Re: 2.6.28-rc9 panics with crashkernel=256M while booting

From: Chandru <chandru@in.ibm.com>
To: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: linuxppc-dev@ozlabs.org,
	Andrew Morton <akpm@linux-foundation.org>,
	Paul Mackerras <paulus@samba.org>,
	linux-kernel@vger.kernel.org
Subject: Re: 2.6.28-rc9 panics with crashkernel=256M while booting
Date: Thu, 22 Jan 2009 13:50:21 +0530	[thread overview]
Message-ID: <200901221350.21601.chandru@in.ibm.com> (raw)
In-Reply-To: <1232584180.27278.86.camel@nimitz>

On Thursday 22 January 2009 05:59:39 Dave Hansen wrote:
> Let's take, for instance, a 1-byte reservation.  With this code, you've
> suddenly turned that into a 0-byte reservation, and that *can't* be
> right.  The same thing happens if you have a reservation that spans two
> pages.  If you unconditionally round it down, then you might miss the
> part that spans a portion of the second page.
> 
> It needs to be rounded down like you are suggesting here, but only in
> the case where we've gone over the *CURRENT* node's boundary.  This is
> kinda what that "if (end_pfn > node_ar.end_pfn)" check is doing.  But,
> it evidently screws it up if the overlap isn't by an entire page or
> something.

I assumed the condition 'while (start_pfn < end_pfn  && .. )' asks for atleast
a PAGE_SIZE difference between them and hence went ahead with that patch.
My guess was a 1-byte , 2-byte or a (PAGE_SIZE -1)-byte reservations may not even
go into that loop.  However we just need a fix for this problem. So if there is a 
better fix that you have please post it to lkml. 

Thanks,
Chandru