From: Zizhi Wo wozizhi@huawei.com
hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/IAC63K CVE: NA
--------------------------------
A concurrent file creation and little writing could unexpectedly return -ENOSPC error since there is a race window that the allocator could get the wrong agf->agf_longest.
Write file process steps: 1) Find the entry that best meets the conditions, then calculate the start address and length of the remaining part of the entry after allocation. 2) Delete this entry and update the agf->agf_longest. 3) Insert the remaining unused parts of this entry based on the calculations in 1), and update the agf->agf_longest again if necessary.
Create file process steps: 1) Check whether there are free inodes in the inode chunk. 2) If there is no free inode, check whether there has space for creating inode chunks, perform the no-lock judgment first. 3) If the judgment succeeds, the judgment is performed again with agf lock held. Otherwire, an error is returned directly.
If the write process is in step 2) but not go to 3) yet, the create file process goes to 2) at this time, it will be mistaken for no space, resulting in the file system still has space but the file creation fails. Because in the previous code, if numrec == 0, longest will be temporarily set to 0 in write process step 2), and we have fixed it in commit 628ab796d8b1 ("xfs: Fix file creation failure") to update the longest in advance.
But we don't fix it all. If numrec is not 0 in xfs_allocbt_update_lastrec, xfs will update the agf_longest to the -current- longest node currently. However, this is not true because the value of the node may be smaller than the remaining part of the original longest extent after part of it is deleted. The agf_longest updated at this moment is not accurate.
Fix it by comparing cur->bc_free_longest with the -current- longest node and taking the maximum value as the agf_longest.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Fixes: 628ab796d8b1 ("xfs: Fix file creation failure") Signed-off-by: Zizhi Wo wozizhi@huawei.com Signed-off-by: Long Li leo.lilong@huawei.com --- fs/xfs/libxfs/xfs_alloc_btree.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-)
diff --git a/fs/xfs/libxfs/xfs_alloc_btree.c b/fs/xfs/libxfs/xfs_alloc_btree.c index d6dbbe85e819..8282a171131e 100644 --- a/fs/xfs/libxfs/xfs_alloc_btree.c +++ b/fs/xfs/libxfs/xfs_alloc_btree.c @@ -141,20 +141,20 @@ xfs_allocbt_update_lastrec( return; ASSERT(ptr == numrecs + 1);
+ /* + * Update in advance to prevent file creation failure + * for concurrent processes even though there is no + * numrec currently. + * And there's no need to worry as the value that not + * less than bc_free_longest will be inserted later. + */ + len = cpu_to_be32(cur->bc_free_longest); if (numrecs) { xfs_alloc_rec_t *rrp;
rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs); - len = rrp->ar_blockcount; - } else { - /* - * Update in advance to prevent file creation failure - * for concurrent processes even though there is no - * numrec currently. - * And there's no need to worry as the value that not - * less than bc_free_longest will be inserted later. - */ - len = cpu_to_be32(cur->bc_free_longest); + len = cpu_to_be32(max_t(xfs_extlen_t, cur->bc_free_longest, + be32_to_cpu(rrp->ar_blockcount))); }
break;