mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Greg KH <gregkh@suse.de>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: torvalds@linux-foundation.org, akpm@linux-foundation.org,
	alan@lxorguk.ukuu.org.uk,
	Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>,
	Jens Axboe <axboe@kernel.dk>
Subject: [07/67] cfq-iosched: fix cfq_cic_link() race confition
Date: Tue, 03 Jan 2012 14:29:25 -0800	[thread overview]
Message-ID: <20120103223031.893661534@clark.kroah.org> (raw)
In-Reply-To: <20120103223043.GA26738@kroah.com>

3.0-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>

commit 5eb46851de3904cd1be9192fdacb8d34deadc1fc upstream.

cfq_cic_link() has race condition. When some processes which shared ioc
issue I/O to same block device simultaneously, cfq_cic_link() returns -EEXIST
sometimes. The race condition might stop I/O by following steps:

step  1: Process A: Issue an I/O to /dev/sda
step  2: Process A: Get an ioc (iocA here) in get_io_context() which does not
		    linked with a cic for the device
step  3: Process A: Get a new cic for the device (cicA here) in
		    cfq_alloc_io_context()

step  4: Process B: Issue an I/O to /dev/sda
step  5: Process B: Get iocA in get_io_context() since process A and B share the
		    same ioc
step  6: Process B: Get a new cic for the device (cicB here) in
		    cfq_alloc_io_context() since iocA has not been linked with a
		    cic for the device yet

step  7: Process A: Link cicA to iocA in cfq_cic_link()
step  8: Process A: Dispatch I/O to driver and finish it

step  9: Process B: Try to link cicB to iocA in cfq_cic_link()
		    But it fails with showing "cfq: cic link failed!" kernel
		    message, since iocA has already linked with cicA at step 7.
step 10: Process B: Wait for finishig I/O in get_request_wait()
		    The function does not wake up, when there is no I/O to the
		    device.

When cfq_cic_link() returns -EEXIST, it means ioc has already linked with cic.
So when cfq_cic_link() return -EEXIST, retry cfq_cic_lookup().

Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

---
 block/cfq-iosched.c |   11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -3169,7 +3169,7 @@ static int cfq_cic_link(struct cfq_data
 		}
 	}
 
-	if (ret)
+	if (ret && ret != -EEXIST)
 		printk(KERN_ERR "cfq: cic link failed!\n");
 
 	return ret;
@@ -3185,6 +3185,7 @@ cfq_get_io_context(struct cfq_data *cfqd
 {
 	struct io_context *ioc = NULL;
 	struct cfq_io_context *cic;
+	int ret;
 
 	might_sleep_if(gfp_mask & __GFP_WAIT);
 
@@ -3192,6 +3193,7 @@ cfq_get_io_context(struct cfq_data *cfqd
 	if (!ioc)
 		return NULL;
 
+retry:
 	cic = cfq_cic_lookup(cfqd, ioc);
 	if (cic)
 		goto out;
@@ -3200,7 +3202,12 @@ cfq_get_io_context(struct cfq_data *cfqd
 	if (cic == NULL)
 		goto err;
 
-	if (cfq_cic_link(cfqd, ioc, cic, gfp_mask))
+	ret = cfq_cic_link(cfqd, ioc, cic, gfp_mask);
+	if (ret == -EEXIST) {
+		/* someone has linked cic to ioc already */
+		cfq_cic_free(cic);
+		goto retry;
+	} else if (ret)
 		goto err_free;
 
 out:



  parent reply	other threads:[~2012-01-03 23:08 UTC|newest]

