Booting of 200811 snapshot on Sun Fire V880

[Beat Gätzi]

Marius Strobl wrote:
> On Tue, Dec 09, 2008 at 11:06:55AM +0100, Beat Gtzi wrote:
>> The new image fixes the data access error but it seems that I run into 
>> another bug:
>>
>> isp1: <Qlogic ISP 10160 PCI SCSI Adapter> port 0x1000-0x10ff mem 
>> 0x2000000-0x2000fff at device 4.0 on pci3
>> pcib2: WARNING: using devices behind PCI-PCI bridges may cause data 
>> corruptionisp1: [ITHREAD]
>> isp1: Board Type 10160, Chip Revision 0x6, loaded F/W Revision 10.4.41
>> isp1: invalid NVRAM header
>> isp2: <Qlogic ISP 10160 PCI SCSI Adapter> port 0x1100-0x11ff mem 
>> 0x2002000-0x2002fff at device 5.0 on pci3
>> pcib2: WARNING: using devices behind PCI-PCI bridges may cause data 
>> corruptionisp2: [ITHREAD]
>> isp2: Board Type 10160, Chip Revision 0x6, loaded F/W Revision 10.4.41
>> isp2: invalid NVRAM header
>> pcib4: <Sun Host-PCI bridge> mem 
>> 0x40004e00000-0x40004e17fff,0x40004c10000-0x40004c1004f,0x7ffe8000000-0x7ffe80000ff 
>> irq
>>  626,624,625,628,630 on nexus0
>> pcib4: Schizo, version 4, IGN 0x9, bus A, 33MHz
>> Timecounter "pcib4" frequency 150000000 Hz quality 100
>> pcib4: DVMA map: 0xc0000000 to 0xffffffff
>> pcib4: [FILTER]
>> pcib4: [FILTER]
>> pcib4: [FILTER]
>> pcib4: [FILTER]
>> pci4: <OFW PCI bus> on pcib4
>> pci4: <bridge> at device 2.0 (no driver attached)
>> hme0: <Sun HME 10/100 Ethernet> mem 0x100000-0x107fff at device 2.1 on pci4
>> miibus2: <MII bus> on hme0
>> ukphy1: <Generic IEEE 802.3u media interface> PHY 1 on miibus2
>> ukphy1:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
>> hme0: Ethernet address: 00:03:ba:xx:xx:xx
>> pcib4: invalid interrupt vector 0x245
>> hme0: couldn't establish interrupt
>> panic: trap: fast data access mmu miss
>> cpuid = 0
>> KDB: enter: panic
>> [thread pid 0 tid 100000 ]
>> Stopped at      kdb_enter+0x80: ta              %xcc, 1
>> db> bt
>> Tracing pid 0 tid 100000 td 0xc0848780
>> panic() at panic+0x20c
>> trap() at trap+0x570
>> -- fast data access mmu miss tar=0 %o7=0xc0328a48 --
>> _rw_rlock() at _rw_rlock+0x170
>> in_pcbpurgeif0() at in_pcbpurgeif0+0x24
>> in_ifdetach() at in_ifdetach+0x1c
>> if_detach() at if_detach+0x180
>> ether_ifdetach() at ether_ifdetach+0x58
>> hme_detach() at hme_detach+0x64
>> hme_pci_attach() at hme_pci_attach+0x388
>> device_attach() at device_attach+0x4a4
>> device_probe_and_attach() at device_probe_and_attach+0x64
>> bus_generic_attach() at bus_generic_attach+0x10
>> ofw_pcibus_attach() at ofw_pcibus_attach+0x72c
>> device_attach() at device_attach+0x4a4
>> device_probe_and_attach() at device_probe_and_attach+0x64
>> bus_generic_attach() at bus_generic_attach+0x10
>> schizo_attach() at schizo_attach+0x13b8
>> device_attach() at device_attach+0x4a4
>> device_probe_and_attach() at device_probe_and_attach+0x64
>> bus_generic_attach() at bus_generic_attach+0x10
>> nexus_attach() at nexus_attach+0x4fc
>> device_attach() at device_attach+0x4a4
>> device_probe_and_attach() at device_probe_and_attach+0x64
>> root_bus_configure() at root_bus_configure+0x28
>> configure() at configure+0x4
>> mi_startup() at mi_startup+0x18c
>> btext() at btext+0x30
>> db>
>>
> 
> You have some real luck with this machine...

Yes and the story continues...

> The problem causing the panic seems to be a bug in the network
> stack which I don't have a clue about, which is triggered by
> hme(4) immediatley detaching again as it can't setup the
> interrupt due to what appears to be a firmware bug. Could
> you please try:
> http://people.freebsd.org/~marius/8.0-20081209-SNAP-sparc64-disc1.iso.gz
> which tries to work around the latter bug. Given that this
> involves some guesswork and I can't tell for sure it's a
> firmware bug without an OFW device tree dump (if there's
> still a working Solaris installation on that machine
> could you please provide the output of `prtconf -Ppv`?)
> I'm unsure whtether this will actually solve it.

With this image the server boots without problems but during the 
installer the server panics:

panic: cheetah_ipi_selected: couldn't send IPI
cpuid = 3
KDB: enter: panic

I was not able to enter to the debugger and it wasn't possible to send a 
break signal and return to OBP. I tried it several times and the panic 
occurred at various points during the installer.

Thanks,
Beat