Linux boot issue: "8254 timer not connected to IO-APIC"

Linux boot issue: "8254 timer not connected to IO-APIC"

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 1293 bytes --]

Hi,

from time to time Linux fails to boot in my kvm boxes, complaining like
this:

[...]
PAT WC disabled due to known CPU erratum.
CPU: L1 I cache: 32K, L1 D cache: 32K
CPU: L2 cache: 2048K
CPU 0/0 -> Node 0
ACPI: Core revision 20080609
Parsing all Control Methods:
Table [DSDT](id 0001) - 295 Objects with 45 Devices 104 Methods 6 Regions
 tbxface-0596 [00] tb_load_namespace     : ACPI Tables successfully acquired
evxfevnt-0091 [00] enable                : Transition to ACPI mode successful
..TIMER: vector=0x30 apic1=0 pin1=0 apic2=-1 pin2=-1
..MP-BIOS bug: 8254 timer not connected to IO-APIC
...trying to set up timer (IRQ0) through the 8259A ...
..... (found apic 0 pin 0) ...
....... failed.
timer doesn't work through the IO-APIC - disabling NMI Watchdog!
...trying to set up timer as Virtual Wire IRQ...
..... failed.
...trying to set up timer as ExtINT IRQ...
..... failed :(.
Kernel panic - not syncing: IO-APIC + timer doesn't work!  Boot with apic=debug and send a report.  Then try booting with the 'noapic' option.

The problem seems to be that timer_irq_works() fails, ie. less than 4
out of 10 PIC ticks made it to the guest within that test window. Gleb,
is this something your anti-drift patches for the PIC would address as
well?

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 257 bytes --]

Re: Linux boot issue: "8254 timer not connected to IO-APIC"

[Gleb Natapov]

On Sun, Nov 02, 2008 at 12:58:52PM +0100, Jan Kiszka wrote:
> Hi,
> 
> from time to time Linux fails to boot in my kvm boxes, complaining like
> this:
> 
> [...]
> PAT WC disabled due to known CPU erratum.
> CPU: L1 I cache: 32K, L1 D cache: 32K
> CPU: L2 cache: 2048K
> CPU 0/0 -> Node 0
> ACPI: Core revision 20080609
> Parsing all Control Methods:
> Table [DSDT](id 0001) - 295 Objects with 45 Devices 104 Methods 6 Regions
>  tbxface-0596 [00] tb_load_namespace     : ACPI Tables successfully acquired
> evxfevnt-0091 [00] enable                : Transition to ACPI mode successful
> ..TIMER: vector=0x30 apic1=0 pin1=0 apic2=-1 pin2=-1
> ..MP-BIOS bug: 8254 timer not connected to IO-APIC
> ...trying to set up timer (IRQ0) through the 8259A ...
> ..... (found apic 0 pin 0) ...
> ....... failed.
> timer doesn't work through the IO-APIC - disabling NMI Watchdog!
> ...trying to set up timer as Virtual Wire IRQ...
> ..... failed.
> ...trying to set up timer as ExtINT IRQ...
> ..... failed :(.
> Kernel panic - not syncing: IO-APIC + timer doesn't work!  Boot with apic=debug and send a report.  Then try booting with the 'noapic' option.
> 
> The problem seems to be that timer_irq_works() fails, ie. less than 4
> out of 10 PIC ticks made it to the guest within that test window. Gleb,
> is this something your anti-drift patches for the PIC would address as
> well?
> 
If the problem is indeed interrupt coalescing then anti-drift patches
will help.

--
			Gleb.