Thread overview: 82+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-01-03 22:30 [00/67] 3.0.16-stable review Greg KH
2012-01-03 22:29 ` [01/67] ARM: OMAP: rx51: fix USB Greg KH
2012-01-03 22:29 ` [02/67] ipip, sit: copy parms.name after register_netdevice Greg KH
2012-01-03 22:29 ` [03/67] rtc: Expire alarms after the time is set Greg KH
2012-01-03 22:46   ` John Stultz
2012-01-03 22:47     ` Greg KH
2012-01-03 23:11       ` John Stultz
2012-01-03 22:29 ` [04/67] rtc: m41t80: Workaround broken alarm functionality Greg KH
2012-01-03 22:29 ` [05/67] drm/i915: prevent division by zero when asking for chipset power Greg KH
2012-01-03 22:29 ` [06/67] cfq-iosched: free cic_index if blkio_alloc_blkg_stats fails Greg KH
2012-01-03 22:29 ` Greg KH [this message]
2012-01-03 22:29 ` [08/67] SCSI: zfcp: return early from slave_destroy if slave_alloc returned early Greg KH
2012-01-03 22:29 ` [09/67] SCSI: mpt2sas: _scsih_smart_predicted_fault uses GFP_KERNEL in interrupt context Greg KH
2012-01-03 22:29 ` [10/67] SCSI: fcoe: Fix preempt count leak in fcoe_filter_frames() Greg KH
2012-01-03 22:29 ` [11/67] mac80211: fix another race in aggregation start Greg KH
2012-01-03 22:29 ` [12/67] block: initialize request_queues numa node during Greg KH
2012-01-03 22:29 ` [13/67] ssb: fix init regression with SoCs Greg KH
2012-01-03 22:29 ` [14/67] MXC PWM: should active during DOZE/WAIT/DBG mode Greg KH
2012-01-03 22:29 ` [15/67] Input: synaptics - fix touchpad not working after S2R on Vostro V13 Greg KH
2012-01-03 22:29 ` [16/67] percpu: fix per_cpu_ptr_to_phys() handling of non-page-aligned addresses Greg KH
2012-01-03 22:29 ` [17/67] binary_sysctl(): fix memory leak Greg KH
2012-01-03 22:29 ` [18/67] oom: fix integer overflow of points in oom_badness Greg KH
2012-01-03 22:29 ` [19/67] oprofile: Fix uninitialized memory access when writing to writing to oprofilefs Greg KH
2012-01-03 22:29 ` [20/67] NFSv4.1: Ensure that we handle _all_ SEQUENCE status bits Greg KH
2012-01-03 22:29 ` [21/67] SELinux: Fix RCU deref check warning in sel_netport_insert() Greg KH
2012-01-03 22:29 ` [22/67] nilfs2: unbreak compat ioctl Greg KH
2012-01-03 22:29 ` [23/67] mmc: vub300: fix type of firmware_rom_wait_states module parameter Greg KH
2012-01-03 22:29 ` [24/67] cgroups: fix a css_set not found bug in cgroup_attach_proc Greg KH
2012-01-03 22:29 ` [25/67] mfd: Fix twl-core oops while calling twl_i2c_* for unbound driver Greg KH
2012-01-03 22:29 ` [26/67] vfs: __read_cache_page should use gfp argument rather than GFP_KERNEL Greg KH
2012-01-03 22:29 ` [27/67] media: s5p-fimc: Use correct fourcc for RGB565 colour format Greg KH
2012-01-03 22:29 ` [28/67] ath9k: fix max phy rate at rate control init Greg KH
2012-01-03 22:29 ` [29/67] iwlwifi: do not set the sequence control bit is not needed Greg KH
2012-01-03 22:29 ` [30/67] iwlwifi: allow to switch to HT40 if not associated Greg KH
2012-01-03 22:29 ` [31/67] memcg: keep root group unchanged if creation fails Greg KH
2012-01-03 22:29 ` [32/67] VFS: Fix race between CPU hotplug and lglocks Greg KH
2012-01-03 22:29 ` [33/67] ARM:imx:fix pwm period value Greg KH
2012-01-03 22:29 ` [34/67] ARM: 7214/1: mmc: mmci: Fixup handling of MCI_STARTBITERR Greg KH
2012-01-03 22:29 ` [35/67] ARM: 7220/1: mmc: mmci: Fixup error handling for dma Greg KH
2012-01-03 22:29 ` [36/67] oprofile, arm/sh: Fix oprofile_arch_exit() linkage issue Greg KH
2012-01-03 22:29 ` [37/67] futex: Fix uninterruptible loop due to gate_area Greg KH
2012-01-03 22:29 ` [38/67] watchdog: hpwdt: Changes to handle NX secure bit in 32bit path Greg KH
2012-01-03 22:29 ` [39/67] drm/radeon/kms: bail on BTC parts if MC ucode is missing Greg KH
