From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tom Rini Date: Mon, 5 May 2014 09:09:49 -0400 Subject: [U-Boot] DA850EVM with USE_NAND config does not pad the AIS file In-Reply-To: <535BEE42.3050105@gmail.com> References: <535BEE42.3050105@gmail.com> Message-ID: <20140505130949.GP22182@bill-the-cat> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de On Sat, Apr 26, 2014 at 01:34:58PM -0400, Tom Taylor wrote: > I'm a U-Boot newbie so please feel free to correct how I'm reporting > this issue.. > > I recently downloaded the 2014.04-rc3 snapshot to build U-Boot for > my custom DA850-based board. The only change was to add a new > target "dav850evm_nand" in boards.cfg with the added parameter > "USE_NAND". > > The resulting AIS file was programmed into EVM-compatible NAND using > standard sfh_OMAP-L138 method. > > The board failed to boot, and stayed in a loop printing the SPL > console message repeatedly. > > After some debugging with CCS 5.5 and an XDS100v2, I found that > incorrect code was being loaded into the 0xc108000 RAM destination. > The da850evm.h file defines CONFIG_SYS_NAND_U_BOOT_OFFS as 0x28000, > which corresponds to an AIS offset of 0x8000 but the u-boot header > did not appear there in the AIS file. A search revealed that the > Makefile catenated u-boot immediately after the SPL without any > padding. > > Further investigation revealed that the target Makefile needs > CONFIG_SPL_MAX_SIZE to be defined as 0x8000 in order for the padding > to be performed properly; however, this constant was apparently > deleted during a series of changes in April, 2013 to accommodate > separate code and BSS size limits for another target. In its place, > CONFIG_SPL_MAX_FOOTPRINT was defined as 32768. Unfortunately, the > da850evm Makefile does not refer to this constant. > > To solve the problem, I added the following 2 lines in my > custom-modified da850evm.h: > #define CONFIG_SPL_PAD_TO 0x8000 > #define CONFIG_SPL_MAX_SIZE 0x8000 Thanks for looking into this. I saw a similar issue on my am18xx EVM but was 5 bugs deep and didn't have time to bisect it then. I think the answer here is to post a patch fixing the boards listed in: $ git grep -l ARM926EJS include/configs/ | xargs grep -l DAVINCI As they all share that set of constraints. Thanks for digging into this! -- Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 836 bytes Desc: Digital signature URL: