Re: [PATCH] x86: only clear node_states for 64bit

From: Yinghai Lu <yinghai-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
To: Andrew Morton <akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org>
Cc: steiner-sJ/iWh9BUns@public.gmane.org,
	cl-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org,
	suresh.b.siddha-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org,
	mel-wPRd99KPJ+uzQB+pC5nmwQ@public.gmane.org,
	containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org,
	rusty-8n+1lVoiYb80n/F98K4Iww@public.gmane.org,
	linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	ntl-e+AXbWqSrlAAvxtiuMwx3w@public.gmane.org,
	viro-RmSDqhL/yNMiFSDQTTA3OLVCufUGDwFn@public.gmane.org,
	hpa-YMNOUZJC4hwAvxtiuMwx3w@public.gmane.org,
	rientjes-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org,
	mingo-X9Un+BFzKDI@public.gmane.org,
	tglx-hfZtesqFncYOwBW4kG4KsQ@public.gmane.org
Subject: Re: [PATCH] x86: only clear node_states for 64bit
Date: Fri, 26 Jun 2009 14:09:18 -0700	[thread overview]
Message-ID: <4A4538FE.2090101__2366.61782623759$1246050807$gmane$org@kernel.org> (raw)
In-Reply-To: <20090626135428.d8f88a70.akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org>

Andrew Morton wrote:
> On Mon, 22 Jun 2009 08:38:50 -0700
> Yinghai Lu <yinghai-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org> wrote:
> 
>> Nathan reported that
>> | commit 73d60b7f747176dbdff826c4127d22e1fd3f9f74
>> | Author: Yinghai Lu <yinghai-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
>> | Date:   Tue Jun 16 15:33:00 2009 -0700
>> |
>> |    page-allocator: clear N_HIGH_MEMORY map before we set it again
>> |    
>> |    SRAT tables may contains nodes of very small size.  The arch code may
>> |    decide to not activate such a node.  However, currently the early boot
>> |    code sets N_HIGH_MEMORY for such nodes.  These nodes therefore seem to be
>> |    active although these nodes have no present pages.
>> |    
>> |    For 64bit N_HIGH_MEMORY == N_NORMAL_MEMORY, so that works for 64 bit too
>>
>> the cpuset.mems cgroup attribute on an i386 kvm guest
>>
>> fix it by only clearing node_states[N_NORMAL_MEMORY] for 64bit only.
>> and need to do save/restore for that in find_zone_movable_pfn
>>
> 
> There appear to be some words omitted from this changelog - it doesn't
> make sense.
> 
> I think that perhaps a line got deleted before "the cpuset.mems cgroup
> ...".  That was the line which actualy describes the bug which we're
> fixing.  Or perhaps it was a single word?  "zeroes".
> 
> 
> I did this:
> 
> Nathan reported that
> : 
> : | commit 73d60b7f747176dbdff826c4127d22e1fd3f9f74
> : | Author: Yinghai Lu <yinghai-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
> : | Date:   Tue Jun 16 15:33:00 2009 -0700
> : |
> : |    page-allocator: clear N_HIGH_MEMORY map before we set it again
> : |
> : |    SRAT tables may contains nodes of very small size.  The arch code may
> : |    decide to not activate such a node.  However, currently the early boot
> : |    code sets N_HIGH_MEMORY for such nodes.  These nodes therefore seem to be
> : |    active although these nodes have no present pages.
> : |
> : |    For 64bit N_HIGH_MEMORY == N_NORMAL_MEMORY, so that works for 64 bit too
> : 

"
> : unintentionally and incorrectly clears the cpuset.mems cgroup attribute on
> : an i386 kvm guest
"
==> 

32bit assume NORMAL_MEMORY bit and HIGH_MEMORY bit are set for Node0 always.

and some code only check if HIGH_MEMORY is there to know if NORMAL_MEMORY is there.

YH