[Avila] IXP4xx watchdog reset hang
David Holmer
dholmer at jhu.edu
Fri Mar 10 12:23:40 EST 2006
I've been experimenting with the external Dallas Semiconductor watchdog on
the Avila board. I've confirmed that the watchdog chip does successfully
reset the system by manualy grounding the "right" pad of R11 for
over a second. I've also writen a small "avila-watchdog" program that uses
the ioctal interface of /dev/gpio to strobe the watchdog. I've verified
that this program (as well as the included gpio program) can
successfully control the GPIO pins on the "GPIO connector" (e.g. GPIO bit
0), but I'm having trouble with bit 14 which is connected to the watchdog
by R11.
It seems like something else in the system is already rapidly "strobing"
this bit. If I measure the "left" pad of R11 with a multi-meter I see
~1.5 volts DC and ~2.4 volts AC, which is consistent with a square wave
going back and forth between 0 and 3.3V. This occurs regardless of my
program, or the gpio utility running. The real problem is that even if I
purposely lock the kernel (i.e. by calling BUG() in a timer) this strobe
doesn't stop, and as a result the watchdog doesn't reset anything.
Anyone have any idea what is strobing bit 14? Perhaps it has something to
do with the "other" GPIO configuration registers, or could it be connected
to some clock signal somewhere?
Thanks,
David Holmer
More information about the Avila
mailing list