From mboxrd@z Thu Jan  1 00:00:00 1970
From: Dirk Behme <dirk.behme@googlemail.com>
Date: Sun, 02 Nov 2008 19:41:08 +0100
Subject: [U-Boot] Creating u-boot-arm-omap3 branch
Message-ID: <490DF444.7030501@googlemail.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


At IRC we had a short discussion what might be the best way how to go 
on with OMAP3 patch series. We are now at version 5 of this patch set, 
unfortunately there are still review comments open [1]. On the other 
hand, from reviewer point of view, sending patch series again and 
again (v6, v7, ...) would need always a complete re-review as reviewer 
can't be sure that no new changes slipped into already reviewed patch 
parts.

Therefore, Jean-Christophe kindly offered to create an 
u-boot-arm-omap3 git branch with v5 patch set just sent. We see 
several advantages:

First, the patch set is more visible for people able to test and 
hopefully willing to help sending patches. Second, only patches to fix 
the open/new review results have to be send to the list and not whole 
patch series. Third, this will avoid need for re-review of whole patch 
series as only the (smaller) fixes have to be reviewed, not the whole 
series again. And fourth, just in case I will disappear, the branch 
can be just deleted if there is no progress to make it mainline ready 
(in contrast to adding OMAP3 now to mainline and trusting that I will 
fix the remaining issues in mainline).

Once we think that OMAP3 patch set in u-boot-arm-omap3 branch is ready 
for mainline, it will be extracted (e.g. by git-rebase -i) and resend 
for final inclusion/review to U-Boot list again. This will avoid git 
pollution with "fix style here" "fix style there" commits.

Thanks for your patience and help, best regards

Dirk

[1] Review comments still open:

1) Use readx/writex with base + offset with base being e.g. from 
uint32_t. NAND driver gpmc_omap.c is already (hopefully correctly) 
converted to this style and can be taken as example.

2) For start.S, function cpu_init_crit():

cpu_init_crit: "why create function that could be call from c also"
cpu_init_crit: "why create function that could be call from c also 
this could be availlable for all cache/mmu/interruptions function"

Jean-Christophe likes that this code can be called from C, too.

3) Replace all hard coded values with macros.

The main places I know are

- misc_init_r() in the board files (evm.c, overo.c, beagle.c)
- dpll_param tables in lowlevel_init.S (to be done like core_dpll_param)
- do_sdrc_init() in mem.c

x) Hopefully all other comments are already fixed. More review 
comments are welcome