From mboxrd@z Thu Jan  1 00:00:00 1970
From: b32955@freescale.com (Huang Shijie)
Date: Wed, 19 Jun 2013 16:47:56 +0800
Subject: [PATCH 1/2] ARM: mmu: fix the hang when we steal a
 section	unaligned size memory
In-Reply-To: <20130619082840.GG2718@n2100.arm.linux.org.uk>
References: <1371113826-1231-1-git-send-email-b32955@freescale.com>
 <20130618152905.GB8534@mudshark.cambridge.arm.com>
 <20130618165246.GC2718@n2100.arm.linux.org.uk>
 <51C11916.2030807@freescale.com>
 <20130619082840.GG2718@n2100.arm.linux.org.uk>
Message-ID: <51C1703C.6010003@freescale.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

? 2013?06?19? 16:28, Russell King - ARM Linux ??:
> On Wed, Jun 19, 2013 at 10:36:06AM +0800, Huang Shijie wrote:
>> ? 2013?06?19? 00:52, Russell King - ARM Linux ??:
>>> This is why I've always said - if you steal memory from memblock, it
>>> _must_  be aligned to 1MB (the section size) to avoid this.  Not only
>> firstly, it's a little waste to _must_ steal 1M for some board with less
>> memory such as 512M memory;
> You're complaining about 512M memory?  Christ.  Some of my machines
> which I run here have as little as 32M of memory!  1M is nothing in
> 512M.
>
My meaning was : we only need 128K in some case, why we waste other 896K?
IMHO, we should treasure the memory.

If you think there is no need to fix this issue, I am ok too.

>> secondly, if we do not support the section unaligned mapping, we should
>> remove the alloc_init_pte() in alloc_init_pmd(),
>> add a comment to tell that we do not support the section unaligned mapping.
> No, because that breaks a whole load of other non-memory mappings.
>
yes. you are right. I forgot the other mappings besides the map_lowmem().

thanks
Huang Shijie