repair: fix the call to search_rt_dup_extent in scan_bmapbt
PATCH 1/1 repair: fix the call to search_rt_dup_extent in scan_bmapbt
search_rt_dup_extent expects an RT extent number and not a fsbno. Convert the units before the call. Without this we are unlikely to ever found a legit duplicate extent on the RT subvolume because the search will always be off the end.
--- repair/scan.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
repair/scan.c +4 -2
@@ -402,8 +402,10 @@ _("bad state %d, inode %" PRIu64 " bmap block 0x%" PRIx64 "\n"), XFS_FSB_TO_AGBNO(mp, bno) + 1))return (1 ); }else { -if (search_rt_dup_extent (mp, bno)) -return (1 ); +xfs_rtblock_t ext = bno / mp->m_sb.sb_rextsize; + +if (search_rt_dup_extent (mp, ext)) +return 1 ; } } (*tot)++;
Signature · 1 line
2.39.2
Quoted 6 lines · Christoph Hellwig
search_rt_dup_extentexpects an RT extent number and not a fsbno. Convert the units before the call. Without this we are unlikely to ever found a legit duplicate extent on the RT subvolume because the search will always be off the end.Signed-off-by Christoph Hellwig <[email protected]>
Looks familiar! :)
In the longer run: whenever the libxfs 6.7 sync hits the list, I'll be ready to go with a pair of broader patches to fix all the confusing / incorrect units and variable names in xfs_repair. This ought to get merged to xfsprogs 6.6.
--D
Quoted 24 lines
--- repair/scan.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)repair/scan.c +4
-2@@ -402,8 +402,10 @@ _("bad state %d, inode %" PRIu64 " bmap block 0x%" PRIx64 "\n"), XFS_FSB_TO_AGBNO(mp, bno) + 1))return (1 ); }else { -if (search_rt_dup_extent (mp, bno)) -return (1 ); +xfs_rtblock_t ext = bno / mp->m_sb.sb_rextsize; + +if (search_rt_dup_extent (mp, ext)) +return 1 ; } } (*tot)++;Signature · 1 line
2.39.2
Quoted 4 lines · Darrick J. Wong
In the longer run: whenever the libxfs 6.7 sync hits the list, I'll be ready to go with a pair of broader patches to fix all the confusing / incorrect units and variable names in
xfs_repair. This ought to get merged to xfsprogs 6.6.
Yes, please!
FYI, I've done a prototype of annotation the rxtnumber_t with __nocast in the kernel, and except for the ugliness in casting all the 0 values it actually looks pretty nice. I hope with the extra work Luc promised we can actually annotate our non-byte offset/bno/len fields with it and get real type checking for them.