Message ID | 154897668914.26065.4021602625860113731.stgit@magnolia (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Series | xfs: incore unlinked list | expand |
On Thu, Jan 31, 2019 at 03:18:09PM -0800, Darrick J. Wong wrote: > From: Darrick J. Wong <darrick.wong@oracle.com> > > Split the AGI unlinked bucket updates into a separate function. No > functional changes. > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> > --- > fs/xfs/xfs_inode.c | 69 ++++++++++++++++++++++++++++++++++++++-------------- > fs/xfs/xfs_trace.h | 26 ++++++++++++++++++++ > 2 files changed, 76 insertions(+), 19 deletions(-) > > > diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c > index 98355f5f9253..3c33136b21ef 100644 > --- a/fs/xfs/xfs_inode.c > +++ b/fs/xfs/xfs_inode.c > @@ -1880,6 +1880,46 @@ xfs_inactive( > xfs_qm_dqdetach(ip); > } > > +/* > + * Point the AGI unlinked bucket at an inode and log the results. The caller > + * is responsible for validating the old value. > + */ > +STATIC int > +xfs_iunlink_update_bucket( > + struct xfs_trans *tp, > + struct xfs_buf *agibp, > + unsigned int bucket_index, > + xfs_agino_t new_agino) > +{ > + struct xfs_agi *agi = XFS_BUF_TO_AGI(agibp); > + xfs_agino_t old_value; > + int offset; > + > + ASSERT(new_agino == NULLAGINO || > + xfs_verify_agino(tp->t_mountp, be32_to_cpu(agi->agi_seqno), > + new_agino)); > + > + old_value = be32_to_cpu(agi->agi_unlinked[bucket_index]); > + trace_xfs_iunlink_update_bucket(tp->t_mountp, > + be32_to_cpu(agi->agi_seqno), bucket_index, old_value, > + new_agino); > + Might as well pass agno as a param, particularly since the callers already have it. > + /* > + * We should never find the head of the list already set to the value > + * passed in because either we're adding or removing ourselves from the > + * head of the list. > + */ > + if (old_value == new_agino) > + return -EFSCORRUPTED; > + This seems a little weird in the NULLAGINO case. That probably should still never happen, but doesn't seem indicative of corruption on its own. *shrug* not a big deal. Brian > + agi->agi_unlinked[bucket_index] = cpu_to_be32(new_agino); > + offset = offsetof(struct xfs_agi, agi_unlinked) + > + (sizeof(xfs_agino_t) * bucket_index); > + xfs_trans_log_buf(tp, agibp, offset, > + (offset + sizeof(xfs_agino_t) - 1)); > + return 0; > +} > + > /* > * This is called when the inode's link count goes to 0 or we are creating a > * tmpfile via O_TMPFILE. In the case of a tmpfile, @ignore_linkcount will be > @@ -1958,15 +1998,10 @@ xfs_iunlink( > xfs_inobp_check(mp, ibp); > } > > - /* > - * Point the bucket head pointer at the inode being inserted. > - */ > - ASSERT(agino != 0); > - agi->agi_unlinked[bucket_index] = cpu_to_be32(agino); > - offset = offsetof(xfs_agi_t, agi_unlinked) + > - (sizeof(xfs_agino_t) * bucket_index); > - xfs_trans_log_buf(tp, agibp, offset, > - (offset + sizeof(xfs_agino_t) - 1)); > + /* Point the head of the list to point to this inode. */ > + error = xfs_iunlink_update_bucket(tp, agibp, bucket_index, agino); > + if (error) > + goto out_unlock; > pag->pagi_unlinked_count++; > > out_unlock: > @@ -2062,16 +2097,12 @@ xfs_iunlink_remove( > } else { > xfs_trans_brelse(tp, ibp); > } > - /* > - * Point the bucket head pointer at the next inode. > - */ > - ASSERT(next_agino != 0); > - ASSERT(next_agino != agino); > - agi->agi_unlinked[bucket_index] = cpu_to_be32(next_agino); > - offset = offsetof(xfs_agi_t, agi_unlinked) + > - (sizeof(xfs_agino_t) * bucket_index); > - xfs_trans_log_buf(tp, agibp, offset, > - (offset + sizeof(xfs_agino_t) - 1)); > + > + /* Point the head of the list to the next unlinked inode. */ > + error = xfs_iunlink_update_bucket(tp, agibp, bucket_index, > + next_agino); > + if (error) > + goto out_unlock; > } else { > /* > * We need to search the list for the inode being freed. > diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h > index 6fcc893dfc91..c10478e7e49a 100644 > --- a/fs/xfs/xfs_trace.h > +++ b/fs/xfs/xfs_trace.h > @@ -3371,6 +3371,32 @@ DEFINE_TRANS_EVENT(xfs_trans_roll); > DEFINE_TRANS_EVENT(xfs_trans_add_item); > DEFINE_TRANS_EVENT(xfs_trans_free_items); > > +TRACE_EVENT(xfs_iunlink_update_bucket, > + TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, unsigned int bucket, > + xfs_agino_t old_ptr, xfs_agino_t new_ptr), > + TP_ARGS(mp, agno, bucket, old_ptr, new_ptr), > + TP_STRUCT__entry( > + __field(dev_t, dev) > + __field(xfs_agnumber_t, agno) > + __field(unsigned int, bucket) > + __field(xfs_agino_t, old_ptr) > + __field(xfs_agino_t, new_ptr) > + ), > + TP_fast_assign( > + __entry->dev = mp->m_super->s_dev; > + __entry->agno = agno; > + __entry->bucket = bucket; > + __entry->old_ptr = old_ptr; > + __entry->new_ptr = new_ptr; > + ), > + TP_printk("dev %d:%d agno %u bucket %u old 0x%x new 0x%x", > + MAJOR(__entry->dev), MINOR(__entry->dev), > + __entry->agno, > + __entry->bucket, > + __entry->old_ptr, > + __entry->new_ptr) > +); > + > #endif /* _TRACE_XFS_H */ > > #undef TRACE_INCLUDE_PATH >
> + struct xfs_agi *agi = XFS_BUF_TO_AGI(agibp); > + xfs_agino_t old_value; > + int offset; > + > + ASSERT(new_agino == NULLAGINO || > + xfs_verify_agino(tp->t_mountp, be32_to_cpu(agi->agi_seqno), > + new_agino)); > + > + old_value = be32_to_cpu(agi->agi_unlinked[bucket_index]); > + trace_xfs_iunlink_update_bucket(tp->t_mountp, > + be32_to_cpu(agi->agi_seqno), bucket_index, old_value, > + new_agino); As already mentioned by Brian I'd rather pass the agno rather than recalculating it twice. > + agi->agi_unlinked[bucket_index] = cpu_to_be32(new_agino); > + offset = offsetof(struct xfs_agi, agi_unlinked) + > + (sizeof(xfs_agino_t) * bucket_index); > + xfs_trans_log_buf(tp, agibp, offset, > + (offset + sizeof(xfs_agino_t) - 1)); No need for the braces here.
