[PATCH] net/irda: fix lockdep annotation

[PATCH] net/irda: fix lockdep annotation

[Dmitry Vyukov]

The current annotation uses a global variable as recursion counter.
The variable is not atomic nor protected with a mutex, but mutated
by multiple threads. This causes lockdep bug reports episodically:

BUG: looking up invalid subclass: 4294967295
...
_raw_spin_lock_irqsave_nested+0x120/0x180
hashbin_delete+0x4fe/0x750
__irias_delete_object+0xab/0x170
irias_delete_object+0x5f/0xc0
ircomm_tty_detach_cable+0x1d5/0x3f0
...

Make the hashbin_lock_depth variable atomic to prevent bug reports.

Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Samuel Ortiz <samuel@sortiz.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Fixes: c7630a4b932af ("[IrDA]: irda lockdep annotation")
---
 net/irda/irqueue.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/net/irda/irqueue.c b/net/irda/irqueue.c
index acbe61c7e683..b9fd74e6ca99 100644
--- a/net/irda/irqueue.c
+++ b/net/irda/irqueue.c
@@ -384,21 +384,23 @@ EXPORT_SYMBOL(hashbin_new);
  *    just supply kfree, which should take care of the job.
  */
 #ifdef CONFIG_LOCKDEP
-static int hashbin_lock_depth = 0;
+static atomic_t hashbin_lock_depth = ATOMIC_INIT(0);
 #endif
 int hashbin_delete( hashbin_t* hashbin, FREE_FUNC free_func)
 {
 	irda_queue_t* queue;
 	unsigned long flags = 0;
-	int i;
+	int i, depth = 0;
 
 	IRDA_ASSERT(hashbin != NULL, return -1;);
 	IRDA_ASSERT(hashbin->magic == HB_MAGIC, return -1;);
 
 	/* Synchronize */
 	if ( hashbin->hb_type & HB_LOCK ) {
-		spin_lock_irqsave_nested(&hashbin->hb_spinlock, flags,
-					 hashbin_lock_depth++);
+#ifdef CONFIG_LOCKDEP
+		depth = atomic_inc_return(&hashbin_lock_depth) - 1;
+#endif
+		spin_lock_irqsave_nested(&hashbin->hb_spinlock, flags, depth);
 	}
 
 	/*
@@ -423,7 +425,7 @@ int hashbin_delete( hashbin_t* hashbin, FREE_FUNC free_func)
 	if ( hashbin->hb_type & HB_LOCK) {
 		spin_unlock_irqrestore(&hashbin->hb_spinlock, flags);
 #ifdef CONFIG_LOCKDEP
-		hashbin_lock_depth--;
+		atomic_dec(&hashbin_lock_depth);
 #endif
 	}
 
-- 
2.11.0.483.g087da7b7c-goog

Re: [PATCH] net/irda: fix lockdep annotation

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 1852 bytes --]

Hi Dmitry,

[auto build test WARNING on net-next/master]
[also build test WARNING on v4.10-rc4 next-20170116]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url:    https://github.com/0day-ci/linux/commits/Dmitry-Vyukov/net-irda-fix-lockdep-annotation/20170116-183405
config: x86_64-randconfig-s2-01161850 (attached as .config)
compiler: gcc-4.4 (Debian 4.4.7-8) 4.4.7
reproduce:
        # save the attached .config to linux build tree
        make ARCH=x86_64 

All warnings (new ones prefixed by >>):

   net/irda/irqueue.c: In function 'hashbin_delete':
>> net/irda/irqueue.c:393: warning: unused variable 'depth'
   At top level:
   cc1: warning: unrecognized command line option "-Wno-maybe-uninitialized"

vim +/depth +393 net/irda/irqueue.c

   377	
   378	
   379	/*
   380	 * Function hashbin_delete (hashbin, free_func)
   381	 *
   382	 *    Destroy hashbin, the free_func can be a user supplied special routine
   383	 *    for deallocating this structure if it's complex. If not the user can
   384	 *    just supply kfree, which should take care of the job.
   385	 */
   386	#ifdef CONFIG_LOCKDEP
   387	static atomic_t hashbin_lock_depth = ATOMIC_INIT(0);
   388	#endif
   389	int hashbin_delete( hashbin_t* hashbin, FREE_FUNC free_func)
   390	{
   391		irda_queue_t* queue;
   392		unsigned long flags = 0;
 > 393		int i, depth = 0;
   394	
   395		IRDA_ASSERT(hashbin != NULL, return -1;);
   396		IRDA_ASSERT(hashbin->magic == HB_MAGIC, return -1;);
   397	
   398		/* Synchronize */
   399		if ( hashbin->hb_type & HB_LOCK ) {
   400	#ifdef CONFIG_LOCKDEP
   401			depth = atomic_inc_return(&hashbin_lock_depth) - 1;

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 25885 bytes --]

Re: [PATCH] net/irda: fix lockdep annotation

[Dmitry Vyukov]

On Mon, Jan 16, 2017 at 3:11 PM, kbuild test robot <lkp@intel.com> wrote:
> Hi Dmitry,
>
> [auto build test WARNING on net-next/master]
> [also build test WARNING on v4.10-rc4 next-20170116]
> [if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
>
> url:    https://github.com/0day-ci/linux/commits/Dmitry-Vyukov/net-irda-fix-lockdep-annotation/20170116-183405
> config: x86_64-randconfig-s2-01161850 (attached as .config)
> compiler: gcc-4.4 (Debian 4.4.7-8) 4.4.7
> reproduce:
>         # save the attached .config to linux build tree
>         make ARCH=x86_64
>
> All warnings (new ones prefixed by >>):
>
>    net/irda/irqueue.c: In function 'hashbin_delete':
>>> net/irda/irqueue.c:393: warning: unused variable 'depth'


Mailed v2 with a fix for the warning.

Thanks, 0-DAY!


>    At top level:
>    cc1: warning: unrecognized command line option "-Wno-maybe-uninitialized"
>
> vim +/depth +393 net/irda/irqueue.c
>
>    377
>    378
>    379  /*
>    380   * Function hashbin_delete (hashbin, free_func)
>    381   *
>    382   *    Destroy hashbin, the free_func can be a user supplied special routine
>    383   *    for deallocating this structure if it's complex. If not the user can
>    384   *    just supply kfree, which should take care of the job.
>    385   */
>    386  #ifdef CONFIG_LOCKDEP
>    387  static atomic_t hashbin_lock_depth = ATOMIC_INIT(0);
>    388  #endif
>    389  int hashbin_delete( hashbin_t* hashbin, FREE_FUNC free_func)
>    390  {
>    391          irda_queue_t* queue;
>    392          unsigned long flags = 0;
>  > 393          int i, depth = 0;
>    394
>    395          IRDA_ASSERT(hashbin != NULL, return -1;);
>    396          IRDA_ASSERT(hashbin->magic == HB_MAGIC, return -1;);
>    397
>    398          /* Synchronize */
>    399          if ( hashbin->hb_type & HB_LOCK ) {
>    400  #ifdef CONFIG_LOCKDEP
>    401                  depth = atomic_inc_return(&hashbin_lock_depth) - 1;
>
> ---
> 0-DAY kernel test infrastructure                Open Source Technology Center
> https://lists.01.org/pipermail/kbuild-all                   Intel Corporation