On Mon, Sep 24, 2012 at 05:30:38PM -0700, Tom Rini wrote:
Hey all,
The following patch series merges the davinci and omap-common SPL frameworks into a single framework, CONFIG_SPL_FRAMEWORK along with a few small cleanups to the code to make it a little smaller and more flexible. The end result is that davinci can now opt in on SPL-boots-Linux by just setting the right defines and other platforms would just need to adapt to this framework. In my testing, davinci gains between 300 and 700 bytes for this (not enabling the SPL OS feature) and omap4/5/etc lose just a little bit (from the printf->puts changes). I've tested this on omap3_beagle (xM and classic) and omap4_panda. I don't have any davinci platforms that were previously using SPL so I can't boot-test those changes but since everyone sets CONFIG_SPL_MAX_SIZE, we're OK in that department.
To make this series easier to test I've placed it on http://github.com/trini/u-boot WIP/spl-improvements
I've looked a little into re-reducing the size and the biggest problem I see is that SPI a lot of informational prints that we don't need, strictly speaking, but are nice in a normal U-Boot context. I'm unsure of the best way to quiet these as I don't like the idea of sprinkling http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54303 and I'm sending another email to discuss that.
v6 is tested on:
- AM335x EVM, Beaglebone, Beagleboard rev C (NAND and MMC), Beagleboard xM, OMAP3EVM (MMC) Devkit8000 (MMC + SPL OS goto U-Boot instead, no kernel so no SPL Linux test), Pandaboard, Pandaboard ES, LogicPD AM1808 EVM
v7 is tested on:
- Beaglebone, Beagleboard rev C (NAND and MMC), Beagleboard xM.
I did more limited testing this time due to the scope of the changes.
'reset' and cold power-on tested in various combinations.
With the one change I have mentioned already, I have now pushed this. I figure going this direcion will hopefully make the merging for Albert and others simpiler.