>Number:         19279
>Category:       kern
>Synopsis:       Lost interrupts with hard disk after turning power on (also see 12693 & 16427)
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Dec 04 12:42:00 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     Teemu Rinta-aho
>Release:        1.6_STABLE
>Organization:
>Environment:
1.6_STABLE / i386, source code from anoncvs.netbsd.org on 2nd Dec 2002
with tag netbsd-1-6.
>Description:
The chipset problem is back - again. There have been changes in the
pciide code over the years and somehow this same problem has once
again reappeared. I have not changed my hardware, just updated
the NetBSD every now and then. See problem report #16427. I just
make this problem report to let everyone know that 1.6 branch also
has the bug as people might think PR #16427 is no more relevant as
it deals with 1.5.3. Well, unfortunately it is.

Funny thing is that if I start up my machine with Windows 2000 first,
which doesn't have any problems with the chipset and the disk, then
after a warm reboot the NetBSD boots up fine also. Also if I start
up NetBSD, then restart, it works. It is only that the first boot
of NetBSD (after the power has been switched on) doesn't somehow seem to set up the Promise ATA-100 controller correctly, and this state which Windows driver somehow sets, stays in the controller over warm reboots. This is one lead which could be followed..
>How-To-Repeat:
I can repeat it any time. Too bad that I don't have a digital camera
so the dmesg is lost every time.. :-)
>Fix:
I don't know exactly, but somehow the Promise ATA-100 controller
is not set up correctly by the driver. It was wrong with 1.5.3,
then it was corrected in 1.6-release and now the bug has reappeared in the 1.6-stable.
>Release-Note:
>Audit-Trail:
>Unformatted: