Hi Wolfgang,
On Mon, Oct 17, 2011 at 1:37 PM, Wolfgang Denk wd@denx.de wrote:
Dear Simon Glass,
In message CAPnjgZ3WBmJBs1zCd4_jQsUZcO2BOAOK865CaL6tWeUX9zVVWg@mail.gmail.com you wrote:
Yes indeed - I did actually submit an 'early panic' patch which I should take another look at - it is called 'Add board_panic_no_console to deal with early critical errors'. I think it was in the same patch set but I'm not sure.
It was (I even commented on it).
It would panic when console output was performed before the console was ready - and therefore give the behaviour you desire.
Well, but there will probably be a large number of boards that don't implement this function (or don;t enable it if we make it configurable). And all these will now fail silently.
Sorry, just emailed again on this thread - actually the behaviour of that patch has already come in due to Graeme's patch.
And I think it is actually the opposite - they will not fail silently, but in fact ignore the pre-console printf() and continue. Previously (before Graeme's patch and this one had it been applied) any pre-console printf() would cause a silent failure / hang. I think this was the intent of the fix.
The intent of my other patch in the series was to provide a way for a board to panic before the console is ready, and hopefully get a message to the user (flashing lights, LCD, init all the UARTs and blast out a msg, etc.). It is true that most boards will not implement this. We could perhaps look at implementing a weak function to do this at the architecture level (but only for SOCs) but it seems risky.
Regards, Simon
Best regards,
Wolfgang Denk
-- DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de panic: kernel trap (ignored)