On Fri, Feb 01, 2019 at 02:00:43PM -0500, Brian Foster wrote: > On Thu, Jan 31, 2019 at 03:18:09PM -0800, Darrick J. Wong wrote: > > From: Darrick J. Wong <darrick.wong@oracle.com> > > > > Split the AGI unlinked bucket updates into a separate function. No > > functional changes. > > > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> > > --- > > fs/xfs/xfs_inode.c | 69 ++++++++++++++++++++++++++++++++++++++-------------- > > fs/xfs/xfs_trace.h | 26 ++++++++++++++++++++ > > 2 files changed, 76 insertions(+), 19 deletions(-) > > > > > > diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c > > index 98355f5f9253..3c33136b21ef 100644 > > --- a/fs/xfs/xfs_inode.c > > +++ b/fs/xfs/xfs_inode.c > > @@ -1880,6 +1880,46 @@ xfs_inactive( > > xfs_qm_dqdetach(ip); > > } > > > > +/* > > + * Point the AGI unlinked bucket at an inode and log the results. The caller > > + * is responsible for validating the old value. > > + */ > > +STATIC int > > +xfs_iunlink_update_bucket( > > + struct xfs_trans *tp, > > + struct xfs_buf *agibp, > > + unsigned int bucket_index, > > + xfs_agino_t new_agino) > > +{ > > + struct xfs_agi *agi = XFS_BUF_TO_AGI(agibp); > > + xfs_agino_t old_value; > > + int offset; > > + > > + ASSERT(new_agino == NULLAGINO || > > + xfs_verify_agino(tp->t_mountp, be32_to_cpu(agi->agi_seqno), > > + new_agino)); > > + > > + old_value = be32_to_cpu(agi->agi_unlinked[bucket_index]); > > + trace_xfs_iunlink_update_bucket(tp->t_mountp, > > + be32_to_cpu(agi->agi_seqno), bucket_index, old_value, > > + new_agino); > > + > > Might as well pass agno as a param, particularly since the callers > already have it. Will fix. > > + /* > > + * We should never find the head of the list already set to the value > > + * passed in because either we're adding or removing ourselves from the > > + * head of the list. > > + */ > > + if (old_value == new_agino) > > + return -EFSCORRUPTED; > > + > > This seems a little weird in the NULLAGINO case. That probably should > still never happen, but doesn't seem indicative of corruption on its > own. *shrug* not a big deal. <nod> We shouldn't ever be pulling the last item off an unlinked bucket only to find that the bucket was empty, but if we do then it's a sign that something has gone seriously wrong with our locking or something. --D > Brian > > > + agi->agi_unlinked[bucket_index] = cpu_to_be32(new_agino); > > + offset = offsetof(struct xfs_agi, agi_unlinked) + > > + (sizeof(xfs_agino_t) * bucket_index); > > + xfs_trans_log_buf(tp, agibp, offset, > > + (offset + sizeof(xfs_agino_t) - 1)); > > + return 0; > > +} > > + > > /* > > * This is called when the inode's link count goes to 0 or we are creating a > > * tmpfile via O_TMPFILE. In the case of a tmpfile, @ignore_linkcount will be > > @@ -1958,15 +1998,10 @@ xfs_iunlink( > > xfs_inobp_check(mp, ibp); > > } > > > > - /* > > - * Point the bucket head pointer at the inode being inserted. > > - */ > > - ASSERT(agino != 0); > > - agi->agi_unlinked[bucket_index] = cpu_to_be32(agino); > > - offset = offsetof(xfs_agi_t, agi_unlinked) + > > - (sizeof(xfs_agino_t) * bucket_index); > > - xfs_trans_log_buf(tp, agibp, offset, > > - (offset + sizeof(xfs_agino_t) - 1)); > > + /* Point the head of the list to point to this inode. */ > > + error = xfs_iunlink_update_bucket(tp, agibp, bucket_index, agino); > > + if (error) > > + goto