On Fri, Sep 27, 2013 at 10:20:52PM -0300, Otavio Salvador wrote:
On Fri, Sep 27, 2013 at 4:26 PM, Tom Rini trini@ti.com wrote:
The following patches depend on the series I posted earlier that adds environment support to SPL as well as the series that converts the davinci bootcount driver to support am335x as well. The most obvious issue to me as that as Scott pointed out with the last series, this would be clearer Kconfig where we wouldn't need to introduce CONFIG_SPL_BOOTCOUNT_SUPPORT and just re-use CONFIG_BOOTCOUNT_SUPPORT. Other than that, I think things look good. It may be possible to re-jigger bootcount drivers to allow counting for both SPL and full U-Boot to happen, if that's strongly desired but the way I see this being used is to fall-back to a failsafe full U-Boot to do what's needed there.
I think we need to count both; the SPL part solves the U-Boot failsafe but the full U-Boot is needed to failsafe the OS.
It's _possible_ we could split the counter up further, 16bit magic, 8 bit SPL count, 8 bit U-Boot count. But I'm not sure if it's useful. Looking at Stefan's patch, we count in SPL, and use altbootcmd in full U-Boot to know we have failure rather than just booting fully via U-Boot.