From mboxrd@z Thu Jan 1 00:00:00 1970 From: Pekka Enberg Subject: Re: [Bug #13319] Page allocation failures with b43 and p54usb Date: Wed, 10 Jun 2009 19:16:44 +0300 Message-ID: <1244650604.23799.35.camel@penberg-laptop> References: <4A2BBC30.2030300@lwfinger.net> <84144f020906070640rf5ab14nbf66d3ca7c97675f@mail.gmail.com> <4A2BCC6F.8090004@redhat.com> <84144f020906070732l31786156r5d9753a0cabfde79@mail.gmail.com> <20090608101739.GA15377@csn.ul.ie> <84144f020906080352k57f12ff9pbd696da5f332ac1a@mail.gmail.com> <20090608110303.GD15377@csn.ul.ie> <20090608141212.GE15070@csn.ul.ie> <1244531201.5024.3.camel@penberg-laptop> <1244534315.5024.34.camel@penberg-laptop> <1244536124.5024.41.camel@penberg-laptop> <4A2FC62C.6030407@lwfinger.net> <1244648659.23799.26.camel@penberg-laptop> <1244648968.23799.27.camel@penberg-laptop> <1244649174.6165.0.camel@johannes.local> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1244649174.6165.0.camel-YfaajirXv2244ywRPIzf9A@public.gmane.org> Sender: kernel-testers-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: text/plain; charset="us-ascii" To: Johannes Berg Cc: Larry Finger , David Rientjes , Mel Gorman , Rik van Riel , "Rafael J. Wysocki" , Linux Kernel Mailing List , Kernel Testers List , Andrew Morton , KOSAKI Motohiro , KAMEZAWA Hiroyuki , Christoph Lameter , npiggin-l3A5Bk7waGM@public.gmane.org, yanmin.zhang-VuQAYsv1563Yd54FQh9/CA@public.gmane.org On Wed, 2009-06-10 at 18:49 +0300, Pekka Enberg wrote: > > On Wed, 2009-06-10 at 18:44 +0300, Pekka Enberg wrote: > > > > > Aha, SLUB thinks the minimum order for 4096 is 1. I guess you have > > > CONFIG_SLUB_DEBUG enabled? If yes, something like to following should > > > help. Christoph, are you okay with this patch? On Wed, 2009-06-10 at 17:52 +0200, Johannes Berg wrote: > > + if ((size + MAX_DEBUG_SIZE) >= PAGE_SIZE) > > && size <= PAGE_SIZE > > ? Or is this a path that only happens for small allocations? > > > + s->flags &= ~(SLAB_POISON|SLAB_RED_ZONE|SLAB_STORE_USER); > > + > > if (!calculate_sizes(s, -1)) > > goto error; Although something like this would probably be even nicer. Pekka diff --git a/mm/slub.c b/mm/slub.c index 65ffda5..a4206ef 100644 --- a/mm/slub.c +++ b/mm/slub.c @@ -2334,6 +2334,16 @@ static int calculate_sizes(struct kmem_cache *s, int forced_order) } +#define MAX_DEBUG_SIZE (3 * sizeof(void *) + 2 * sizeof(struct track)) + +static bool must_disable_debug(size_t size) +{ + /* + * Disable debugging if it increases the minimum page order. + */ + return get_order(size + MAX_DEBUG_SIZE) > get_order(size); +} + static int kmem_cache_open(struct kmem_cache *s, gfp_t gfpflags, const char *name, size_t size, size_t align, unsigned long flags, @@ -2346,6 +2356,9 @@ static int kmem_cache_open(struct kmem_cache *s, gfp_t gfpflags, s->align = align; s->flags = kmem_cache_flags(size, flags, name, ctor); + if (must_disable_debug(size)) + s->flags &= ~(SLAB_POISON|SLAB_RED_ZONE|SLAB_STORE_USER); + if (!calculate_sizes(s, -1)) goto error;