On Mon, Jun 20, 2011 at 3:58 AM, Michiel Konstapel <[email protected]>wrote:

> ** **
>
> *From:* [email protected] [mailto:
> [email protected]] *On Behalf Of *Peter Bigot
> *Sent:* zaterdag 18 juni 2011 22:08
>
>
> To highlight: be aware that mspgcc no longer automatically turns off the
> watchdog in its C runtime support infrastructure.  You get about 30msec
> after power-up to do something about it.  See the notes below for further
> details; I chose not to try to figure out how to determine in the .platform
> files whether the newer mspgcc was being used and thus whether
> -mdisable-watchdog was valid.****
>
> ** **
>
> This does the trick for me:****
>
> ** **
>
> # uniarch by default enables the watchdog****
>
> push @opts, "-mdisable-watchdog" if `msp430-gcc --target-help | grep
> mdisable-watchdog`;
>

Thanks.  MSP430 platform maintainers should consider whether they want to
take that approach.

Pro: system won't "bark" during startup if it takes too long to get to the
TinyOS msp430 clock init where the watchdog is currently disabled.

Con: Robust applications that want watchdog support from the instant the CPU
starts will have a hard time getting that (unless there's a way to override
that push; maybe as simple as -menable-watchdog later on the command line).

If anybody does add support for detecting the new mspgcc to a .platform
file, consider also updating the -mmcu= parameter to name the exact chip
(e.g., msp430f1611 vice msp430x1611).  As TI introduces chips where the MCU
identifier portion replaced by "x" has significance to the compiler/linker,
mspgcc may stop supporting the corresponding generic identifiers.  This is
already true for many chips in the G2xx value-line suite, which have
different memory layouts compared to the corresponding F2xx chips.

Peter
_______________________________________________
Tinyos-devel mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-devel

Reply via email to