xfs: don't commit sunit/swidth updates to disk if that would cause repair failures
PATCH v2 1/1 xfs: don't commit sunit/swidth updates to disk if that would cause repair failures
From: Darrick J. Wong <[email protected]>
Alex Lyakas reported[1] that mounting an xfs filesystem with new sunit and swidth values could cause xfs_repair to fail loudly. The problem here is that repair calculates the where mkfs should have allocated the root inode, based on the superblock geometry. The allocation decisions depend on sunit, which means that we really can't go updating sunit if it would lead to a subsequent repair failure on an otherwise correct filesystem.
Port from xfs_repair some code that computes the location of the root inode and teach mount to skip the ondisk update if it would cause problems for repair. Along the way we'll update the documentation, provide a function for computing the minimum AGFL size instead of open-coding it, and cut down some indenting in the mount code.
Note that we allow the mount to proceed (and new allocations will reflect this new geometry) because we've never screened this kind of thing before. We'll have to wait for a new future incompat feature to enforce correct behavior, alas.
Note that the geometry reporting always uses the superblock values, not the incore ones, so that is what xfs_info and xfs_growfs will report.
---
v2: refactor the agfl length calculations, clarify the fsgeometry ioctl behavior, fix a bunch of the comments and make it clearer how we compute the rootino location
--- fs/xfs/libxfs/xfs_alloc.c | 18 ++++++--- fs/xfs/libxfs/xfs_ialloc.c | 70 ++++++++++++++++++++++++++++++++++ fs/xfs/libxfs/xfs_ialloc.h | 1 fs/xfs/xfs_mount.c | 90 ++++++++++++++++++++++++++++++-------------- fs/xfs/xfs_trace.h | 21 ++++++++++ 5 files changed, 166 insertions(+), 34 deletions(-)
fs/xfs/libxfs/xfs_alloc.c +13 -5
@@ -2248,24 +2248,32 @@ xfs_alloc_longest_free_extent(return pag->pagf_flcount >0 || pag->pagf_longest >0 ; } +/* +* Compute the minimum length of the AGFL in the given AG. If @pag is NULL, +* return the largest possible minimum length. +*/ unsigned int xfs_alloc_min_freelist (struct xfs_mount *mp,struct xfs_perag *pag) { +/* AG btrees have at least 1 level. */ +static const uint8_t fake_levels[XFS_BTNUM_AGF ] = {1 ,1 ,1 }; +const uint8_t *levels = pag ? pag->pagf_levels : fake_levels;unsigned int min_free; +ASSERT (mp->m_ag_maxlevels >0 ); +/* space needed by-bno freespace btree */ - min_free = min_t(unsigned int, pag->pagf_levels[XFS_BTNUM_BNOi] + 1, + min_free = min_t(unsigned int, levels[XFS_BTNUM_BNOi] + 1, mp->m_ag_maxlevels);/* space needed by-size freespace btree */ - min_free += min_t(unsigned int, pag->pagf_levels[XFS_BTNUM_CNTi] + 1, + min_free += min_t(unsigned int, levels[XFS_BTNUM_CNTi] + 1, mp->m_ag_maxlevels);/* space needed reverse mapping used space btree */ if (xfs_sb_version_hasrmapbt (&mp->m_sb)) - min_free += min_t(unsigned int, - pag->pagf_levels[XFS_BTNUM_RMAPi] +1 , - mp->m_rmap_maxlevels); + min_free += min_t(unsigned int, levels[XFS_BTNUM_RMAPi] + 1, + mp->m_rmap_maxlevels);return min_free; }
fs/xfs/libxfs/xfs_ialloc.c +70 -0
73 lines not on this page · raw message
fs/xfs/libxfs/xfs_ialloc.h +1 -0
@@ -152,5 +152,6 @@ int xfs_inobt_insert_rec(struct xfs_btree_cur *cur, uint16_t holemask,int xfs_ialloc_cluster_alignment (struct xfs_mount *mp);void xfs_ialloc_setup_geometry (struct xfs_mount *mp); +xfs_ino_t xfs_ialloc_calc_rootino (struct xfs_mount *mp,int sunit);#endif /* __XFS_IALLOC_H__ */
fs/xfs/xfs_mount.c +61 -29
131 lines not on this page · raw message
fs/xfs/xfs_trace.h +21 -0
@@ -3573,6 +3573,27 @@ DEFINE_KMEM_EVENT(kmem_alloc_large);DEFINE_KMEM_EVENT (kmem_realloc);DEFINE_KMEM_EVENT (kmem_zone_alloc); +TRACE_EVENT (xfs_check_new_dalign, + TP_PROTO(struct xfs_mount *mp, int new_dalign, xfs_ino_t calc_rootino), +TP_ARGS (mp, new_dalign, calc_rootino), +TP_STRUCT__entry ( + __field(dev_t, dev) + __field(int, new_dalign) + __field(xfs_ino_t, sb_rootino) +__field (xfs_ino_t, calc_rootino) + ), +TP_fast_assign ( + __entry->dev = mp->m_super->s_dev; + __entry->new_dalign = new_dalign; + __entry->sb_rootino = mp->m_sb.sb_rootino; + __entry->calc_rootino = calc_rootino; + ), +TP_printk ("dev %d:%d new_dalign %d sb_rootino %llu calc_rootino %llu" , +MAJOR (__entry->dev),MINOR (__entry->dev), + __entry->new_dalign, __entry->sb_rootino, + __entry->calc_rootino) +) +#endif /* _TRACE_XFS_H */ #undef TRACE_INCLUDE_PATH
Quoted 40 lines · Darrick J. Wong
From: Darrick J. Wong <[email protected]>
Alex Lyakas reported[1] that mounting an xfs filesystem with new sunit and swidth values could cause
xfs_repairto fail loudly. The problem here is that repair calculates the where mkfs should have allocated the root inode, based on the superblock geometry. The allocation decisions depend on sunit, which means that we really can't go updating sunit if it would lead to a subsequent repair failure on an otherwise correct filesystem.Port from
xfs_repairsome code that computes the location of the root inode and teach mount to skip the ondisk update if it would cause problems for repair. Along the way we'll update the documentation, provide a function for computing the minimum AGFL size instead of open-coding it, and cut down some indenting in the mount code.Note that we allow the mount to proceed (and new allocations will reflect this new geometry) because we've never screened this kind of thing before. We'll have to wait for a new future incompat feature to enforce correct behavior, alas.
Note that the geometry reporting always uses the superblock values, not the incore ones, so that is what
xfs_infoandxfs_growfswill report.Reported-by Alex Lyakas <[email protected]>Signed-off-by Darrick J. Wong <[email protected]>---
v2: refactor the agfl length calculations, clarify the fsgeometry ioctl behavior, fix a bunch of the comments and make it clearer how we compute the rootino location--- fs/xfs/libxfs/xfs_alloc.c | 18 ++++++--- fs/xfs/libxfs/xfs_ialloc.c | 70 ++++++++++++++++++++++++++++++++++ fs/xfs/libxfs/xfs_ialloc.h | 1 fs/xfs/xfs_mount.c | 90 ++++++++++++++++++++++++++++++-------------- fs/xfs/xfs_trace.h | 21 ++++++++++ 5 files changed, 166 insertions(+), 34 deletions(-)
...
Quoted 4 lines
fs/xfs/libxfs/xfs_ialloc.h +0
-0
...
Quoted 24 lines
fs/xfs/libxfs/xfs_ialloc.h +2
-14@@ -398,28 +431,26 @@ xfs_update_alignment(xfs_mount_t *mp) } } -/* -* Update superblock with new values -* and log changes -*/ -if (xfs_sb_version_hasdalign (sbp)) { -if (sbp->sb_unit != mp->m_dalign) { - sbp->sb_unit = mp->m_dalign; - mp->m_update_sb = true; - } -if (sbp->sb_width != mp->m_swidth) { - sbp->sb_width = mp->m_swidth; - mp->m_update_sb = true; - } - }else { +/* Update superblock with new values and log changes. */ +if (!xfs_sb_version_hasdalign (sbp)) {xfs_warn (mp,"cannot change alignment: superblock does not support data alignment" );return -EINVAL ; }
fs/xfs/libxfs/xfs_ialloc.h xfs_update_alignment(xfs_mount_t *mp) · line 449
Quoted 10 lines
fs/xfs/libxfs/xfs_ialloc.h +10
-0+ +if (sbp->sb_unit == mp->m_dalign && + sbp->sb_width == mp->m_swidth) +return 0 ; + +xfs_check_new_dalign (mp, mp->m_dalign); + + sbp->sb_unit = mp->m_dalign; + sbp->sb_width = mp->m_swidth; + mp->m_update_sb = true;
Isn't this supposed to conditionally update the superblock based on the rootino calculation?
Brian
Quoted 77 lines
fs/xfs/libxfs/xfs_ialloc.h +13
-12}else if ((mp->m_flags &XFS_MOUNT_NOALIGN ) !=XFS_MOUNT_NOALIGN &&xfs_sb_version_hasdalign (&mp->m_sb)) { - mp->m_dalign = sbp->sb_unit; - mp->m_swidth = sbp->sb_width; + mp->m_dalign = sbp->sb_unit; + mp->m_swidth = sbp->sb_width; }return 0 ;@@ -647,16 +678,6 @@ xfs_mountfs( mp->m_update_sb = true; } -/* -* Check if sb_agblocks is aligned at stripe boundary -* If sb_agblocks is NOT aligned turn off m_dalign since -* allocator alignment is within an ag, therefore ag has -* to be aligned at stripe boundary. -*/ - error =xfs_update_alignment (mp); -if (error) - goto out; -xfs_alloc_compute_maxlevels (mp);xfs_bmap_compute_maxlevels (mp,XFS_DATA_FORK );xfs_bmap_compute_maxlevels (mp,XFS_ATTR_FORK );@@ -664,6 +685,17 @@ xfs_mountfs(xfs_rmapbt_compute_maxlevels (mp);xfs_refcountbt_compute_maxlevels (mp); +/* +* Check if sb_agblocks is aligned at stripe boundary. If sb_agblocks +* is NOT aligned turn off m_dalign since allocator alignment is within +* an ag, therefore ag has to be aligned at stripe boundary. Note that +* we must compute the free space and rmap btree geometry before doing +* this. +*/ + error =xfs_update_alignment (mp); +if (error) + goto out; +/* enable fail_at_unmount as default */ mp->m_fail_unmount = true;fs/xfs/xfs_trace.h +21
-0@@ -3573,6 +3573,27 @@ DEFINE_KMEM_EVENT(kmem_alloc_large);DEFINE_KMEM_EVENT (kmem_realloc);DEFINE_KMEM_EVENT (kmem_zone_alloc); +TRACE_EVENT (xfs_check_new_dalign, + TP_PROTO(struct xfs_mount *mp, int new_dalign, xfs_ino_t calc_rootino), +TP_ARGS (mp, new_dalign, calc_rootino), +TP_STRUCT__entry ( + __field(dev_t, dev) + __field(int, new_dalign) + __field(xfs_ino_t, sb_rootino) +__field (xfs_ino_t, calc_rootino) + ), +TP_fast_assign ( + __entry->dev = mp->m_super->s_dev; + __entry->new_dalign = new_dalign; + __entry->sb_rootino = mp->m_sb.sb_rootino; + __entry->calc_rootino = calc_rootino; + ), +TP_printk ("dev %d:%d new_dalign %d sb_rootino %llu calc_rootino %llu" , +MAJOR (__entry->dev),MINOR (__entry->dev), + __entry->new_dalign, __entry->sb_rootino, + __entry->calc_rootino) +) +#endif /* _TRACE_XFS_H */ #undef TRACE_INCLUDE_PATH
Quoted 76 lines · Brian Foster
Quoted 40 lines · Darrick J. Wong
From: Darrick J. Wong <[email protected]>
Alex Lyakas reported[1] that mounting an xfs filesystem with new sunit and swidth values could cause
xfs_repairto fail loudly. The problem here is that repair calculates the where mkfs should have allocated the root inode, based on the superblock geometry. The allocation decisions depend on sunit, which means that we really can't go updating sunit if it would lead to a subsequent repair failure on an otherwise correct filesystem.Port from
xfs_repairsome code that computes the location of the root inode and teach mount to skip the ondisk update if it would cause problems for repair. Along the way we'll update the documentation, provide a function for computing the minimum AGFL size instead of open-coding it, and cut down some indenting in the mount code.Note that we allow the mount to proceed (and new allocations will reflect this new geometry) because we've never screened this kind of thing before. We'll have to wait for a new future incompat feature to enforce correct behavior, alas.
Note that the geometry reporting always uses the superblock values, not the incore ones, so that is what
xfs_infoandxfs_growfswill report.Reported-by Alex Lyakas <[email protected]>Signed-off-by Darrick J. Wong <[email protected]>---
v2: refactor the agfl length calculations, clarify the fsgeometry ioctl behavior, fix a bunch of the comments and make it clearer how we compute the rootino location--- fs/xfs/libxfs/xfs_alloc.c | 18 ++++++--- fs/xfs/libxfs/xfs_ialloc.c | 70 ++++++++++++++++++++++++++++++++++ fs/xfs/libxfs/xfs_ialloc.h | 1 fs/xfs/xfs_mount.c | 90 ++++++++++++++++++++++++++++++-------------- fs/xfs/xfs_trace.h | 21 ++++++++++ 5 files changed, 166 insertions(+), 34 deletions(-)...
Quoted 4 lines
fs/xfs/libxfs/xfs_ialloc.h +0
-0...
