sata_fsl: save irqs while coalescing
authorAnthony Foiani <anthony.foiani@gmail.com>
Tue, 20 Aug 2013 01:20:30 +0000 (19:20 -0600)
committerTejun Heo <tj@kernel.org>
Tue, 20 Aug 2013 12:38:23 +0000 (08:38 -0400)
commit99bbdfa6bdcb4bdf5be914a48e9b46941bf30819
tree7458928e58efb2a35cc2b8b0b96cdaf64967ca56
parent8ffff94d20b7eb446e848e0046107d51b17a20a8
sata_fsl: save irqs while coalescing

Before this patch, I was seeing the following lockdep splat on my
MPC8315 (PPC32) target:

  [    9.086051] =================================
  [    9.090393] [ INFO: inconsistent lock state ]
  [    9.094744] 3.9.7-ajf-gc39503d #1 Not tainted
  [    9.099087] ---------------------------------
  [    9.103432] inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage.
  [    9.109431] scsi_eh_1/39 [HC1[1]:SC0[0]:HE0:SE1] takes:
  [    9.114642]  (&(&host->lock)->rlock){?.+...}, at: [<c02f4168>] sata_fsl_interrupt+0x50/0x250
  [    9.123137] {HARDIRQ-ON-W} state was registered at:
  [    9.128004]   [<c006cdb8>] lock_acquire+0x90/0xf4
  [    9.132737]   [<c043ef04>] _raw_spin_lock+0x34/0x4c
  [    9.137645]   [<c02f3560>] fsl_sata_set_irq_coalescing+0x68/0x100
  [    9.143750]   [<c02f36a0>] sata_fsl_init_controller+0xa8/0xc0
  [    9.149505]   [<c02f3f10>] sata_fsl_probe+0x17c/0x2e8
  [    9.154568]   [<c02acc90>] driver_probe_device+0x90/0x248
  [    9.159987]   [<c02acf0c>] __driver_attach+0xc4/0xc8
  [    9.164964]   [<c02aae74>] bus_for_each_dev+0x5c/0xa8
  [    9.170028]   [<c02ac218>] bus_add_driver+0x100/0x26c
  [    9.175091]   [<c02ad638>] driver_register+0x88/0x198
  [    9.180155]   [<c0003a24>] do_one_initcall+0x58/0x1b4
  [    9.185226]   [<c05aeeac>] kernel_init_freeable+0x118/0x1c0
  [    9.190823]   [<c0004110>] kernel_init+0x18/0x108
  [    9.195542]   [<c000f6b8>] ret_from_kernel_thread+0x64/0x6c
  [    9.201142] irq event stamp: 160
  [    9.204366] hardirqs last  enabled at (159): [<c043f778>] _raw_spin_unlock_irq+0x30/0x50
  [    9.212469] hardirqs last disabled at (160): [<c000f414>] reenable_mmu+0x30/0x88
  [    9.219867] softirqs last  enabled at (144): [<c002ae5c>] __do_softirq+0x168/0x218
  [    9.227435] softirqs last disabled at (137): [<c002b0d4>] irq_exit+0xa8/0xb4
  [    9.234481]
  [    9.234481] other info that might help us debug this:
  [    9.240995]  Possible unsafe locking scenario:
  [    9.240995]
  [    9.246898]        CPU0
  [    9.249337]        ----
  [    9.251776]   lock(&(&host->lock)->rlock);
  [    9.255878]   <Interrupt>
  [    9.258492]     lock(&(&host->lock)->rlock);
  [    9.262765]
  [    9.262765]  *** DEADLOCK ***
  [    9.262765]
  [    9.268684] no locks held by scsi_eh_1/39.
  [    9.272767]
  [    9.272767] stack backtrace:
  [    9.277117] Call Trace:
  [    9.279589] [cfff9da0] [c0008504] show_stack+0x48/0x150 (unreliable)
  [    9.285972] [cfff9de0] [c0447d5c] print_usage_bug.part.35+0x268/0x27c
  [    9.292425] [cfff9e10] [c006ace4] mark_lock+0x2ac/0x658
  [    9.297660] [cfff9e40] [c006b7e4] __lock_acquire+0x754/0x1840
  [    9.303414] [cfff9ee0] [c006cdb8] lock_acquire+0x90/0xf4
  [    9.308745] [cfff9f20] [c043ef04] _raw_spin_lock+0x34/0x4c
  [    9.314250] [cfff9f30] [c02f4168] sata_fsl_interrupt+0x50/0x250
  [    9.320187] [cfff9f70] [c0079ff0] handle_irq_event_percpu+0x90/0x254
  [    9.326547] [cfff9fc0] [c007a1fc] handle_irq_event+0x48/0x78
  [    9.332220] [cfff9fe0] [c007c95c] handle_level_irq+0x9c/0x104
  [    9.337981] [cfff9ff0] [c000d978] call_handle_irq+0x18/0x28
  [    9.343568] [cc7139f0] [c000608c] do_IRQ+0xf0/0x1a8
  [    9.348464] [cc713a20] [c000fc8c] ret_from_except+0x0/0x14
  [    9.353983] --- Exception: 501 at _raw_spin_unlock_irq+0x40/0x50
  [    9.353983]     LR = _raw_spin_unlock_irq+0x30/0x50
  [    9.364839] [cc713af0] [c043db10] wait_for_common+0xac/0x188
  [    9.370513] [cc713b30] [c02ddee4] ata_exec_internal_sg+0x2b0/0x4f0
  [    9.376699] [cc713be0] [c02de18c] ata_exec_internal+0x68/0xa8
  [    9.382454] [cc713c20] [c02de4b8] ata_dev_read_id+0x158/0x594
  [    9.388205] [cc713ca0] [c02ec244] ata_eh_recover+0xd88/0x13d0
  [    9.393962] [cc713d20] [c02f2520] sata_pmp_error_handler+0xc0/0x8ac
  [    9.400234] [cc713dd0] [c02ecdc8] ata_scsi_port_error_handler+0x464/0x5e8
  [    9.407023] [cc713e10] [c02ecfd0] ata_scsi_error+0x84/0xb8
  [    9.412528] [cc713e40] [c02c4974] scsi_error_handler+0xd8/0x47c
  [    9.418457] [cc713eb0] [c004737c] kthread+0xa8/0xac
  [    9.423355] [cc713f40] [c000f6b8] ret_from_kernel_thread+0x64/0x6c

This fix was suggested by Bhushan Bharat <R65777@freescale.com>, and
was discussed in email at:

  http://linuxppc.10917.n7.nabble.com/MPC8315-reboot-failure-lockdep-splat-possibly-related-tp75162.html

Same patch successfully tested with 3.9.7.  linux-next compiled but
not tested on hardware.

This patch is based off linux-next tag next-20130819
(which is commit 66a01bae29d11916c09f9f5a937cafe7d402e4a5 )

Signed-off-by: Anthony Foiani <anthony.foiani@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
drivers/ata/sata_fsl.c