[v2] erofs: fix missing xas_retry() in fscache mode
Commit Message
The xarray iteration only holds the RCU read lock and thus may encounter
XA_RETRY_ENTRY if there's process modifying the xarray concurrently.
This will cause oops when referring to the invalid entry.
Fix this by adding the missing xas_retry(), which will make the
iteration wind back to the root node if XA_RETRY_ENTRY is encountered.
Fixes: d435d53228dd ("erofs: change to use asynchronous io for fscache readpage/readahead")
Suggested-by: David Howells <dhowells@redhat.com>
Reviewed-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Reviewed-by: Jia Zhu <zhujia.zj@bytedance.com>
Signed-off-by: Jingbo Xu <jefflexu@linux.alibaba.com>
---
v2:
- update the commit message suggested by David
- add "Suggested-by" tag, as this patch actually fixes the same issue
as that fixed by David's patch [1]. Sorry I forgot in v1 ;-\
- add "Reviewed-by" tags
[1] https://lore.kernel.org/lkml/084d78a4-6052-f2ec-72f2-af9c4979f5dc@linux.alibaba.com/T/#m036d0221dacde491a1bcf2f4428e8b1450179df9
---
fs/erofs/fscache.c | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)
@@ -75,11 +75,15 @@ static void erofs_fscache_rreq_unlock_folios(struct netfs_io_request *rreq)
rcu_read_lock();
xas_for_each(&xas, folio, last_page) {
- unsigned int pgpos =
- (folio_index(folio) - start_page) * PAGE_SIZE;
- unsigned int pgend = pgpos + folio_size(folio);
+ unsigned int pgpos, pgend;
bool pg_failed = false;
+ if (xas_retry(&xas, folio))
+ continue;
+
+ pgpos = (folio_index(folio) - start_page) * PAGE_SIZE;
+ pgend = pgpos + folio_size(folio);
+
for (;;) {
if (!subreq) {
pg_failed = true;