linux/arch
Srivatsa S. Bhat b11e3d782b x86, mce: Fix rcu splat in drain_mce_log_buffer()
While booting, the following message is seen:

[   21.665087] ===============================
[   21.669439] [ INFO: suspicious RCU usage. ]
[   21.673798] 3.2.0-0.0.0.28.36b5ec9-default #2 Not tainted
[   21.681353] -------------------------------
[   21.685864] arch/x86/kernel/cpu/mcheck/mce.c:194 suspicious rcu_dereference_index_check() usage!
[   21.695013]
[   21.695014] other info that might help us debug this:
[   21.695016]
[   21.703488]
[   21.703489] rcu_scheduler_active = 1, debug_locks = 1
[   21.710426] 3 locks held by modprobe/2139:
[   21.714754]  #0:  (&__lockdep_no_validate__){......}, at: [<ffffffff8133afd3>] __driver_attach+0x53/0xa0
[   21.725020]  #1:
[   21.725323] ioatdma: Intel(R) QuickData Technology Driver 4.00
[   21.733206]  (&__lockdep_no_validate__){......}, at: [<ffffffff8133afe1>] __driver_attach+0x61/0xa0
[   21.743015]  #2:  (i7core_edac_lock){+.+.+.}, at: [<ffffffffa01cfa5f>] i7core_probe+0x1f/0x5c0 [i7core_edac]
[   21.753708]
[   21.753709] stack backtrace:
[   21.758429] Pid: 2139, comm: modprobe Not tainted 3.2.0-0.0.0.28.36b5ec9-default #2
[   21.768253] Call Trace:
[   21.770838]  [<ffffffff810977cd>] lockdep_rcu_suspicious+0xcd/0x100
[   21.777366]  [<ffffffff8101aa41>] drain_mcelog_buffer+0x191/0x1b0
[   21.783715]  [<ffffffff8101aa78>] mce_register_decode_chain+0x18/0x20
[   21.790430]  [<ffffffffa01cf8db>] i7core_register_mci+0x2fb/0x3e4 [i7core_edac]
[   21.798003]  [<ffffffffa01cfb14>] i7core_probe+0xd4/0x5c0 [i7core_edac]
[   21.804809]  [<ffffffff8129566b>] local_pci_probe+0x5b/0xe0
[   21.810631]  [<ffffffff812957c9>] __pci_device_probe+0xd9/0xe0
[   21.816650]  [<ffffffff813362e4>] ? get_device+0x14/0x20
[   21.822178]  [<ffffffff81296916>] pci_device_probe+0x36/0x60
[   21.828061]  [<ffffffff8133ac8a>] really_probe+0x7a/0x2b0
[   21.833676]  [<ffffffff8133af23>] driver_probe_device+0x63/0xc0
[   21.839868]  [<ffffffff8133b01b>] __driver_attach+0x9b/0xa0
[   21.845718]  [<ffffffff8133af80>] ? driver_probe_device+0xc0/0xc0
[   21.852027]  [<ffffffff81339168>] bus_for_each_dev+0x68/0x90
[   21.857876]  [<ffffffff8133aa3c>] driver_attach+0x1c/0x20
[   21.863462]  [<ffffffff8133a64d>] bus_add_driver+0x16d/0x2b0
[   21.869377]  [<ffffffff8133b6dc>] driver_register+0x7c/0x160
[   21.875220]  [<ffffffff81296bda>] __pci_register_driver+0x6a/0xf0
[   21.881494]  [<ffffffffa01fe000>] ? 0xffffffffa01fdfff
[   21.886846]  [<ffffffffa01fe047>] i7core_init+0x47/0x1000 [i7core_edac]
[   21.893737]  [<ffffffff810001ce>] do_one_initcall+0x3e/0x180
[   21.899670]  [<ffffffff810a9b95>] sys_init_module+0xc5/0x220
[   21.905542]  [<ffffffff8149bc39>] system_call_fastpath+0x16/0x1b

Fix this by using ACCESS_ONCE() instead of rcu_dereference_check_mce()
over mcelog.next. Since the access to each entry is controlled by the
->finished field, ACCESS_ONCE() should work just fine. An rcu_dereference
is unnecessary here.

Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Suggested-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Tony Luck <tony.luck@intel.com>
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
2012-03-07 11:44:29 +01:00
..
alpha Merge branch 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci 2012-01-11 18:50:26 -08:00
arm Another set of arm-soc bug fixes on top of v3.3-rc5. The few larger 2012-02-29 11:24:39 -08:00
avr32 avr32: select generic atomic64_t support 2012-02-03 16:16:40 -08:00
blackfin Merge branch 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci 2012-01-11 18:50:26 -08:00
c6x Kbuild: Use dtc's -d (dependency) option 2012-02-14 21:14:44 -05:00
cris lib: use generic pci_iomap on all architectures 2012-01-10 18:04:27 -08:00
frv Merge branch 'for-next' of git://git.infradead.org/users/dhowells/linux-headers 2012-01-14 18:03:30 -08:00
h8300 Merge branch 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci 2012-01-11 18:50:26 -08:00
hexagon
ia64 Merge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux 2012-01-18 15:51:48 -08:00
m32r Merge branch 'for-next' of git://git.infradead.org/users/dhowells/linux-headers 2012-01-14 18:03:30 -08:00
m68k m68k: Do not set global share for non-kernel shared pages 2012-02-07 11:25:57 +10:00
microblaze Revert "microblaze: Add topology init" 2012-01-31 09:06:52 +01:00
mips mips: use the the PCI controller's io_map_base 2012-01-31 23:20:30 +02:00
mn10300 Merge branch 'for-next' of git://git.infradead.org/users/dhowells/linux-headers 2012-01-14 18:03:30 -08:00
openrisc openrisc: Fix up audit_syscall_[entry|exit]() usage 2012-02-17 09:53:12 +01:00
parisc [PARISC] don't unconditionally override CROSS_COMPILE for 64 bit. 2012-02-28 09:39:28 -06:00
powerpc powerpc: Fix various issues with return to userspace 2012-02-22 16:48:53 +11:00
s390 Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux 2012-03-01 18:22:55 -08:00
score score: fix off-by-one index into syscall table 2012-01-23 08:38:49 -08:00
sh SuperH fixes for 3.3-rc5 2012-02-24 08:56:51 -08:00
sparc lib: Fix multiple definitions of clz_tab 2012-02-02 10:34:23 +11:00
tile Merge branch 'for-next' of git://git.infradead.org/users/dhowells/linux-headers 2012-01-14 18:03:30 -08:00
um Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/audit 2012-01-17 16:41:31 -08:00
unicore32 Merge branch 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci 2012-01-11 18:50:26 -08:00
x86 x86, mce: Fix rcu splat in drain_mce_log_buffer() 2012-03-07 11:44:29 +01:00
xtensa xtensa: fix memscan() 2012-02-03 16:16:40 -08:00
.gitignore
Kconfig mm,x86,um: move CMPXCHG_DOUBLE config option 2012-01-12 20:13:03 -08:00