[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 6 Jun 2016 16:59:02 -0700
From: Linus Torvalds <torvalds@...ux-foundation.org>
To: Al Viro <viro@...iv.linux.org.uk>
Cc: Dave Hansen <dave.hansen@...el.com>,
"Chen, Tim C" <tim.c.chen@...el.com>,
Ingo Molnar <mingo@...hat.com>,
Davidlohr Bueso <dbueso@...e.de>,
"Peter Zijlstra (Intel)" <peterz@...radead.org>,
Jason Low <jason.low2@...com>,
Michel Lespinasse <walken@...gle.com>,
"Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>,
Waiman Long <waiman.long@...com>,
LKML <linux-kernel@...r.kernel.org>
Subject: Re: performance delta after VFS i_mutex=>i_rwsem conversion
On Mon, Jun 6, 2016 at 4:50 PM, Linus Torvalds
<torvalds@...ux-foundation.org> wrote:
>
>
> And I do want to repeat that the patch is entirely untested. It compiles.
> I looked at the assembly it generated. It looks fine to me, but I might
> have had a brainfart and done something completely broken.
Dammit.
I did that whole dcache_emit_entry() thing because we need to do the
final target cursor move regardless, and dcache_emit_entry() will
always return with the spinlock held again.
But then I didn't actually fix the "return" to a "break".
So the "return 0" here:
> + /* This will drop and re-take the dentry lock .. */
> + if (!dcache_emit_entry(ctx, dentry, child))
> return 0;
is fatal and woudl return with the spinlock held. It *should* have
been a "break" to exit the readdir loop.
So the patch I sent out was indeed a terminal brainfart, but with that
fix to change that return to a "break" it *might* work.
Linus
Powered by blists - more mailing lists