xfs: always init fdblocks in mount

hulk inclusion
category: bugfix
bugzilla: 31088
CVE: NA

---------------------------

Use fuzz(hydra) to test XFS and automatically generate
tmp.img(XFS v5 format, but some metadata is wrong)

xfs_repair information(just one AG):
agf_freeblks 0, counted 3224 in ag 0
agf_longest 0, counted 3224 in ag 0
sb_fdblocks 3228, counted 3224

Test as follows:
mount tmp.img tmpdir
cp file1M tmpdir
sync

In 4.19-stable, sync will stuck, while in linux-next, sync not stuck.
The reason is same to commit d0c7feaf8767
("xfs: add agf freeblocks verify in xfs_agf_verify"), cause agf_longest
is 0, we can not block this in xfs_agf_verify.

Make sure fdblocks is always inited in mount(also init ifree, icount).

xfs_mountfs
  xfs_check_summary_counts
    xfs_initialize_perag_data
Signed-off-by: NZheng Bin <zhengbin13@huawei.com>
Reviewed-by: NHou Tao <houtao1@huawei.com>
Signed-off-by: NYang Yingliang <yangyingliang@huawei.com>

fs/xfs/libxfs/xfs_ag_resv.c

@@ -309,10 +309,13 @@ xfs_ag_resv_init(
 	}
 
 #ifdef DEBUG
-	/* need to read in the AGF for the ASSERT below to work */
-	error = xfs_alloc_pagf_init(pag->pag_mount, tp, pag->pag_agno, 0);
-	if (error)
-		return error;
+	if (!pag->pagf_init) {
+		/* need to read in the AGF for the ASSERT below to work */
+		error = xfs_alloc_pagf_init(pag->pag_mount, tp,
+					    pag->pag_agno, 0);
+		if (error)
+			return error;
+	}
 
 	ASSERT(xfs_perag_resv(pag, XFS_AG_RESV_METADATA)->ar_reserved +
 	       xfs_perag_resv(pag, XFS_AG_RESV_RMAPBT)->ar_reserved <=

fs/xfs/xfs_mount.c

@@ -624,39 +624,6 @@ xfs_check_summary_counts(
 		return -EFSCORRUPTED;
 	}
 
-	/*
-	 * Now the log is mounted, we know if it was an unclean shutdown or
-	 * not. If it was, with the first phase of recovery has completed, we
-	 * have consistent AG blocks on disk. We have not recovered EFIs yet,
-	 * but they are recovered transactionally in the second recovery phase
-	 * later.
-	 *
-	 * If the log was clean when we mounted, we can check the summary
-	 * counters.  If any of them are obviously incorrect, we can recompute
-	 * them from the AGF headers in the next step.
-	 */
-	if (XFS_LAST_UNMOUNT_WAS_CLEAN(mp) &&
-	    (mp->m_sb.sb_fdblocks > mp->m_sb.sb_dblocks ||
-	     !xfs_verify_icount(mp, mp->m_sb.sb_icount) ||
-	     mp->m_sb.sb_ifree > mp->m_sb.sb_icount))
-		mp->m_flags |= XFS_MOUNT_BAD_SUMMARY;
-
-	/*
-	 * We can safely re-initialise incore superblock counters from the
-	 * per-ag data. These may not be correct if the filesystem was not
-	 * cleanly unmounted, so we waited for recovery to finish before doing
-	 * this.
-	 *
-	 * If the filesystem was cleanly unmounted or the previous check did
-	 * not flag anything weird, then we can trust the values in the
-	 * superblock to be correct and we don't need to do anything here.
-	 * Otherwise, recalculate the summary counters.
-	 */
-	if ((!xfs_sb_version_haslazysbcount(&mp->m_sb) ||
-	     XFS_LAST_UNMOUNT_WAS_CLEAN(mp)) &&
-	    !(mp->m_flags & XFS_MOUNT_BAD_SUMMARY))
-		return 0;
-
 	return xfs_initialize_perag_data(mp, mp->m_sb.sb_agcount);
 }