On 05/25/2010 09:07 PM, Sukadev Bhattiprolu wrote: > If process P1 has a F_WRLCK lease on file F1 and process P2 opens the > file, P2's open() blocks for lease_break_time (45 seconds) and P1 gets > a SIGIO to cleanup it lease in preparation for P2's open. If the two > processes are checkpointed/restarted in this window, we should address > following two issues: > > - P1 should get a SIGIO only once for the lease (i.e if P1 got the > SIGIO before checkpoint, it should not get the SIGIO after restart). > > - If R seconds remain in the lease, P2's open should be blocked for > at least the R seconds, so P1 has the time to clean up its lease. > The previous patch gives P1 the entire lease_break_time but that > can leave P2 stalled for 2*lease_break_time. > > To address first, we add a field ->fl_break_notified to "remember" if we > notified the lease-holder already. We save this field in the checkpoint > image and when restarting, we notify the lease-holder only if this field > is not set. > > To address the second issue, we also checkpoint the ->fl_break_time for > an in-progress lease. When restarting the process, we ensure that the > lease-holder sleeps only for the remaining-lease rather than the entire > lease. > > These two fixes sound like an approximation (see comments in do_setlease() > and __break_lease() below) and are also a bit kludgy (hence a separate patch > for now). > > Appreciate comments on how we can do this better. Specifically: > > - do we even need to try and address the second issue above or > just let P1 have the entire lease_break_time again ? > > - theoretically, the R seconds should start counting after *all* > processes in the application-process tree have been restarted, > since P1 waits inside the kernel for a portion of the remaining > lease - should we then add a delta to R ? [...] > @@ -1084,7 +1088,8 @@ static int restore_file_locks(struct ckpt_ctx *ctx, struct file *file, int fd) > type = h->fl_type; > if (h->fl_type & F_INPROGRESS) > type = h->fl_type_prev; > - ret = do_setlease(fd, file, type, h->fl_rem_lease); > + ret = do_setlease(fd, file, type, h->fl_rem_lease, > + h->fl_break_notified); Is h->fl_break_notified sanitized ? Oren. -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
