Tested on Dell GX280, ICH6 instead of ICH5; your workaround seems to work:
framboos:~# hwclock --show
Tue 21 Dec 2004 03:05:02 AM CET -0.672601 seconds
framboos:~# hwclock-2.12 --show
select() to /dev/rtc to wait for clock tick timed out
framboos:~#
Notes:
- util-linux-2.12h no longer hangs on broken rtc but times out.
That test was done with a business card CD when debian installer rc2 was
two days old; see http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=282861
so there's a decent chance a non-hanging hwclock on the full ISO.
- hwclock(8) states --directisa has no effect except
on i386 and alpha, so risk for other architectures from your workaround seems
limited.
- As noted by debacle in http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=277298
selfcompiled kernels seem to work fine. I tested this only with 2.6.10-rc2,
with make-kpkg, but without debian patches.
On Mon, Dec 20, 2004 at 10:27:05PM +0100, Geert Stappers wrote:
Hello,
In this bugreport there is the hint to modify the hwclock.sh files.
But that is a change at several places. I'm not fond of that.
However the addition of '--directisa' is indeed where it seems to hurt.
(kernel, hwclock and hardware mismatch)
I have this (untested) workaround in mind:
# cd /sbin
# ./hwclock --version
hwclock from util-linux-2.12
# mv hwclock hwclock-2.12
# cat hwclock HERE
#!/bin/bash
/sbin/hwclock-2.12 --directisa \$@
HERE
# chmod +x hwclock
it is a wrapper for the hwclock binary which appends directisa
That is is untested is because I have currently no access
to this Dell who is affect by this bug.
Cheers
Geert Stappers