ACPI/PCI-bus issue with compaq evo n160

[Fredrik Lindberg]

Hi,

I tried removing cbb and related stuff from the kernel, no luck.
The machine locks up just like before, output from db

http://shapeshifter.se/~fredde/tmp/db.acpi-cbb-disabled

That db-output showed uchi still on irq10, so I disabled all usb-related stuff
from the kernel, and guess what, the machine booted with working acpi.
Some output with acpi enabled and cbb and uhci disabled

http://shapeshifter.se/~fredde/tmp/db.acpi-cbb-uhci.disabled
http://shapeshifter.se/~fredde/tmp/dmesg.acpi-cbb-uhci.disabled

I also tried to load the usb kernelmoduels (usbd) after boot but that hangs the machine and a show intrcnt says irq10: uhci1 3391124, same irq-storm.

   Fredrik


On Mon, Dec 15, 2003 at 01:47:32PM -0500, John Baldwin wrote:
> 
> On 12-Dec-2003 Fredrik Lindberg wrote:
> > Hi,
> > Thank you for your reply.
> > The information you requested is avaiable at
> > 
> > http://shapeshifter.se/~fredde/tmp/db
> > http://shapeshifter.se/~fredde/tmp/dmesg.acpi-pci.enabled
> > 
> >   Fredrik 
> 
> I think your problem is cbb(4) related.  Probably cbb(4) is using an
> address that conflicts with another device and thus it's interrupt
> (on irq 10) is never being handled and it storms.  Try removing 'cbb'
> from your kernel as a test.  Warner (cc'd) can help you find a memory
> region for your cbb(4) device that you can set using 'hw.cbb.start_memory'
> from the loader.
> 
> > On Thu, Dec 11, 2003 at 01:42:06PM -0500, John Baldwin wrote:
> >> [ Resent due to NO.*SPAM bounce the first time ]
> >> 
> >> On 10-Dec-2003 Fredrik Lindberg wrote:
> >> > Hi,
> >> > 
> >> > I have a Compaq evo n160 running 
> >> > FreeBSD biocandy 5.2-RC FreeBSD 5.2-RC #10: Mon Dec  8 19:08:38 CET 2003
> >> > 
> >> > The machine fails (and always has) to boot with acpi enabled 
> >> > (locks up when mounting /), however, I managed to find out that 
> >> > booting with the following option 
> >> > 
> >> > debug.acpi.disable="pci" 
> >> > 
> >> > in /boot/loader.conf made the machine boot correctly and acpi related 
> >> > functions such as battry monitoring worked just fine.
> >> > 
> >> > But, and a huge but, no pci devices are detected during boot 
> >> > (maybe quite obvious because of that debug option) 
> >> > All pci-devices works perfectly with acpi disabled.
> >> > 
> >> > Now, is there any chance to make freebsd use acpi and the 
> >> > "normal" pci-bus driver at the same time, overriding the 
> >> > acpi pci-bus implementation?
> >> > 
> >> > I believe linux has a kernel option called pci=noacpi (atleast acording to google),
> >> > which does this.
> >> > 
> >> > With acpi enabled scanpci reports all the pci devices, but pciconf -l
> >> > doesn't return anything.
> >> > With acpi disabled, scanpci reports all pci devices, pciconf -l 
> >> > reports all devices.
> >> > 
> >> > dmesg output with acpi enabled 
> >> > http://shapeshifter.se/~fredde/tmp/dmesg.acpi.enabled
> >> > 
> >> > dmesg out with acpi disabled
> >> > http://shapeshifter.se/~fredde/tmp/dmesg.acpi.disabled
> >> > 
> >> > Any, and I mean any, help on this will be very appreciated.
> >> 
> >> If you can drop into ddb and do a 'show intrcnt' when the machine
> >> locks up that might help fix the hang.  It sounds like the interrupt
> >> routing may not have worked correctly.  Also, a dmesg of acpi
> >> with pci enabled would be helpful.
> >> 
> >> -- 
> >> 
> >> John Baldwin <>  <><  http://www.FreeBSD.org/~jhb/
> >> "Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/
> 
> -- 
> 
> John Baldwin <>  <><  http://www.FreeBSD.org/~jhb/
> "Power Users Use the Power to Serve!"  -  http://www.FreeBSD.org/