From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ilya Averyanov Date: Mon, 23 Apr 2012 18:27:11 +0400 Subject: [U-Boot] [PATCH v3] Add support for MINI2440 (s3c2440). Documentation about the product can be found on: http://www.friendlyarm.net/products/mini2440 In-Reply-To: <201204231607.37587.marex@denx.de> References: <20120416174633.GA7463@debian> <201204211519.30964.marex@denx.de> <201204231607.37587.marex@denx.de> Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de 2012/4/23 Marek Vasut > Dear Ilya Averyanov, > > > Dear Marek Vasut, > > > > Would you please explain, why do you want to initialize the sdram has > been > > in C code. > > Ok, no problem: > - You'll be able to call get_ram_size() to scan for ram size as any other > board > - SDRAM driver will fit properly into the driver model once it's in place > (WIP, > wanna volunteer for review process?) > - It's much cleaner > > > If for provide this feature should be written assembly code. And it would > > sometimes even harder than the initialization code sdram? > > What do you mean? > If we use 4kb of SRAM (steppingstone), then at least need to edit u-boot/arch/arm/cpu/arm920t/start.S, to add a test from there was a boot (NAND or NOR). Edit the address stack. etc ... Use the MMU cache for the stack, even a little better, but it is more difficult. > > Ah yes, just a remark, please don't top-post. Notice how other people > don't do > it -- I know gmail does it out-of-the-box, but that doesn't mean it's > correct. > > Ok. > > 2012/4/21 Marek Vasut > > > > > Dear Ilya Averyanov, > > > > > > > 2012/4/21 Vasily Khoruzhick > > > > > > > > > 2012/4/21 Ilya Averyanov : > > > > > > No have problem. > > > > > > Available only if you boot from NOR flash. > > > > > > If you boot from the NAND 4kb of SRAM (steppingstone) not > > > > > > available. > > > > > > > > > > For NAND boot it's mapped at 0x0 > > > > > For NOR boot it's mapped at 0x40000000 > > > > > > > > > > NAND controller uses it for booting purpose, but anyway for SPL we > > > > > should fit code into > > > > > 4k (hardware limitation), so I think using assembly here for SDRAM > > > > > init is OK. > > > > > > > > > > And how will you use the internal SRAM when you boot from NAND, if > > > > > the SoC > > > > > > > > writes in the SDRAM read data from NAND, and your data will be > > > > > > overwritten. > > > > > > Why? You'll load SPL into SRAM, then in SPL you'll copy the rest of > uboot > > > from > > > NAND to - then already inited - SDRAM and execute it. We even have > > > ready-to-use > > > NAND SPL. > > > > > > > > Regards > > > > > Vasily > > > > > > Best regards, > > > Marek Vasut > > Best regards, > Marek Vasut >