[PATCH v2] xfs: fix spurious spin_is_locked() assert failures on non-smp kernels

[PATCH v2] xfs: fix spurious spin_is_locked() assert failures on non-smp kernels

[Brian Foster]

The 0-day kernel test robot reports assertion failures on
!CONFIG_SMP kernels due to failed spin_is_locked() checks. As it
turns out, spin_is_locked() is hardcoded to return zero on
!CONFIG_SMP kernels and so this function cannot be relied on to
verify spinlock state in this configuration.

To avoid this problem, replace the associated asserts with lockdep
variants that do the right thing regardless of kernel configuration.
Drop the one assert that checks for an unlocked lock as there is no
suitable lockdep variant for that case. This moves the spinlock
checks from XFS debug code to lockdep, but generally provides the
same level of protection.

Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Brian Foster <bfoster@redhat.com>
---

Here's another version that uses lockdep calls as suggested by
Christoph.

Brian

v2:
- Use lockdep asserts instead of config check.
- Drop !spin_is_locked() assert from inode initialization.
v1: http://www.spinics.net/lists/linux-xfs/msg07463.html

 fs/xfs/xfs_buf.c    | 2 +-
 fs/xfs/xfs_icache.c | 5 ++---
 2 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
index 07b77b7..16d6a57 100644
--- a/fs/xfs/xfs_buf.c
+++ b/fs/xfs/xfs_buf.c
@@ -117,7 +117,7 @@ static inline void
 __xfs_buf_ioacct_dec(
 	struct xfs_buf	*bp)
 {
-	ASSERT(spin_is_locked(&bp->b_lock));
+	lockdep_assert_held(&bp->b_lock);
 
 	if (bp->b_state & XFS_BSTATE_IN_FLIGHT) {
 		bp->b_state &= ~XFS_BSTATE_IN_FLIGHT;
diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
index f61c84f8..990210f 100644
--- a/fs/xfs/xfs_icache.c
+++ b/fs/xfs/xfs_icache.c
@@ -66,7 +66,6 @@ xfs_inode_alloc(
 
 	XFS_STATS_INC(mp, vn_active);
 	ASSERT(atomic_read(&ip->i_pincount) == 0);
-	ASSERT(!spin_is_locked(&ip->i_flags_lock));
 	ASSERT(!xfs_isiflocked(ip));
 	ASSERT(ip->i_ino == 0);
 
@@ -190,7 +189,7 @@ xfs_perag_set_reclaim_tag(
 {
 	struct xfs_mount	*mp = pag->pag_mount;
 
-	ASSERT(spin_is_locked(&pag->pag_ici_lock));
+	lockdep_assert_held(&pag->pag_ici_lock);
 	if (pag->pag_ici_reclaimable++)
 		return;
 
@@ -212,7 +211,7 @@ xfs_perag_clear_reclaim_tag(
 {
 	struct xfs_mount	*mp = pag->pag_mount;
 
-	ASSERT(spin_is_locked(&pag->pag_ici_lock));
+	lockdep_assert_held(&pag->pag_ici_lock);
 	if (--pag->pag_ici_reclaimable)
 		return;
 
-- 
2.7.5

Re: [PATCH v2] xfs: fix spurious spin_is_locked() assert failures on non-smp kernels

[Christoph Hellwig]

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v2] xfs: fix spurious spin_is_locked() assert failures on non-smp kernels

[Darrick J. Wong]

On Wed, Jun 07, 2017 at 09:00:55AM -0400, Brian Foster wrote:
> The 0-day kernel test robot reports assertion failures on
> !CONFIG_SMP kernels due to failed spin_is_locked() checks. As it
> turns out, spin_is_locked() is hardcoded to return zero on
> !CONFIG_SMP kernels and so this function cannot be relied on to
> verify spinlock state in this configuration.
> 
> To avoid this problem, replace the associated asserts with lockdep
> variants that do the right thing regardless of kernel configuration.
> Drop the one assert that checks for an unlocked lock as there is no
> suitable lockdep variant for that case. This moves the spinlock
> checks from XFS debug code to lockdep, but generally provides the
> same level of protection.
> 
> Reported-by: kbuild test robot <fengguang.wu@intel.com>
> Signed-off-by: Brian Foster <bfoster@redhat.com>
> ---
> 
> Here's another version that uses lockdep calls as suggested by
> Christoph.

