mlx4_en mlx4_core ext3
jbd mbcache sd_mod crc_t10dif usb_storage ata_generic pata_acpi ata_piix
libsas(U) scsi_transport_sas dm_mod [last unloaded: pm8001]

Modules linked in: pm8001(U) ses enclosure fuse nfsd exportfs autofs4
ipmi_devintf ipmi_si ipmi_msghandler nfs lockd fscache nfs_acl
auth_rpcgss 8021q fcoe libfcoe garp libfc scsi_transport_fc stp scsi_tgt
llc sunrpc cpufreq_ondemand acpi_cpufreq freq_table ipv6 sr_mod cdrom
dm_mirror dm_region_hash dm_log uinput sg i2c_i801 i2c_core iTCO_wdt
iTCO_vendor_support e1000e mlx4_ib ib_mad ib_core mlx4_en mlx4_core ext3
jbd mbcache sd_mod crc_t10dif usb_storage ata_generic pata_acpi ata_piix
libsas(U) scsi_transport_sas dm_mod [last unloaded: pm8001]
Pid: 5146, comm: scsi_wq_5 Not tainted
2.6.32-71.29.1.el6.lustre.7.x86_64 #1 Storage Server
RIP: 0010:[<ffffffff814ca9b1>]  [<ffffffff814ca9b1>]
mutex_lock+0x21/0x50
RSP: 0018:ffff8803e4e33d30  EFLAGS: 00010246
RAX: 0000000000000000 RBX: 0000000000000238 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffff8803e664c800 RDI: 0000000000000238
RBP: ffff8803e4e33d40 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000000000
R13: 0000000000000238 R14: ffff88041acb7200 R15: ffff88041c51ada0
FS:  0000000000000000(0000) GS:ffff880028200000(0000)
knlGS:0000000000000000
CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
CR2: 0000000000000238 CR3: 0000000410143000 CR4: 00000000000006f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process scsi_wq_5 (pid: 5146, threadinfo ffff8803e4e32000, task
ffff8803e4e294a0)
Stack:
 ffff8803e664c800 0000000000000000 ffff8803e4e33d70 ffffffffa001f06e
<0> ffff8803e4e33d60 ffff88041c51ada0 ffff88041acb7200 ffff88041bc0aa00
<0> ffff8803e4e33d90 ffffffffa0032b6c 0000000000000014 ffff88041acb7200
Call Trace:
 [<ffffffffa001f06e>] sas_port_delete_phy+0x2e/0xa0 [scsi_transport_sas]
 [<ffffffffa0032b6c>] sas_unregister_devs_sas_addr+0xac/0xe0 [libsas]
 [<ffffffffa0034914>] sas_ex_revalidate_domain+0x204/0x330 [libsas]
 [<ffffffffa00307f0>] ? sas_revalidate_domain+0x0/0x90 [libsas]
 [<ffffffffa0030855>] sas_revalidate_domain+0x65/0x90 [libsas]
 [<ffffffff8108c7d0>] worker_thread+0x170/0x2a0
 [<ffffffff81091ea0>] ? autoremove_wake_function+0x0/0x40
 [<ffffffff8108c660>] ? worker_thread+0x0/0x2a0
 [<ffffffff81091b36>] kthread+0x96/0xa0
 [<ffffffff810141ca>] child_rip+0xa/0x20
 [<ffffffff81091aa0>] ? kthread+0x0/0xa0
 [<ffffffff810141c0>] ? child_rip+0x0/0x20
Code: ff ff 85 c0 75 ed eb d6 66 90 55 48 89 e5 48 83 ec 10 48 89 1c 24
4c 89 64 24 08 0f 1f 44 00 00 48 89 fb e8 92 f4 ff ff 48 89 df <f0> ff
0f 79 05 e8 25 00 00 00 65 48 8b 04 25 08 cc 00 00 48 2d
RIP  [<ffffffff814ca9b1>] mutex_lock+0x21/0x50
 RSP <ffff8803e4e33d30>
CR2: 0000000000000238

The following patch is admittedly a band-aid, and does not solve the
root cause, but it still is a good candidate for hardening as a pointer
check before reference.

Signed-off-by: Mark Salyzyn <mark_salyzyn@us.xyratex.com>
Tested-by: Jack Wang <jack_wang@usish.com>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
1 parent 9bfacd0
Raw File
Tip revision: a73914c35b05d80f8ce78288e10056c91090b666 authored by Mark Salyzyn on 22 September 2011, 15:32:23 UTC
[SCSI] libsas: fix panic when single phy is disabled on a wide port
Tip revision: a73914c
blk-timeout.c
/*
 * Functions related to generic timeout handling of requests.
 */
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/blkdev.h>
#include <linux/fault-inject.h>

#include "blk.h"

#ifdef CONFIG_FAIL_IO_TIMEOUT

static DECLARE_FAULT_ATTR(fail_io_timeout);

static int __init setup_fail_io_timeout(char *str)
{
	return setup_fault_attr(&fail_io_timeout, str);
}
__setup("fail_io_timeout=", setup_fail_io_timeout);

int blk_should_fake_timeout(struct request_queue *q)
{
	if (!test_bit(QUEUE_FLAG_FAIL_IO, &q->queue_flags))
		return 0;

	return should_fail(&fail_io_timeout, 1);
}

static int __init fail_io_timeout_debugfs(void)
{
	struct dentry *dir = fault_create_debugfs_attr("fail_io_timeout",
						NULL, &fail_io_timeout);

	return IS_ERR(dir) ? PTR_ERR(dir) : 0;
}

late_initcall(fail_io_timeout_debugfs);

ssize_t part_timeout_show(struct device *dev, struct device_attribute *attr,
			  char *buf)
{
	struct gendisk *disk = dev_to_disk(dev);
	int set = test_bit(QUEUE_FLAG_FAIL_IO, &disk->queue->queue_flags);

