Re: [PATCH] jbd2: wake up journal waiters in FIFO order, not LIFO

From: "Ritesh Harjani (IBM)" <ritesh.list@gmail.com>
To: Alexey Lyashkov <alexey.lyashkov@gmail.com>
Cc: linux-ext4@vger.kernel.org, Theodore Ts'o <tytso@mit.edu>,
	Andreas Dilger <adilger@dilger.ca>,
	Artem Blagodarenko <artem.blagodarenko@gmail.com>,
	Andrew Perepechko <anserper@ya.ru>
Subject: Re: [PATCH] jbd2: wake up journal waiters in FIFO order, not  LIFO
Date: Thu, 8 Sep 2022 14:43:32 +0530	[thread overview]
Message-ID: <20220908091332.lowhekyxcs23kgtu@riteshh-domain> (raw)
In-Reply-To: <20220908054611.vjcb27wmq4dggqmv@riteshh-domain>

On 22/09/08 11:16AM, Ritesh Harjani (IBM) wrote:
> On 22/09/07 07:59PM, Alexey Lyashkov wrote:
> > From: Andrew Perepechko <anserper@ya.ru>
> > 
> > LIFO wakeup order is unfair and sometimes leads to a journal
> > user not being able to get a journal handle for hundreds of
> > transactions in a row.
> > 
> > FIFO wakeup can make things more fair.
> 
> prepare_to_wait() will always add the task to the head of the list.
> While prepare_to_wait_exclusive() will add the task to the tail since all of the
> exclusive tasks are added to the tail.
> wake_up() function will wake up all non-exclusive and single exclusive task 
> v/s
> wake_up_all() function will wake up all tasks irrespective.
> 
> So your change does makes the ordering to FIFO, in which the task which came in 
> first will be woken up first. 

With all other details which got discussed in other threads. 
This looks like the right thing to do.

Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>