public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	stable@vger.kernel.org, Seiji Aguchi <seiji.aguchi@hds.com>,
	Don Zickus <dzickus@redhat.com>, Tony Luck <tony.luck@intel.com>,
	CAI Qian <caiqian@redhat.com>
Subject: [ 34/46] pstore: Avoid deadlock in panic and emergency-restart path
Date: Fri,  1 Mar 2013 11:45:22 -0800	[thread overview]
Message-ID: <20130301194435.868452642@linuxfoundation.org> (raw)
In-Reply-To: <20130301194432.263409302@linuxfoundation.org>

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

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

From: Seiji Aguchi <seiji.aguchi@hds.com>

commit 9f244e9cfd70c7c0f82d3c92ce772ab2a92d9f64 upstream.

[Issue]

When pstore is in panic and emergency-restart paths, it may be blocked
in those paths because it simply takes spin_lock.

This is an example scenario which pstore may hang up in a panic path:

 - cpuA grabs psinfo->buf_lock
 - cpuB panics and calls smp_send_stop
 - smp_send_stop sends IRQ to cpuA
 - after 1 second, cpuB gives up on cpuA and sends an NMI instead
 - cpuA is now in an NMI handler while still holding buf_lock
 - cpuB is deadlocked

This case may happen if a firmware has a bug and
cpuA is stuck talking with it more than one second.

Also, this is a similar scenario in an emergency-restart path:

 - cpuA grabs psinfo->buf_lock and stucks in a firmware
 - cpuB kicks emergency-restart via either sysrq-b or hangcheck timer.
   And then, cpuB is deadlocked by taking psinfo->buf_lock again.

[Solution]

This patch avoids the deadlocking issues in both panic and emergency_restart
paths by introducing a function, is_non_blocking_path(), to check if a cpu
can be blocked in current path.

With this patch, pstore is not blocked even if another cpu has
taken a spin_lock, in those paths by changing from spin_lock_irqsave
to spin_trylock_irqsave.

In addition, according to a comment of emergency_restart() in kernel/sys.c,
spin_lock shouldn't be taken in an emergency_restart path to avoid
deadlock. This patch fits the comment below.

<snip>
/**
 *      emergency_restart - reboot the system
 *
 *      Without shutting down any hardware or taking any locks
 *      reboot the system.  This is called when we know we are in
 *      trouble so this is our best effort to reboot.  This is
 *      safe to call in interrupt context.
 */
void emergency_restart(void)
<snip>

Signed-off-by: Seiji Aguchi <seiji.aguchi@hds.com>
Acked-by: Don Zickus <dzickus@redhat.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Cc: CAI Qian <caiqian@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 fs/pstore/platform.c   |   35 +++++++++++++++++++++++++++++------
 include/linux/pstore.h |    6 ++++++
 2 files changed, 35 insertions(+), 6 deletions(-)

--- a/fs/pstore/platform.c
+++ b/fs/pstore/platform.c
@@ -88,6 +88,27 @@ static const char *get_reason_str(enum k
 	}
 }
 
+bool pstore_cannot_block_path(enum kmsg_dump_reason reason)
+{
+	/*
+	 * In case of NMI path, pstore shouldn't be blocked
+	 * regardless of reason.
+	 */
+	if (in_nmi())
+		return true;
+
+	switch (reason) {
+	/* In panic case, other cpus are stopped by smp_send_stop(). */
+	case KMSG_DUMP_PANIC:
+	/* Emergency restart shouldn't be blocked by spin lock. */
+	case KMSG_DUMP_EMERG:
+		return true;
+	default:
+		return false;
+	}
+}
+EXPORT_SYMBOL_GPL(pstore_cannot_block_path);
+
 /*
  * callback from kmsg_dump. (s2,l2) has the most recently
  * written bytes, older bytes are in (s1,l1). Save as much
@@ -111,10 +132,12 @@ static void pstore_dump(struct kmsg_dump
 
 	why = get_reason_str(reason);
 
-	if (in_nmi()) {
-		is_locked = spin_trylock(&psinfo->buf_lock);
-		if (!is_locked)
-			pr_err("pstore dump routine blocked in NMI, may corrupt error record\n");
+	if (pstore_cannot_block_path(reason)) {
+		is_locked = spin_trylock_irqsave(&psinfo->buf_lock, flags);
+		if (!is_locked) {
+			pr_err("pstore dump routine blocked in %s path, may corrupt error record\n"
+				       , in_nmi() ? "NMI" : why);
+		}
 	} else
 		spin_lock_irqsave(&psinfo->buf_lock, flags);
 	oopscount++;
@@ -145,9 +168,9 @@ static void pstore_dump(struct kmsg_dump
 		total += l1_cpy + l2_cpy;
 		part++;
 	}
-	if (in_nmi()) {
+	if (pstore_cannot_block_path(reason)) {
 		if (is_locked)
-			spin_unlock(&psinfo->buf_lock);
+			spin_unlock_irqrestore(&psinfo->buf_lock, flags);
 	} else
 		spin_unlock_irqrestore(&psinfo->buf_lock, flags);
 }
--- a/include/linux/pstore.h
+++ b/include/linux/pstore.h
@@ -54,12 +54,18 @@ struct pstore_info {
 
 #ifdef CONFIG_PSTORE
 extern int pstore_register(struct pstore_info *);
+extern bool pstore_cannot_block_path(enum kmsg_dump_reason reason);
 #else
 static inline int
 pstore_register(struct pstore_info *psi)
 {
 	return -ENODEV;
 }
+static inline bool
+pstore_cannot_block_path(enum kmsg_dump_reason reason)
+{
+	return false;
+}
 #endif
 
 #endif /*_LINUX_PSTORE_H*/



  parent reply	other threads:[~2013-03-01 19:58 UTC|newest]

