atkbdc broken on current ?

[John Baldwin]

On Sunday, June 19, 2011 7:05:04 pm Damjan Marion wrote:
> 
> On Jun 17, 2011, at 11:21 PM, John Baldwin wrote:
> 
> > On Friday, May 06, 2011 11:47:33 am John Baldwin wrote:
> >> On Thursday, May 05, 2011 5:04:54 pm Damjan Marion wrote:
> >>> 
> >>> On May 5, 2011, at 7:43 PM, John Baldwin wrote:
> >>> 
> >>>> On Thursday, May 05, 2011 9:21:04 am Damjan Marion wrote:
> >>>>> 
> >>>>> Hi,
> >>>>> 
> >>>>> I have issue with old HP DL380G3 server. When I use ILO virtual console to 
> >>>> manage server. Seems that 9-CURRENT fails to detect atkbdc.
> >>>>> When I boot 8.2-RELEASE it works well.
> >>>>> 
> >>>>> 8.2 dmesg shows:
> >>>>> 
> >>>>> atkbdc0: <Keyboard controller (i8042)> port 0x60,0x64 irq 1 on acpi0
> >>>>> 
> >>>>> 9.0:
> >>>>> 
> >>>>> atkbdc0: <Keyboard controller (i8042)> failed to probe at port 0x60 on isa0
> >>>>> 
> >>>>> Is this a known issue?
> >>>>> 
> >>>>> Should I enable some additional outputs, like KBDIO_DEBUG?
> >>>> 
> >>>> I suspect this is a resource issue stemming from changes I made to the acpi(4) 
> >>>> bus driver quite a while ago to make it use rman_reserve_resource().  Can you
> >>>> capture a full verbose dmesg from 9 along with devinfo -rv and devinfo -ur 
> >>>> output from 9?
> >>> 
> >>> Here it is:
> >>> 
> >>> http://web.me.com/dmarion/atkbdc.txt
> >> 
> >> Ohh, hmm.  Your BIOS has done "odd" things:
> >> 
> >>        isab0 pnpinfo vendor=0x1166 device=0x0201 subvendor=0x1166 subdevice=0x0201 class=0x060100 at slot=15 function=0 
handle=\_SB_.PCI0.IBRG
> >>          isa0
> >>              I/O ports:
> >>                  0x0-0xf
> >>                  0x20-0x21
> >>                  0x40-0x43
> >>                  0x60
> >>                  0x61
> >>                  0x64
> >>                  0x80-0x8f
> >>                  0xa0-0xa1
> >>                  0xc0-0xdf
> >>                  0x4d6
> >> 
> >> Still, I don't know how the ISA bus is actually allocating resources.  Can
> >> you add some code to the x86 nexus driver to drop into kdb when it receives
> >> a SYS_RES_IOPORT allocation request from "isa0" and get a stack trace from
> >> DDB and reply with the trace?
> > 
> > So I think I just found the explanation for this and I think the change I
> > just committed will fix your system:
> > 
> > Author: jhb
> > Date: Fri Jun 17 21:19:01 2011
> > New Revision: 223207
> > URL: http://svn.freebsd.org/changeset/base/223207
> 
> Hi John,
> 
> sorry for not coming back to you on your previous email, it was on my todo
> list but I haven't time to do it.
> 
> I tried now new kernel and it works well. Thanks for fixing this.

Thanks.  Can you capture an acpdump from your box when you get a chance and
send it to me?  I want to make sure if I make any other changes in this area
that your box will still work.

-- 
John Baldwin