[PATCH] add option to override redboot abort script check

[Gary Thomas]

Andrew Lunn wrote:
> On Mon, Aug 25, 2008 at 10:02:37PM -0400, Jose Vasconcellos wrote:
>> This patch factors out the test for ^C that is used to bypass the  
>> start-up script.
>> By creating the function redboot_abort_script and declaring weak, it's  
>> possible
>> to easily replace it for custom needs.
>>
>> In my case, there's no easily accessible serial port so get to the  
>> redboot prompt
>> telnet is used. However, if there's a boot script, there needs to be a  
>> way for the
>> user to abort it in an easy manner. So, in my platform specific code, the
>> redboot_abort_script is replaced by code to check for to see button if a  
>> button
>> is pressed.
> 
> What i don't like about this is that it removes the normal ^C
> handling. For you that is probably O.K. however in other situations,
> it would be better to add additional mechanism, rather than replace
> mechanisms. So in the field you can use a jumper on a GPIO, but the
> developers desk which does have a serial port, maybe via a bed of
> nails, can continue to use ^C etc.

Indeed, the initial discussion did not mention using weak functions
(although I suspected this).  I'm also not keen on a wholesale replacement.

> Redboot makes a lot of use of HAL tables, eg for adding new commands,
> idle handlers, etc. What i would suggest is adding a new table for
> script abort functions. The ^C handler would be the first entry in the
> table. HALs or other packages could then add more entries, eg read a
> GPIO line etc. If any of these functions return true, the script is
> aborted. 

Another possibility would be to use a HAL defined macro, in the
same spirit as HAL_MEM_REAL_REGION_TOP which is used in that same
file (main.c)



-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------