Looks ok, will test:
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>

Hmm.... do you want me to put this into 4.12?  It's sort of a regression
introduced in -rc4, but on the other hand this seems to have been broken
for quite a while for SMP=n && XFS_DEBUG=y and nobody complained...

--D

> 
> Brian
> 
> v2:
> - Use lockdep asserts instead of config check.
> - Drop !spin_is_locked() assert from inode initialization.
> v1: http://www.spinics.net/lists/linux-xfs/msg07463.html
> 
>  fs/xfs/xfs_buf.c    | 2 +-
>  fs/xfs/xfs_icache.c | 5 ++---
>  2 files changed, 3 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
> index 07b77b7..16d6a57 100644
> --- a/fs/xfs/xfs_buf.c
> +++ b/fs/xfs/xfs_buf.c
> @@ -117,7 +117,7 @@ static inline void
>  __xfs_buf_ioacct_dec(
>  	struct xfs_buf	*bp)
>  {
> -	ASSERT(spin_is_locked(&bp->b_lock));
> +	lockdep_assert_held(&bp->b_lock);
>  
>  	if (bp->b_state & XFS_BSTATE_IN_FLIGHT) {
>  		bp->b_state &= ~XFS_BSTATE_IN_FLIGHT;
> diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> index f61c84f8..990210f 100644
> --- a/fs/xfs/xfs_icache.c
> +++ b/fs/xfs/xfs_icache.c
> @@ -66,7 +66,6 @@ xfs_inode_alloc(
>  
>  	XFS_STATS_INC(mp, vn_active);
>  	ASSERT(atomic_read(&ip->i_pincount) == 0);
> -	ASSERT(!spin_is_locked(&ip->i_flags_lock));
>  	ASSERT(!xfs_isiflocked(ip));
>  	ASSERT(ip->i_ino == 0);
>  
> @@ -190,7 +189,7 @@ xfs_perag_set_reclaim_tag(
>  {
>  	struct xfs_mount	*mp = pag->pag_mount;
>  
> -	ASSERT(spin_is_locked(&pag->pag_ici_lock));
> +	lockdep_assert_held(&pag->pag_ici_lock);
>  	if (pag->pag_ici_reclaimable++)
>  		return;
>  
> @@ -212,7 +211,7 @@ xfs_perag_clear_reclaim_tag(
>  {
>  	struct xfs_mount	*mp = pag->pag_mount;
>  
> -	ASSERT(spin_is_locked(&pag->pag_ici_lock));
> +	lockdep_assert_held(&pag->pag_ici_lock);
>  	if (--pag->pag_ici_reclaimable)
>  		return;
>  
> -- 
> 2.7.5
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2] xfs: fix spurious spin_is_locked() assert failures on non-smp kernels

[Brian Foster]

On Thu, Jun 08, 2017 at 08:26:07AM -0700, Darrick J. Wong wrote:
> On Wed, Jun 07, 2017 at 09:00:55AM -0400, Brian Foster wrote:
> > The 0-day kernel test robot reports assertion failures on
> > !CONFIG_SMP kernels due to failed spin_is_locked() checks. As it
> > turns out, spin_is_locked() is hardcoded to return zero on
> > !CONFIG_SMP kernels and so this function cannot be relied on to
> > verify spinlock state in this configuration.
> > 
> > To avoid this problem, replace the associated asserts with lockdep
> > variants that do the right thing regardless of kernel configuration.
> > Drop the one assert that checks for an unlocked lock as there is no
> > suitable lockdep variant for that case. This moves the spinlock
> > checks from XFS debug code to lockdep, but generally provides the
> > same level of protection.
> > 
> > Reported-by: kbuild test robot <fengguang.wu@intel.com>
> > Signed-off-by: Brian Foster <bfoster@redhat.com>
> > ---
> > 
> > Here's another version that uses lockdep calls as suggested by
> > Christoph.
> 
> Looks ok, will test:
> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Hmm.... do you want me to put this into 4.12?  It's sort of a regression
> introduced in -rc4, but on the other hand this seems to have been broken
> for quite a while for SMP=n && XFS_DEBUG=y and nobody complained...
> 

I don't have a major preference either way. I generally agree that the
issue is old and that the recent patch just widened the scope slightly
such that the 0-day test caught it, so I'm fine with deferring it to
next from a technical perspective.

