On Mon, Dec 07, 2009 at 10:41:15PM +0300, Dmitry Monakhov wrote: > Absolutely right. I've fixed an issue, but overlooked the BIGGEST one. > So off course my patch is wrong, even if we will acquire lock in > different order " dqptr_sem > i_block_reservation_lock" > we sill getting in to sleeping spin lock problems by following scenario: > ext4_da_update_reserve_space() > ->dquot_claim_space() > ASSUMES that we hold i_block_reservation_lock here. > -->mark_dquot_dirty() > --->ext4_write_dquot() > if (journalled quota) ext4_write_dquot(); > ---->dquot_commit() > ----->mutex_lock(&dqopt->dqio_mutt's); <<< sleep here. > > This means that we have fully redesign quota reservation locking. > As i already suggested previously here: > http://thread.gmane.org/gmane.comp.file-systems.ext4/16576/focus=16587 Given this, should I include this patch for now, given that it does fix _one_ race, or should I hold off until you redo the locking? How long do you think to send a revised/new patch? - Ted -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