Quoted 29 lines
fs/xfs/libxfs/xfs_ialloc.h +7
-14@@ -398,28 +431,26 @@ xfs_update_alignment(xfs_mount_t *mp) } } -/* -* Update superblock with new values -* and log changes -*/ -if (xfs_sb_version_hasdalign (sbp)) { -if (sbp->sb_unit != mp->m_dalign) { - sbp->sb_unit = mp->m_dalign; - mp->m_update_sb = true; - } -if (sbp->sb_width != mp->m_swidth) { - sbp->sb_width = mp->m_swidth; - mp->m_update_sb = true; - } - }else { +/* Update superblock with new values and log changes. */ +if (!xfs_sb_version_hasdalign (sbp)) {xfs_warn (mp,"cannot change alignment: superblock does not support data alignment" );return -EINVAL ; } + +if (sbp->sb_unit == mp->m_dalign && + sbp->sb_width == mp->m_swidth) +return 0 ; +
Quoted 8 lines
fs/xfs/libxfs/xfs_ialloc.h xfs_update_alignment(xfs_mount_t *mp) · line 449
Quoted 5 lines
fs/xfs/libxfs/xfs_ialloc.h +5
-0+xfs_check_new_dalign (mp, mp->m_dalign); + + sbp->sb_unit = mp->m_dalign; + sbp->sb_width = mp->m_swidth; + mp->m_update_sb = true;Isn't this supposed to conditionally update the superblock based on the rootino calculation?
D'oh. V3 it is then. :(
--D
Quoted 80 lines
Brian
Quoted 77 lines
fs/xfs/libxfs/xfs_ialloc.h +13
-12}else if ((mp->m_flags &XFS_MOUNT_NOALIGN ) !=XFS_MOUNT_NOALIGN &&xfs_sb_version_hasdalign (&mp->m_sb)) { - mp->m_dalign = sbp->sb_unit; - mp->m_swidth = sbp->sb_width; + mp->m_dalign = sbp->sb_unit; + mp->m_swidth = sbp->sb_width; }return 0 ;@@ -647,16 +678,6 @@ xfs_mountfs( mp->m_update_sb = true; } -/* -* Check if sb_agblocks is aligned at stripe boundary -* If sb_agblocks is NOT aligned turn off m_dalign since -* allocator alignment is within an ag, therefore ag has -* to be aligned at stripe boundary. -*/ - error =xfs_update_alignment (mp); -if (error) - goto out; -xfs_alloc_compute_maxlevels (mp);xfs_bmap_compute_maxlevels (mp,XFS_DATA_FORK );xfs_bmap_compute_maxlevels (mp,XFS_ATTR_FORK );@@ -664,6 +685,17 @@ xfs_mountfs(xfs_rmapbt_compute_maxlevels (mp);xfs_refcountbt_compute_maxlevels (mp); +/* +* Check if sb_agblocks is aligned at stripe boundary. If sb_agblocks +* is NOT aligned turn off m_dalign since allocator alignment is within +* an ag, therefore ag has to be aligned at stripe boundary. Note that +* we must compute the free space and rmap btree geometry before doing +* this. +*/ + error =xfs_update_alignment (mp); +if (error) + goto out; +/* enable fail_at_unmount as default */ mp->m_fail_unmount = true;fs/xfs/xfs_trace.h +21
-0@@ -3573,6 +3573,27 @@ DEFINE_KMEM_EVENT(kmem_alloc_large);DEFINE_KMEM_EVENT (kmem_realloc);DEFINE_KMEM_EVENT (kmem_zone_alloc); +TRACE_EVENT (xfs_check_new_dalign, + TP_PROTO(struct xfs_mount *mp, int new_dalign, xfs_ino_t calc_rootino), +TP_ARGS (mp, new_dalign, calc_rootino), +TP_STRUCT__entry ( + __field(dev_t, dev) + __field(int, new_dalign) + __field(xfs_ino_t, sb_rootino) +__field (xfs_ino_t, calc_rootino) + ), +TP_fast_assign ( + __entry->dev = mp->m_super->s_dev; + __entry->new_dalign = new_dalign; + __entry->sb_rootino = mp->m_sb.sb_rootino; + __entry->calc_rootino = calc_rootino; + ), +TP_printk ("dev %d:%d new_dalign %d sb_rootino %llu calc_rootino %llu" , +MAJOR (__entry->dev),MINOR (__entry->dev), + __entry->new_dalign, __entry->sb_rootino, + __entry->calc_rootino) +) +#endif /* _TRACE_XFS_H */ #undef TRACE_INCLUDE_PATH