out_unlock; > > pag->pagi_unlinked_count++; > > > > out_unlock: > > @@ -2062,16 +2097,12 @@ xfs_iunlink_remove( > > } else { > > xfs_trans_brelse(tp, ibp); > > } > > - /* > > - * Point the bucket head pointer at the next inode. > > - */ > > - ASSERT(next_agino != 0); > > - ASSERT(next_agino != agino); > > - agi->agi_unlinked[bucket_index] = cpu_to_be32(next_agino); > > - offset = offsetof(xfs_agi_t, agi_unlinked) + > > - (sizeof(xfs_agino_t) * bucket_index); > > - xfs_trans_log_buf(tp, agibp, offset, > > - (offset + sizeof(xfs_agino_t) - 1)); > > + > > + /* Point the head of the list to the next unlinked inode. */ > > + error = xfs_iunlink_update_bucket(tp, agibp, bucket_index, > > + next_agino); > > + if (error) > > + goto out_unlock; > > } else { > > /* > > * We need to search the list for the inode being freed. > > diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h > > index 6fcc893dfc91..c10478e7e49a 100644 > > --- a/fs/xfs/xfs_trace.h > > +++ b/fs/xfs/xfs_trace.h > > @@ -3371,6 +3371,32 @@ DEFINE_TRANS_EVENT(xfs_trans_roll); > > DEFINE_TRANS_EVENT(xfs_trans_add_item); > > DEFINE_TRANS_EVENT(xfs_trans_free_items); > > > > +TRACE_EVENT(xfs_iunlink_update_bucket, > > + TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, unsigned int bucket, > > + xfs_agino_t old_ptr, xfs_agino_t new_ptr), > > + TP_ARGS(mp, agno, bucket, old_ptr, new_ptr), > > + TP_STRUCT__entry( > > + __field(dev_t, dev) > > + __field(xfs_agnumber_t, agno) > > + __field(unsigned int, bucket) > > + __field(xfs_agino_t, old_ptr) > > + __field(xfs_agino_t, new_ptr) > > + ), > > + TP_fast_assign( > > + __entry->dev = mp->m_super->s_dev; > > + __entry->agno = agno; > > + __entry->bucket = bucket; > > + __entry->old_ptr = old_ptr; > > + __entry->new_ptr = new_ptr; > > + ), > > + TP_printk("dev %d:%d agno %u bucket %u old 0x%x new 0x%x", > > + MAJOR(__entry->dev), MINOR(__entry->dev), > > + __entry->agno, > > + __entry->bucket, > > + __entry->old_ptr, > > + __entry->new_ptr) > > +); > > + > > #endif /* _TRACE_XFS_H */ > > > > #undef TRACE_INCLUDE_PATH > >
On Sat, Feb 02, 2019 at 08:21:21AM -0800, Christoph Hellwig wrote: > > + struct xfs_agi *agi = XFS_BUF_TO_AGI(agibp); > > + xfs_agino_t old_value; > > + int offset; > > + > > + ASSERT(new_agino == NULLAGINO || > > + xfs_verify_agino(tp->t_mountp, be32_to_cpu(agi->agi_seqno), > > + new_agino)); > > + > > + old_value = be32_to_cpu(agi->agi_unlinked[bucket_index]); > > + trace_xfs_iunlink_update_bucket(tp->t_mountp, > > + be32_to_cpu(agi->agi_seqno), bucket_index, old_value, > > + new_agino); > > As already mentioned by Brian I'd rather pass the agno rather than > recalculating it twice. > > > + agi->agi_unlinked[bucket_index] = cpu_to_be32(new_agino); > > + offset = offsetof(struct xfs_agi, agi_unlinked) + > > + (sizeof(xfs_agino_t) * bucket_index); > > + xfs_trans_log_buf(tp, agibp, offset, > > + (offset + sizeof(xfs_agino_t) - 1)); > > No need for the braces here. Will fix. --D
