@@ -21,6 +21,34 @@
#define NEED_RETRY 0x10 /* A front op requests retrying */
#define SAW_FAILURE 0x20 /* One stream or hit a permanent failure */
+static void netfs_dump_request(const struct netfs_io_request *rreq)
+{
+ pr_err("Request R=%08x r=%d fl=%lx or=%x e=%ld\n",
+ rreq->debug_id, refcount_read(&rreq->ref), rreq->flags,
+ rreq->origin, rreq->error);
+ pr_err(" st=%llx tsl=%zx/%llx/%llx\n",
+ rreq->start, rreq->transferred, rreq->submitted, rreq->len);
+ pr_err(" cci=%llx/%llx/%llx\n",
+ rreq->cleaned_to, rreq->collected_to, atomic64_read(&rreq->issued_to));
+ pr_err(" iw=%pSR\n", rreq->netfs_ops->issue_write);
+ for (int i = 0; i < NR_IO_STREAMS; i++) {
+ const struct netfs_io_subrequest *sreq;
+ const struct netfs_io_stream *s = &rreq->io_streams[i];
+
+ pr_err(" str[%x] s=%x e=%d acnf=%u,%u,%u,%u\n",
+ s->stream_nr, s->source, s->error,
+ s->avail, s->active, s->need_retry, s->failed);
+ pr_err(" str[%x] ct=%llx t=%zx\n",
+ s->stream_nr, s->collected_to, s->transferred);
+ list_for_each_entry(sreq, &s->subrequests, rreq_link) {
+ pr_err(" sreq[%x:%x] sc=%u s=%llx t=%zx/%zx r=%d f=%lx\n",
+ sreq->stream_nr, sreq->debug_index, sreq->source,
+ sreq->start, sreq->transferred, sreq->len,
+ refcount_read(&sreq->ref), sreq->flags);
+ }
+ }
+}
+
/*
* Successful completion of write of a folio to the server and/or cache. Note
* that we are not allowed to lock the folio here on pain of deadlocking with
@@ -87,6 +115,12 @@ static void netfs_writeback_unlock_folios(struct netfs_io_request *wreq,
unsigned long long collected_to = wreq->collected_to;
unsigned int slot = wreq->buffer.first_tail_slot;
+ if (WARN_ON_ONCE(!folioq)) {
+ pr_err("[!] Writeback unlock found empty rolling buffer!\n");
+ netfs_dump_request(wreq);
+ return;
+ }
+
if (wreq->origin == NETFS_PGPRIV2_COPY_TO_CACHE) {
if (netfs_pgpriv2_unlock_copied_folios(wreq))
*notes |= MADE_PROGRESS;
It seems that it's possible to get to netfs_writeback_unlock_folios() with an empty rolling buffer during buffered writes. This should not be possible as the rolling buffer is initialised as the write request is set up and thereafter maintains at least one folio_queue struct therein until it gets destroyed. This allows lockless addition and removal of folio_queue structs in the buffer because, unlike with a ring buffer, the producer and consumer each only need to look at and alter one pointer into the buffer. Now, the rolling buffer is only used for buffered I/O operations as netfs_collect_write_results() should only call netfs_writeback_unlock_folios() if the request is of origin type NETFS_WRITEBACK, NETFS_WRITETHROUGH or NETFS_PGPRIV2_COPY_TO_CACHE. So it would seem that one of the following occurred: (1) I/O started before the request was fully initialised, (2) the origin got switched mid-flow or (3) the request has already been freed and this is a UAF error. I think the last is the most likely. Make netfs_writeback_unlock_folios() report information about the request and subrequests if folioq is seen to be NULL to try and help debug this, throw a warning and return. Note that this does not try to fix the problem. Reported-by: syzbot+af5c06208fa71bf31b16@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?extid=af5c06208fa71bf31b16 Signed-off-by: David Howells <dhowells@redhat.com> cc: Chang Yu <marcus.yu.56@gmail.com> Link: https://lore.kernel.org/r/ZxshMEW4U7MTgQYa@gmail.com/ cc: Jeff Layton <jlayton@kernel.org> cc: netfs@lists.linux.dev cc: linux-fsdevel@vger.kernel.org --- fs/netfs/write_collect.c | 34 ++++++++++++++++++++++++++++++++++ 1 file changed, 34 insertions(+)