Please don't reply to lustre-devel. Instead, comment in Bugzilla by using the
following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11301
What |Removed |Added
----------------------------------------------------------------------------
Attachment #9225 is|1 |0
patch| |
Attachment #9225 is|1 |0
private| |
Attachment #9225|review?([EMAIL PROTECTED]|review-
Flag|) |
(From update of attachment 9225)
1. ldlm_run_bl_ast_work(), ldlm_run_cb_ast_work()
@arg is shared between all the requests, which are sent out at
ptlrpc_set_wait() and interpreter is called when a request reply
is received. by that time @arg is initialized by the data of the last
prepared request, thus arg->lock_handle is wrong.
2. ldlm_server_blocking_ast()
if (unlikely(instant_cancel)) {}
error handling is missed for this case. it looks still needed because
we may fail to send the request and want to evict the client.
3. ldlm_server_blocking_ast()
if (unlikely(instant_cancel)) {}
ptlrpc_req_finished(req) is missed.
4. hld says if the amount of locks is huge, not more than some maximum
amount of requests will be sent in parallel. is it missed or considered
as not important?
_______________________________________________
Lustre-devel mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-devel
