This is the mail archive of the
ecos-maintainers@sources.redhat.com
mailing list for the eCos project.
Re: at91 watchdog on eb42
- From: Nick Garnett <nickg at ecoscentric dot com>
- To: davarn at free dot fr
- Cc: ecos-patches at ecos dot sourceware dot org,ecos-maintainers at ecos dot sourceware dot org
- Date: 02 Aug 2004 14:23:14 +0100
- Subject: Re: at91 watchdog on eb42
- References: <1090845473.4104fb212a10b@imp5-q.free.fr><m3hdruhfzi.fsf@xl5.calivar.com><1091001934.41075e4ea029d@imp6-q.free.fr><m3r7qwdzju.fsf@xl5.calivar.com><1091172281.4109f7b959cff@imp5-q.free.fr><m3pt69airj.fsf@xl5.calivar.com><1091448136.410e2d48adf27@imp5-q.free.fr>
davarn@free.fr writes:
> Hello,
>
> The change is really the minimum, even if it is quite important.
> Tis is due to the fact that on eb42 board, if Interrupt is used in init_hw, the
> object is not already intialized so the attach function assert.
> Moving it in the start function make the thing work, and does not change the
> behavior at all.
> To be closer to the original code, it is necessary to investigate object
> intialization on eb42 board.
I was also concerned about changes to var_io.h. Your patch seems to be
deleting large numbers of definitions. I'm also not sure what the
change to AT91_US_BAUD() is going to do.
The problem with the interrupt object can be fixed by adding
CYGBLD_ATTRIB_INIT_PRI() to the interrupt object definition:
static Cyg_Interrupt wdint(
CYGNUM_HAL_INTERRUPT_WATCHDOG,
INT_PRIO,
0,
isr,
NULL
) CYGBLD_ATTRIB_INIT_PRI( CYG_INIT_DRIVERS );
This only seems to affect some targets, we haven't worked out why
this is yet.
--
Nick Garnett eCos Kernel Architect
http://www.ecoscentric.com/ The eCos and RedBoot experts