Hi Stephen,
On Wed, Mar 28, 2012 at 2:11 PM, Stephen Warren swarren@wwwdotorg.org wrote:
On 03/28/2012 02:08 PM, Simon Glass wrote:
When using CONFIG_OF_CONTROL, add a check that we have a valid FDT and panic() if not. This must be done after the console is ready.
Signed-off-by: Simon Glass sjg@chromium.org
Excellent. The behavior after this series is exactly what I was looking for.
Thanks, I'm pleased. The behavior was never in question, only the means to make it work.
One query on this patch though: Is there a need for arch/arm/lib/board.c:init_sequence[] to still include the call to fdtdec_check_fdt()? After all, if the console doesn't come from FDT, then the panic() call this patch adds will show the message so there's no need to check it earlier, and if the console does come from FDT, there's little point executing that early assert() since the message goes nowhere - presumably the code that extracts the console from FDT would perform this panic if required. So, I would have just moved the call to the existing fdtdec_check_fdt() myself, rather than splitting it into two.
I did certainly look at this, but I know we are going to hit this problem again. As you say the console-from-FDT code will currently have to do its own check, and I'm not entirely comfortable with that - it would be nice it the fdtdec_check_fdt() could actually do something useful. But I'm going to wait until another sub-arch has FDT support (and I have put up my fdt_serial series) before worrying about it further.
Still, as far as I'm concerned, this comment can be addressed in a later cleanup patch. So, the series:
Tested-by: Stephen Warren swarren@wwwdotorg.org Acked-by: Stephen Warren swarren@wwwdotorg.org
Great!
Regards, Simon