I'm actually not sure if the 0-day test thing is going to continue to
complain about the issue on subsequent merges or -rc drops, or if it's
just a one time informational thing..? If the former I suppose it might
make sense to drop this into 4.12 to quiet the tests and fix the
"regression." If the latter, perhaps just defer it..?

Thanks for the review..

Brian

> --D
> 
> > 
> > Brian
> > 
> > v2:
> > - Use lockdep asserts instead of config check.
> > - Drop !spin_is_locked() assert from inode initialization.
> > v1: http://www.spinics.net/lists/linux-xfs/msg07463.html
> > 
> >  fs/xfs/xfs_buf.c    | 2 +-
> >  fs/xfs/xfs_icache.c | 5 ++---
> >  2 files changed, 3 insertions(+), 4 deletions(-)
> > 
> > diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c
> > index 07b77b7..16d6a57 100644
> > --- a/fs/xfs/xfs_buf.c
> > +++ b/fs/xfs/xfs_buf.c
> > @@ -117,7 +117,7 @@ static inline void
> >  __xfs_buf_ioacct_dec(
> >  	struct xfs_buf	*bp)
> >  {
> > -	ASSERT(spin_is_locked(&bp->b_lock));
> > +	lockdep_assert_held(&bp->b_lock);
> >  
> >  	if (bp->b_state & XFS_BSTATE_IN_FLIGHT) {
> >  		bp->b_state &= ~XFS_BSTATE_IN_FLIGHT;
> > diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> > index f61c84f8..990210f 100644
> > --- a/fs/xfs/xfs_icache.c
> > +++ b/fs/xfs/xfs_icache.c
> > @@ -66,7 +66,6 @@ xfs_inode_alloc(
> >  
> >  	XFS_STATS_INC(mp, vn_active);
> >  	ASSERT(atomic_read(&ip->i_pincount) == 0);
> > -	ASSERT(!spin_is_locked(&ip->i_flags_lock));
> >  	ASSERT(!xfs_isiflocked(ip));
> >  	ASSERT(ip->i_ino == 0);
> >  
> > @@ -190,7 +189,7 @@ xfs_perag_set_reclaim_tag(
> >  {
> >  	struct xfs_mount	*mp = pag->pag_mount;
> >  
> > -	ASSERT(spin_is_locked(&pag->pag_ici_lock));
> > +	lockdep_assert_held(&pag->pag_ici_lock);
> >  	if (pag->pag_ici_reclaimable++)
> >  		return;
> >  
> > @@ -212,7 +211,7 @@ xfs_perag_clear_reclaim_tag(
> >  {
> >  	struct xfs_mount	*mp = pag->pag_mount;
> >  
> > -	ASSERT(spin_is_locked(&pag->pag_ici_lock));
> > +	lockdep_assert_held(&pag->pag_ici_lock);
> >  	if (--pag->pag_ici_reclaimable)
> >  		return;
> >  
> > -- 
> > 2.7.5
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2] xfs: fix spurious spin_is_locked() assert failures on non-smp kernels

[Christoph Hellwig]

On Thu, Jun 08, 2017 at 08:26:07AM -0700, Darrick J. Wong wrote:
> Hmm.... do you want me to put this into 4.12?  It's sort of a regression
> introduced in -rc4, but on the other hand this seems to have been broken
> for quite a while for SMP=n && XFS_DEBUG=y and nobody complained...

Please add it to 4.12.  It's harmless, and the thing that triggered it
only was merged in 4.12-rc.