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index 98355f5f9253..3c33136b21ef 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -1880,6 +1880,46 @@ xfs_inactive( xfs_qm_dqdetach(ip); } +/* + * Point the AGI unlinked bucket at an inode and log the results. The caller + * is responsible for validating the old value. + */ +STATIC int +xfs_iunlink_update_bucket( + struct xfs_trans *tp, + struct xfs_buf *agibp, + unsigned int bucket_index, + xfs_agino_t new_agino) +{ + struct xfs_agi *agi = XFS_BUF_TO_AGI(agibp); + xfs_agino_t old_value; + int offset; + + ASSERT(new_agino == NULLAGINO || + xfs_verify_agino(tp->t_mountp, be32_to_cpu(agi->agi_seqno), + new_agino)); + + old_value = be32_to_cpu(agi->agi_unlinked[bucket_index]); + trace_xfs_iunlink_update_bucket(tp->t_mountp, + be32_to_cpu(agi->agi_seqno), bucket_index, old_value, + new_agino); + + /* + * We should never find the head of the list already set to the value + * passed in because either we're adding or removing ourselves from the + * head of the list. + */ + if (old_value == new_agino) + return -EFSCORRUPTED; + + agi->agi_unlinked[bucket_index] = cpu_to_be32(new_agino); + offset = offsetof(struct xfs_agi, agi_unlinked) + + (sizeof(xfs_agino_t) * bucket_index); + xfs_trans_log_buf(tp, agibp, offset, + (offset + sizeof(xfs_agino_t) - 1)); + return 0; +} + /* * This is called when the inode's link count goes to 0 or we are creating a * tmpfile via O_TMPFILE. In the case of a tmpfile, @ignore_linkcount will be @@ -1958,15 +1998,10 @@ xfs_iunlink( xfs_inobp_check(mp, ibp); } - /* - * Point the bucket head pointer at the inode being inserted. - */ - ASSERT(agino != 0); - agi->agi_unlinked[bucket_index] = cpu_to_be32(agino); - offset = offsetof(xfs_agi_t, agi_unlinked) + - (sizeof(xfs_agino_t) * bucket_index); - xfs_trans_log_buf(tp, agibp, offset, - (offset + sizeof(xfs_agino_t) - 1)); + /* Point the head of the list to point to this inode. */ + error = xfs_iunlink_update_bucket(tp, agibp, bucket_index, agino); + if (error) + goto out_unlock; pag->pagi_unlinked_count++; out_unlock: @@ -2062,16 +2097,12 @@ xfs_iunlink_remove( } else { xfs_trans_brelse(tp, ibp); } - /* - * Point the bucket head pointer at the next inode. - */ - ASSERT(next_agino != 0); - ASSERT(next_agino != agino); - agi->agi_unlinked[bucket_index] = cpu_to_be32(next_agino); - offset = offsetof(xfs_agi_t, agi_unlinked) + - (sizeof(xfs_agino_t) * bucket_index); - xfs_trans_log_buf(tp, agibp, offset, - (offset + sizeof(xfs_agino_t) - 1)); + + /* Point the head of the list to the next unlinked inode. */ + error = xfs_iunlink_update_bucket(tp, agibp, bucket_index, + next_agino); + if (error) + goto out_unlock; } else { /* * We need to search the list for the inode being freed. diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h index 6fcc893dfc91..c10478e7e49a 100644 --- a/fs/xfs/xfs_trace.h +++ b/fs/xfs/xfs_trace.h @@ -3371,6 +3371,32 @@ DEFINE_TRANS_EVENT(xfs_trans_roll); DEFINE_TRANS_EVENT(xfs_trans_add_item); DEFINE_TRANS_EVENT(xfs_trans_free_items); +TRACE_EVENT(xfs_iunlink_update_bucket, + TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, unsigned int bucket, + xfs_agino_t old_ptr, xfs_agino_t new_ptr), + TP_ARGS(mp, agno, bucket, old_ptr, new_ptr), + TP_STRUCT__entry( + __field(dev_t, dev) + __field(xfs_agnumber_t, agno) + __field(unsigned int, bucket) + __field(xfs_agino_t, old_ptr) + __field(xfs_agino_t, new_ptr) + ), + TP_fast_assign( + __entry->dev = mp->m_super->s_dev; + __entry->agno = agno; + __entry->bucket = bucket; + __entry->old_ptr = old_ptr; + __entry->new_ptr = new_ptr; + ), + TP_printk("dev %d:%d agno %u bucket %u old 0x%x new 0x%x", + MAJOR(__entry->dev), MINOR(__entry->dev), + __entry->agno, + __entry->bucket, + __entry->old_ptr, + __entry->new_ptr) +); + #endif /* _TRACE_XFS_H */ #undef TRACE_INCLUDE_PATH