Re: [PATCH RT 05/12] slab: Prevent local lock deadlock

From
Frank Rowand <hidden>
Date
2012-07-27 00:15:47
Also in
lkml
Reply to
Steven Rostedt, 2012-07-18
On 07/18/12 15:39, Steven Rostedt wrote:
From: Thomas Gleixner <redacted>

On RT we avoid the cross cpu function calls and take the per cpu local
locks instead. Now the code missed that taking the local lock on the
cpu which runs the code must use the proper local lock functions and
not a simple spin_lock(). Otherwise it deadlocks later when trying to
acquire the local lock with the proper function.

Reported-and-tested-by: Chris Pringle <redacted>
Signed-off-by: Thomas Gleixner <redacted>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 mm/slab.c |   26 ++++++++++++++++++++++----
 1 file changed, 22 insertions(+), 4 deletions(-)

This patch leads to a warning during boot on the ARM pandaboard:

[    0.225097] Brought up 2 CPUs
[    0.225097] SMP: Total of 2 processors activated (2007.19 BogoMIPS).
[    0.225952] 
[    0.225982] =============================================
[    0.225982] [ INFO: possible recursive locking detected ]
[    0.225982] 3.0.36-rt58 #1
[    0.225982] ---------------------------------------------
[    0.225982] swapper/0/1 is trying to acquire lock:
[    0.226013]  (&per_cpu(slab_lock, __cpu).lock){+.+...}, at: [<c0147544>] do_ccupdate_local+0x18/0x44
[    0.226043] 
[    0.226043] but task is already holding lock:
[    0.226043]  (&per_cpu(slab_lock, __cpu).lock){+.+...}, at: [<c014737c>] lock_slab_on+0x48/0x134
[    0.226074] 
[    0.226074] other info that might help us debug this:
[    0.226074]  Possible unsafe locking scenario:
[    0.226074] 
[    0.226074]        CPU0
[    0.226074]        ----
[    0.226074]   lock(&per_cpu(slab_lock, __cpu).lock);
[    0.226104]   lock(&per_cpu(slab_lock, __cpu).lock);
[    0.226104] 
[    0.226104]  *** DEADLOCK ***
[    0.226104] 
[    0.226104]  May be due to missing lock nesting notation
[    0.226104] 
[    0.226104] 2 locks held by swapper/0/1:
[    0.226135]  #0:  (cache_chain_mutex){+.+.+.}, at: [<c014a618>] kmem_cache_create+0x74/0x4bc
[    0.226135]  #1:  (&per_cpu(slab_lock, __cpu).lock){+.+...}, at: [<c014737c>] lock_slab_on+0x48/0x134
[    0.226165] 
[    0.226165] stack backtrace:
[    0.226196] [<c00681f8>] (unwind_backtrace+0x0/0xf0) from [<c00da918>] (__lock_acquire+0x1984/0x1ce8)
[    0.226196] [<c00da918>] (__lock_acquire+0x1984/0x1ce8) from [<c00db29c>] (lock_acquire+0x100/0x120)
[    0.226226] [<c00db29c>] (lock_acquire+0x100/0x120) from [<c0485c10>] (rt_spin_lock+0x4c/0x5c)
[    0.226257] [<c0485c10>] (rt_spin_lock+0x4c/0x5c) from [<c0147544>] (do_ccupdate_local+0x18/0x44)
[    0.226257] [<c0147544>] (do_ccupdate_local+0x18/0x44) from [<c01476e8>] (slab_on_each_cpu+0x2c/0x64)
[    0.226287] [<c01476e8>] (slab_on_each_cpu+0x2c/0x64) from [<c0149c70>] (do_tune_cpucache+0xd8/0x3e8)
[    0.226287] [<c0149c70>] (do_tune_cpucache+0xd8/0x3e8) from [<c014a154>] (enable_cpucache+0x50/0xcc)
[    0.226318] [<c014a154>] (enable_cpucache+0x50/0xcc) from [<c014a974>] (kmem_cache_create+0x3d0/0x4bc)
[    0.226318] [<c014a974>] (kmem_cache_create+0x3d0/0x4bc) from [<c0021e54>] (init_tmpfs+0x3c/0xe8)
[    0.226348] [<c0021e54>] (init_tmpfs+0x3c/0xe8) from [<c00083b4>] (kernel_init+0x80/0x150)
[    0.226379] [<c00083b4>] (kernel_init+0x80/0x150) from [<c0061e30>] (kernel_thread_exit+0x0/0x8)
[    0.239776] omap_hwmod: _populate_mpu_rt_base found no _mpu_rt_va for emif_fw
[    0.239776] omap_hwmod: _populate_mpu_rt_base found no _mpu_rt_va for l3_instr



Config is from arch/arm/configs/omap2plus_defconfig
plus:

   CONFIG_USB_EHCI_HCD=y
   CONFIG_USB_NET_SMSC95XX=y
   CONFIG_PREEMPT_RT_FULL=y


-Frank
Thread (25 messages) 25 messages, 3 authors, last reply 2012-07-31
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help