	return sprintf(buf, "%d\n", set != 0);
}

ssize_t part_timeout_store(struct device *dev, struct device_attribute *attr,
			   const char *buf, size_t count)
{
	struct gendisk *disk = dev_to_disk(dev);
	int val;

	if (count) {
		struct request_queue *q = disk->queue;
		char *p = (char *) buf;

		val = simple_strtoul(p, &p, 10);
		spin_lock_irq(q->queue_lock);
		if (val)
			queue_flag_set(QUEUE_FLAG_FAIL_IO, q);
		else
			queue_flag_clear(QUEUE_FLAG_FAIL_IO, q);
		spin_unlock_irq(q->queue_lock);
	}

	return count;
}

#endif /* CONFIG_FAIL_IO_TIMEOUT */

/*
 * blk_delete_timer - Delete/cancel timer for a given function.
 * @req:	request that we are canceling timer for
 *
 */
void blk_delete_timer(struct request *req)
{
	list_del_init(&req->timeout_list);
}

static void blk_rq_timed_out(struct request *req)
{
	struct request_queue *q = req->q;
	enum blk_eh_timer_return ret;

	ret = q->rq_timed_out_fn(req);
	switch (ret) {
	case BLK_EH_HANDLED:
		__blk_complete_request(req);
		break;
	case BLK_EH_RESET_TIMER:
		blk_clear_rq_complete(req);
		blk_add_timer(req);
		break;
	case BLK_EH_NOT_HANDLED:
		/*
		 * LLD handles this for now but in the future
		 * we can send a request msg to abort the command
		 * and we can move more of the generic scsi eh code to
		 * the blk layer.
		 */
		break;
	default:
		printk(KERN_ERR "block: bad eh return: %d\n", ret);
		break;
	}
}

void blk_rq_timed_out_timer(unsigned long data)
{
	struct request_queue *q = (struct request_queue *) data;
	unsigned long flags, next = 0;
	struct request *rq, *tmp;
	int next_set = 0;

	spin_lock_irqsave(q->queue_lock, flags);

	list_for_each_entry_safe(rq, tmp, &q->timeout_list, timeout_list) {
		if (time_after_eq(jiffies, rq->deadline)) {
			list_del_init(&rq->timeout_list);

			/*
			 * Check if we raced with end io completion
			 */
			if (blk_mark_rq_complete(rq))
				continue;
			blk_rq_timed_out(rq);
		} else if (!next_set || time_after(next, rq->deadline)) {
			next = rq->deadline;
			next_set = 1;
		}
	}

	if (next_set)
		mod_timer(&q->timeout, round_jiffies_up(next));

	spin_unlock_irqrestore(q->queue_lock, flags);
}

/**
 * blk_abort_request -- Request request recovery for the specified command
 * @req:	pointer to the request of interest
 *
 * This function requests that the block layer start recovery for the
 * request by deleting the timer and calling the q's timeout function.
 * LLDDs who implement their own error recovery MAY ignore the timeout
 * event if they generated blk_abort_req. Must hold queue lock.
 */
void blk_abort_request(struct request *req)
{
	if (blk_mark_rq_complete(req))
		return;
	blk_delete_timer(req);
	blk_rq_timed_out(req);
}
EXPORT_SYMBOL_GPL(blk_abort_request);

/**
 * blk_add_timer - Start timeout timer for a single request
 * @req:	request that is about to start running.
 *
 * Notes:
 *    Each request has its own timer, and as it is added to the queue, we
 *    set up the timer. When the request completes, we cancel the timer.
 */
void blk_add_timer(struct request *req)
{
	struct request_queue *q = req->q;
	unsigned long expiry;

	if (!q->rq_timed_out_fn)
		return;

	BUG_ON(!list_empty(&req->timeout_list));
	BUG_ON(test_bit(REQ_ATOM_COMPLETE, &req->atomic_flags));

	/*
	 * Some LLDs, like scsi, peek at the timeout to prevent a
	 * command from being retried forever.
	 */
	if (!req->timeout)
		req->timeout = q->rq_timeout;

	req->deadline = jiffies + req->timeout;
	list_add_tail(&req->timeout_list, &q->timeout_list);

	/*
	 * If the timer isn't already pending or this timeout is earlier
	 * than an existing one, modify the timer. Round up to next nearest
	 * second.
	 */
	expiry = round_jiffies_up(req->deadline);

	if (!timer_pending(&q->timeout) ||
	    time_before(expiry, q->timeout.expires))
		mod_timer(&q->timeout, expiry);
}

/**
 * blk_abort_queue -- Abort all request on given queue
 * @queue:	pointer to queue
 *
 */
void blk_abort_queue(struct request_queue *q)
{
	unsigned long flags;
	struct request *rq, *tmp;
	LIST_HEAD(list);

	/*
	 * Not a request based block device, nothing to abort
	 */
	if (!q->request_fn)
		return;

	spin_lock_irqsave(q->queue_lock, flags);

	elv_abort_queue(q);

	/*
	 * Splice entries to local list, to avoid deadlocking if entries
	 * get readded to the timeout list by error handling
	 */
	list_splice_init(&q->timeout_list, &list);

	list_for_each_entry_safe(rq, tmp, &list, timeout_list)
		blk_abort_request(rq);

	/*
	 * Occasionally, blk_abort_request() will return without
	 * deleting the element from the list. Make sure we add those back
	 * instead of leaving them on the local stack list.
	 */
	list_splice(&list, &q->timeout_list);

	spin_unlock_irqrestore(q->queue_lock, flags);

}
EXPORT_SYMBOL_GPL(blk_abort_queue);
back to top