Thread overview: 49+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-03-01 19:44 [ 00/46] 3.4.35-stable review Greg Kroah-Hartman
2013-03-01 19:44 ` [ 01/46] ALSA: hda - hdmi: Make jacks phantom, if theyre not detectable Greg Kroah-Hartman
2013-03-01 19:44 ` [ 02/46] quota: autoload the quota_v2 module for QFMT_VFS_V1 quota format Greg Kroah-Hartman
2013-03-01 19:44 ` [ 03/46] iommu/amd: Initialize device table after dma_ops Greg Kroah-Hartman
2013-03-01 19:44 ` [ 04/46] posix-timer: Dont call idr_find() with out-of-range ID Greg Kroah-Hartman
2013-03-01 19:44 ` [ 05/46] ftrace: Call ftrace cleanup module notifier after all other notifiers Greg Kroah-Hartman
2013-03-01 19:44 ` [ 06/46] x86, efi: Make "noefi" really disable EFI runtime serivces Greg Kroah-Hartman
2013-03-01 19:44 ` [ 07/46] doc, xen: Mention earlyprintk=xen in the documentation Greg Kroah-Hartman
2013-03-01 19:44 ` [ 08/46] doc, kernel-parameters: Document console=hvc<n> Greg Kroah-Hartman
2013-03-01 19:44 ` [ 09/46] x86: Make sure we can boot in the case the BDA contains pure garbage Greg Kroah-Hartman
2013-03-01 19:44 ` [ 10/46] target: Fix lookup of dynamic NodeACLs during cached demo-mode operation Greg Kroah-Hartman
2013-03-01 19:44 ` [ 11/46] target: Add missing mapped_lun bounds checking during make_mappedlun setup Greg Kroah-Hartman
2013-03-01 19:45 ` [ 12/46] ocfs2: fix possible use-after-free with AIO Greg Kroah-Hartman
2013-03-01 19:45 ` [ 13/46] ocfs2: fix ocfs2_init_security_and_acl() to initialize acl correctly Greg Kroah-Hartman
2013-03-01 19:45 ` [ 14/46] ocfs2: ac->ac_allow_chain_relink=0 wont disable group relink Greg Kroah-Hartman
2013-03-01 19:45 ` [ 15/46] block: fix ext_devt_idr handling Greg Kroah-Hartman
2013-03-01 19:45 ` [ 16/46] xen-blkback: do not leak mode property Greg Kroah-Hartman
2013-03-01 19:45 ` [ 17/46] xen/blkback: Dont trust the handle from the frontend Greg Kroah-Hartman
2013-03-01 19:45 ` [ 18/46] idr: fix a subtle bug in idr_get_next() Greg Kroah-Hartman
2013-03-01 19:45 ` [ 19/46] block: fix synchronization and limit check in blk_alloc_devt() Greg Kroah-Hartman
2013-03-01 19:45 ` [ 20/46] firewire: add minor number range check to fw_device_init() Greg Kroah-Hartman
2013-03-01 19:45 ` [ 21/46] sysctl: fix null checking in bin_dn_node_address() Greg Kroah-Hartman
2013-03-01 19:45 ` [ 22/46] fs: Fix possible use-after-free with AIO Greg Kroah-Hartman
2013-03-01 19:45 ` [ 23/46] media: rc: unlock on error in show_protocols() Greg Kroah-Hartman
2013-03-01 19:45 ` [ 24/46] ext4: check bh in ext4_read_block_bitmap() Greg Kroah-Hartman
2013-03-01 19:45 ` [ 25/46] ext4: fix race in ext4_mb_add_n_trim() Greg Kroah-Hartman
2013-03-01 19:45 ` [ 26/46] ext4: fix xattr block allocation/release with bigalloc Greg Kroah-Hartman
2013-03-01 19:45 ` [ 27/46] ext4: fix free clusters calculation in bigalloc filesystem Greg Kroah-Hartman
2013-03-01 19:45 ` [ 28/46] nfsd: Fix memleak Greg Kroah-Hartman
2013-03-01 19:45 ` [ 29/46] svcrpc: make svc_age_temp_xprts enqueue under sv_lock Greg Kroah-Hartman
2013-03-01 19:45 ` [ 30/46] vhost: fix length for cross region descriptor Greg Kroah-Hartman
2013-03-01 19:45 ` [ 31/46] fuse: dont WARN when nlink is zero Greg Kroah-Hartman
2013-03-01 19:45 ` [ 32/46] unbreak automounter support on 64-bit kernel with 32-bit userspace (v2) Greg Kroah-Hartman
2013-03-01 19:45 ` [ 33/46] ath9k_hw: fix calibration issues on chainmask that dont include chain 0 Greg Kroah-Hartman
2013-03-01 19:45 ` Greg Kroah-Hartman [this message]
2013-03-01 19:45 ` [ 35/46] cpuset: fix cpuset_print_task_mems_allowed() vs rename() race Greg Kroah-Hartman
2013-03-01 19:45 ` [ 36/46] cgroup: fix exit() vs rmdir() race Greg Kroah-Hartman
2013-03-01 19:45 ` [ 37/46] ab8500-chargalg: Only root should have write permission on sysfs file Greg Kroah-Hartman
2013-03-01 19:45 ` [ 38/46] ab8500_btemp: Demote initcall sequence Greg Kroah-Hartman
2013-03-01 19:45 ` [ 39/46] ACPI: Add DMI entry for Sony VGN-FW41E_H Greg Kroah-Hartman
2013-03-01 19:45 ` [ 40/46] staging: comedi: ni_labpc: correct differential channel sequence for AI commands Greg Kroah-Hartman
2013-03-01 19:45 ` [ 41/46] staging: comedi: ni_labpc: set up command4 register *after* command3 Greg Kroah-Hartman
2013-03-01 19:45 ` [ 42/46] staging: comedi: check s->async for poll(), read() and write() Greg Kroah-Hartman
2013-03-01 19:45 ` [ 43/46] perf tools: Fix build with bison 2.3 and older Greg Kroah-Hartman
2013-03-01 19:45 ` [ 44/46] ata_piix: IDE-mode SATA patch for Intel Avoton DeviceIDs Greg Kroah-Hartman
2013-03-01 19:45 ` [ 45/46] ata_piix: Add Device IDs for Intel Wellsburg PCH Greg Kroah-Hartman
2013-03-01 19:45 ` [ 46/46] [hid] usb hid quirks for Masterkit MA901 usb radio Greg Kroah-Hartman
2013-03-02  3:59 ` [ 00/46] 3.4.35-stable review Shuah Khan
2013-03-03 11:48 ` Satoru Takeuchi

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=20130301194435.868452642@linuxfoundation.org \
    --to=gregkh@linuxfoundation.org \
    --cc=caiqian@redhat.com \
    --cc=dzickus@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=seiji.aguchi@hds.com \
    --cc=stable@vger.kernel.org \
    --cc=tony.luck@intel.com \
    /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