| Message ID | 159854066219.1382667.2526502599415329370.stgit@warthog.procyon.org.uk |
|---|---|
| State | Accepted |
| Delegated to: | David Miller |
| Headers | show |
| Series | rxrpc, afs: Fix probing issues | expand |
diff --git a/fs/afs/vl_rotate.c b/fs/afs/vl_rotate.c index ab2beb4ba20e..c0458c903b31 100644 --- a/fs/afs/vl_rotate.c +++ b/fs/afs/vl_rotate.c @@ -263,10 +263,14 @@ bool afs_select_vlserver(struct afs_vl_cursor *vc) for (i = 0; i < vc->server_list->nr_servers; i++) { struct afs_vlserver *s = vc->server_list->servers[i].server; + if (test_bit(AFS_VLSERVER_FL_RESPONDING, &s->flags)) + e.responded = true; afs_prioritise_error(&e, READ_ONCE(s->probe.error), s->probe.abort_code); } + error = e.error; + failed_set_error: vc->error = error; failed:
The error handling in the VL server rotation in the case of there being no contactable servers is not correct. In such a case, the records of all the servers in the list are scanned and the errors and abort codes are mapped and prioritised and one error is chosen. This is then forgotten and the default error is used (EDESTADDRREQ). Fix this by using the calculated error. Also we need to note whether a server responded on one of its endpoints so that we can priorise an error from an abort message over local and network errors. Fixes: 4584ae96ae30 ("afs: Fix missing net error handling") Signed-off-by: David Howells <dhowells@redhat.com> --- fs/afs/vl_rotate.c | 4 ++++ 1 file changed, 4 insertions(+)