From mboxrd@z Thu Jan  1 00:00:00 1970
From: Stephen Warren <swarren@wwwdotorg.org>
Date: Fri, 30 May 2014 11:07:07 -0600
Subject: [U-Boot] [PATCH] mmc: add wrappers for MMC block_{read, write,
 erase}
In-Reply-To: <5388B820.7030806@broadcom.com>
References: <1401315346-30231-1-git-send-email-srae@broadcom.com>
	<53875F6D.5010708@wwwdotorg.org>
	<5387753E.9060106@broadcom.com> <538781B6.90703@wwwdotorg.org>
	<53878E10.4010609@broadcom.com> <538798E1.2040703@wwwdotorg.org>
	<5387AEB8.8030304@broadcom.com> <5388AA9C.8030504@wwwdotorg.org>
	<5388B820.7030806@broadcom.com>
Message-ID: <5388BABB.9020808@wwwdotorg.org>
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

On 05/30/2014 10:56 AM, Steve Rae wrote:
> On 14-05-30 08:58 AM, Stephen Warren wrote:
...
>> What code are you
>> looking at that handles multiple devices sequentially under program
>> control rather than user command control?
>
> Cannot go into too much detail here (yet) -- but imagine the situation
> where:
> - lookup the GPT partition name (in User, Boot1, Boot2)
> - do a block_write to the desired location...

So this is all to support some non-upstream code that you can't discuss?
That doesn't sound good...

> So after discussing with a colleague, we would propose the following.
> Does this implement what you were proposing?:
> 
> 
> Usage (example):
> 
> mmc->part_num_next_block_op = 1;          /* specify Boot1 partition */
> mmc->block_dev.block_read(0, 0, 1, buf);  /* read first block from Boot1
> partition */
> mmc->part_num_next_block_op = 0;          /* specify User partition */
> mmc->block_dev.block_read(0, 0, 1, buf);  /* read first block from User
> partition */

No. I would propose:

get_device("mmc", "0.1", &bdev_boot1);
bdev_boot1->block_read(...);
get_device("mmc", "0.0", &bdev_user);
bdev_user->block_read(...);

That way, nothing needs to change in block_dev_desc_t, get_device(), etc.