[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20090612100756.GA25185@elte.hu>
Date: Fri, 12 Jun 2009 12:07:56 +0200
From: Ingo Molnar <mingo@...e.hu>
To: Pekka Enberg <penberg@...helsinki.fi>
Cc: linux-mm@...ck.org, linux-kernel@...r.kernel.org, npiggin@...e.de,
benh@...nel.crashing.org, akpm@...ux-foundation.org,
cl@...ux-foundation.org, torvalds@...ux-foundation.org
Subject: Re: [PATCH v2] slab,slub: ignore __GFP_WAIT if we're booting or
suspending
* Pekka Enberg <penberg@...helsinki.fi> wrote:
> On Fri, Jun 12, 2009 at 12:10 PM, Ingo Molnar<mingo@...e.hu> wrote:
> >> @@ -1548,6 +1548,20 @@ new_slab:
> >> goto load_freelist;
> >> }
> >>
> >> + /*
> >> + * Lets not wait if we're booting up or suspending even if the user
> >> + * asks for it.
> >> + */
> >> + if (system_state != SYSTEM_RUNNING)
> >> + gfpflags &= ~__GFP_WAIT;
> >
> > Hiding that bug like that is not particularly clean IMO. We should
> > not let system_state hacks spread like that.
> >
> > We emit a debug warning but dont crash, so all should be fine and
> > the culprits can then be fixed, right?
>
> OK, lets not use system_state then and go with Ben's approach
> then. Again, neither of the patches are about "hiding buggy
> callers" but changing allocation policy wrt. gfp flags during boot
> (and later on during suspend).
IMHO such invisible side-channels modifying the semantics of GFP
flags is a bit dubious.
We could do GFP_INIT or GFP_BOOT. These can imply other useful
modifiers as well: panic-on-failure for example. (this would clean
up a fair amount of init code that currently checks for an panics on
allocation failure.)
Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists