lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
Open Source and information security mailing list archives
| ||
|
Message-Id: <1199916092.4011.3.camel@localhost.localdomain> Date: Wed, 09 Jan 2008 14:01:32 -0800 From: Mingming Cao <cmm@...ibm.com> To: "Aneesh Kumar K.V" <aneesh.kumar@...ux.vnet.ibm.com> Cc: Jan Kara <jack@...e.cz>, tytso@....edu, adilger@....com, bzzz@....com, linux-ext4@...r.kernel.org Subject: Re: [PATCH] ext4: Fix the soft lockup with multi block allocator. On Thu, 2008-01-10 at 00:41 +0530, Aneesh Kumar K.V wrote: > On Wed, Jan 09, 2008 at 07:44:30PM +0100, Jan Kara wrote: > > On Wed 09-01-08 23:54:28, Aneesh Kumar K.V wrote: > > > On Wed, Jan 09, 2008 at 01:10:41PM +0100, Jan Kara wrote: > > > > > With the multi block allocator when we don't have prealloc space we discard > > > > > @@ -3790,7 +3782,9 @@ repeat: > > > > > > > > > > /* if we still need more blocks and some PAs were used, try again */ > > > > > if (free < needed && busy) { > > > > > + busy = 0; > > > > > ext4_unlock_group(sb, group); > > > > > + schedule_timeout(HZ); > > > > > goto repeat; > > > > > } > > > > Hmm, wouldn't just schedule() be enough here? That would give a good > > > > chance to other processes to proceed and we would avoid this artificial > > > > wait of 1s which is quite ugly IMO. > > > > > > > > > > But then who will wake up the task ?. I have the below comment added to > > > the patch in the patch queue. > > As far as I know, you don't have to wake-up the task explicitely. > > Scheduler will simply schedule the task sometime in future (it is a similar > > situation as if the task got preempted in the kernel). > > I missed the current->state = TASK_UNINTERRUPTIBLE; in that code piece. > So yes without setting the task state yes it would put it back to the run > queue.Infact schedule_timeout() without changing the task state also > behaves similarly. Now that that we know that it if fine just to have a > schedule() there since schedule_timeout() was just behaving as > schedule(). I guess we should make the change you suggested. In that > case we can remove the comment which says we need to add the wait queue. > > Mingming, > > Do you want me to send a patch or can you make the modification ? > I could make the changes and update the mballoc patch in the queue. Mingming > -aneesh > - To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@...r.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists