From mboxrd@z Thu Jan  1 00:00:00 1970
From: Derek Ou <derek@siconix.com>
Date: Tue, 17 Feb 2009 13:54:00 -0700
Subject: [U-Boot] [PATCH] Pad data length for nand write
In-Reply-To: <20090217185155.GA31737@ld0162-tx32.am.freescale.net>
References: <1234890907-934-1-git-send-email-dou@siconix.com>
	<1234890907-934-2-git-send-email-dou@siconix.com>
	<20090217185155.GA31737@ld0162-tx32.am.freescale.net>
Message-ID: <499B23E8.30607@siconix.com>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

This routine is copied from nand_write_opts() at nand_util.c of v1.3.4.  
It maybe better
to pad data length and data buffer in do_nand() at cmd_nand.c.  But I 
don't see real
difference.  I didn't allocate a new buffer since the data length is 
defined in the command
arguments and I have no knowledge of a safe location and large enough 
memory space
to store data.

Maybe we should implement, like Wolfgang said, a "+length" syntax in the 
"nand write"
command to indicates "round up to the next page boundary".  In this 
case, do_nand()
can assume that it's safe to write pass memory address (add + length) 
and Flash offset
(off + length).

Derek
Scott Wood wrote:
> On Tue, Feb 17, 2009 at 10:15:07AM -0700, Derek Ou wrote:
>   
>> +	/* now, pad data with 0xff */
>> +	if (page_offset != 0)
>> +		memset(buffer + *length, 0xff, pad_len);
>>     
> You cannot write to the caller's buffer (or worse, past the end of the
> caller's buffer) like this.  You'll need to allocate a new, padded
> buffer.
>
> -Scott
>