2012-01-03 22:29 ` [40/67] mm: hugetlb: fix non-atomic enqueue of huge page Greg KH
2012-01-03 22:29 ` [41/67] mpt2sas crashes on shutdown Greg KH
2012-01-03 22:30 ` [42/67] sparc64: Fix MSIQ HV call ordering in pci_sun4v_msiq_build_irq() Greg KH
2012-01-03 22:30 ` [43/67] sparc32: Be less strict in matching %lo part of relocation Greg KH
2012-01-03 22:30 ` [44/67] sparc64: Patch sun4v code sequences properly on module load Greg KH
2012-01-03 22:30 ` [45/67] sparc: Kill custom io_remap_pfn_range() Greg KH
2012-01-03 22:30 ` [46/67] sparc32: Remove non-kernel code from memcpy implementation Greg KH
2012-01-03 22:30 ` [47/67] sparc32: Remove uses of %g7 in " Greg KH
2012-01-03 22:30 ` [48/67] sparc32: Correct the return value of memcpy Greg KH
2012-01-03 22:30 ` [49/67] sparc64: Fix masking and shifting in VIS fpcmp emulation Greg KH
2012-01-03 22:30 ` [50/67] sparc: Fix handling of orig_i0 wrt. debugging when restarting syscalls Greg KH
2012-01-03 22:30 ` [51/67] net: bpf_jit: fix an off-one bug in x86_64 cond jump target Greg KH
2012-01-03 22:30 ` [52/67] ppp: fix pptp double release_sock in pptp_bind() Greg KH
2012-01-03 22:30 ` [53/67] llc: llc_cmsg_rcv was getting called after sk_eat_skb Greg KH
2012-01-03 22:30 ` [54/67] mqprio: Avoid panic if no options are provided Greg KH
2012-01-03 22:30 ` [55/67] net: have ipconfig not wait if no dev is available Greg KH
2012-01-03 22:30 ` [56/67] sch_gred: should not use GFP_KERNEL while holding a spinlock Greg KH
2012-01-03 22:30 ` [57/67] sctp: fix incorrect overflow check on autoclose Greg KH
2012-01-03 22:30 ` [58/67] sctp: Do not account for sizeof(struct sk_buff) in estimated rwnd Greg KH
2012-01-03 22:30 ` [59/67] ipv4: flush route cache after change accept_local Greg KH
2012-01-03 22:30 ` [60/67] ipv4: reintroduce route cache garbage collector Greg KH
2012-01-03 22:30 ` [61/67] ipv4: using prefetch requires including prefetch.h Greg KH
2012-01-03 22:30 ` [62/67] iwlwifi: update SCD BC table for all SCD queues Greg KH
2012-01-03 22:30 ` [63/67] mfd: Fix mismatch in twl4030 mutex lock-unlock Greg KH
2012-01-03 22:30 ` [64/67] mfd: Copy the device pointer to the twl4030-madc structure Greg KH
2012-01-03 22:30 ` [65/67] mfd: Check for twl4030-madc NULL pointer Greg KH
2012-01-03 22:30 ` [66/67] mfd: Turn on the twl4030-madc MADC clock Greg KH
2012-01-03 22:30 ` [67/67] xen/swiotlb: Use page alignment for early buffer allocation Greg KH
2012-01-05 19:18 ` [00/67] 3.0.16-stable review Greg KH
2012-01-05 19:26   ` Greg KH
2012-01-05 22:46   ` Greg KH
2012-01-05 22:44     ` [40/73] mpt2sas: fix non-x86 crash on shutdown Greg KH
2012-01-05 22:44     ` [67/73] xfs: log the inode in ->write_inode calls for kupdate Greg KH
2012-01-05 22:44     ` [68/73] xfs: log all dirty inodes in xfs_fs_sync_fs Greg KH
2012-01-05 22:44     ` [69/73] drm/radeon/kms/atom: fix possible segfault in pm setup Greg KH
2012-01-05 22:44     ` [70/73] hung_task: fix false positive during vfork Greg KH
2012-01-05 22:44     ` [71/73] Revert "rtc: Disable the alarm in the hardware" Greg KH
2012-01-05 22:44     ` [72/73] ptrace: partially fix the do_wait(WEXITED) vs EXIT_DEAD->EXIT_ZOMBIE race Greg KH
2012-01-05 22:44     ` [73/73] ath9k: Fix kernel panic in AR2427 in AP mode Greg KH

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=20120103223031.893661534@clark.kroah.org \
    --to=gregkh@suse.de \
    --cc=akpm@linux-foundation.org \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=axboe@kernel.dk \
    --cc=isimatu.yasuaki@jp.fujitsu.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=stable@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    /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: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®