John Baldwin
2014-Dec-22 14:57 UTC
FreeBSD 10.1-amd64 -> booting on a HP DL380 Gen9 results in panic
On Thursday, December 18, 2014 7:27:13 pm Rainer Duffner wrote:> Hi, > > we got one to test and it booted using the UEFI memory stick image. > > However, I get a panic after > Event time ?LAPIC? quality 600 > ACPI APIC Table: <HP Proliant> > panic: APIC: CPU with APIC ID 0 is not enabled > cpuid = 0 > > and then a stack backtrace > > > What does that mean? > > AFAIK, I have a single E5-2620V3 CPU and 16 GB RAM in there. > > > It?s primarily intended as a test-system - but earlier or later I will haveto put one into production because we?ll likely stop procuring Gen8 systems sometime next year (when they simply stop becoming available).> > I haven?t tried a snapshot of current.Eh, the table that ACPI provides that lists the available CPUs in the system claims that that the CPU that the kernel booted from is disabled. We assume that the boot processor is valid and enabled (since it is running the kernel already!) This is almost certainly a firmware bug. Booting without UEFI would be a good test as Mark suggested. If you can capture a verbose dmesg, that would also include enough details about what we found in the ACPI table to debug this further perhaps. -- John Baldwin
Nagy, Attila
2015-Feb-25 15:22 UTC
FreeBSD 10.1-amd64 -> booting on a HP DL380 Gen9 results in panic
Hi, On 12/22/14 15:57, John Baldwin wrote:> On Thursday, December 18, 2014 7:27:13 pm Rainer Duffner wrote: >> Hi, >> >> we got one to test and it booted using the UEFI memory stick image. >> >> However, I get a panic after >> Event time ?LAPIC? quality 600 >> ACPI APIC Table: <HP Proliant> >> panic: APIC: CPU with APIC ID 0 is not enabled >> cpuid = 0 >> >> and then a stack backtrace >> >> >> What does that mean? >> >> AFAIK, I have a single E5-2620V3 CPU and 16 GB RAM in there. >> >> > > Eh, the table that ACPI provides that lists the available CPUs in the system > claims that that the CPU that the kernel booted from is disabled. We assume > that the boot processor is valid and enabled (since it is running the kernel > already!) This is almost certainly a firmware bug. Booting without UEFI > would be a good test as Mark suggested. If you can capture a verbose dmesg, > that would also include enough details about what we found in the ACPI table > to debug this further perhaps. >One of my colleagues have captured a boot -v output for this machine. One line (where it prints the CPU) is missing, the others are intact: https://plus.google.com/photos/104147045962330059540/albums/6119801748474522833/6119801755965284786?banner=pwa&pid=6119801755965284786&oid=104147045962330059540 The machine is PXE-booted. I can try anything to help fixing this issue. UEFI and AFAIK the x2APIC support, which Rainer mentioned is disabled here.