From mboxrd@z Thu Jan  1 00:00:00 1970
From: Tom Rini <trini@ti.com>
Date: Mon, 27 Aug 2012 08:43:12 -0700
Subject: [U-Boot] [PATCHv1] ARM: Add Altera SOCFPGA Cyclone5
In-Reply-To: <20120825115601.GA19418@elf.ucw.cz>
References: <1345157139-19585-1-git-send-email-dinguyen@altera.com>
	<201208230026.53389.marex@denx.de>
	<20120823011353.GB3476@bill-the-cat>
	<20120823104342.GA28214@elf.ucw.cz> <5037D383.2080105@ti.com>
	<20120825115601.GA19418@elf.ucw.cz>
Message-ID: <503B9590.7000403@ti.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

On 08/25/2012 04:56 AM, Pavel Machek wrote:
> Hi!
> 
>> [snip]
>>>> It's the omap-common/spl.c code with the bits for IH_OS_LINUX taken out
>>>> :)  Can you please re-do your series on top of the SPL series I just
>>>> posted that provides a common SPL framework?  Thanks!
>>>
>>> I'll take a look. OTOH, ammount of code duplication here is quite low,
>>> and redoing it on top of SPL series will mean delaying merge, right?
>>
>> My hope is that since I got the SPL patches posted in time we can get
>> them in for v2012.10 so no, this will only help your chances :)
> 
> Hopefully.
> 
> Some observations:
> 
> Not sure what to do with CONFIG_SPL_STACK: altera was just using
> symbol from linker script. Is there way to keep it like that? I'd hate
> to invent magic constant.

Well, lets think about this.  Previously, everyone has been saying "use
this spot in SRAM" or "use this spot in DRAM (which has been setup
previously)".  But I think you should be able to do
#define CONFIG_SPL_STACK __linker_symbol and be done with it.

> Same issue with CONFIG_SYS_SPL_MALLOC_START... I was doing:
> 
>         /* setup MALLOC after clocks going faster */
> 	mem_malloc_init((ulong) &__malloc_start,
> 					(&__malloc_end -
> 	&__malloc_start));
> 
> I guess defines allow me to keep my definition. Good.

Should be fine, yes.

> spl_parse_image_header(): We have "modern" system, we'd prefer not to
> do
>                 /* Signature not found - assume u-boot.bin */
> ...but to hang in that case, in order not to jump to random code.
> 
> Ok, I should have something publishable, soon ;-).

The only reason I like the assumption idea is that I've already seen at
least one distribution ship new MLO (based on SPL) and u-boot.bin rather
than u-boot.img for their beagleboard images.

-- 
Tom