DMA and IRQ Issues in Gentoo with Kernel 2.6.34-gentoo-r12

[mark@baggywrinkle.co.uk]

I have Gentoo installed in virtual machine on a MAC Pro. The MAC is running 10.6.6 and Parallels 6.0.11990 Rev 621014. I dropped back to here due to disk corruption issues. The issues below I also had on the later versions of Parallels.

During the creation of a RAID array within the virtual machine and during heavy disk usage I am getting ATA READ DMA errors as follows... These are on disks configured in the guest as SATA, module ata_piix and ahci are being used for the SATA and IDE disks.

All disks are files set to 64Gb, and have no problems storing or retrieving data under OS X


ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
ata2.00: failed command: READ DMA
ata2.00: cmd c8/00:48:00:29:5e/00:00:00:00:00/e0 tag 0 dma 36864 in
res 40/00:01:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
ata2.00: status: { DRDY }
ata2: hard resetting link
ata2: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
ata2.00: configured for UDMA/100
ata2.00: device reported invalid CHS sector 0
ata2: EH complete


After a period I also get IRQ16 being disabled...

irq 16: nobody cared (try booting with the "irqpoll" option)
Pid: 0, comm: swapper Not tainted 2.6.34-gentoo-r12 #1
Call Trace:
<IRQ> [<ffffffff810523bd>] ? __report_bad_irq+0x30/0x7d
[<ffffffff81052511>] ? note_interrupt+0x107/0x16e
[<ffffffff81052c2f>] ? handle_fasteoi_irq+0x95/0xb5
[<ffffffff810046a6>] ? handle_irq+0x17/0x1d
[<ffffffff81003d1a>] ? do_IRQ+0x54/0xb9
[<ffffffff81394313>] ? ret_from_intr+0x0/0xa
[<ffffffff81031024>] ? __do_softirq+0x54/0x10c
[<ffffffff810161e7>] ? ack_apic_level+0x86/0x1dc
[<ffffffff81002c4c>] ? call_softirq+0x1c/0x28
[<ffffffff8100465d>] ? do_softirq+0x31/0x63
[<ffffffff81003d69>] ? do_IRQ+0xa3/0xb9
[<ffffffff81394313>] ? ret_from_intr+0x0/0xa
<EOI> [<ffffffff810089f7>] ? default_idle+0x20/0x34
[<ffffffff8100117a>] ? cpu_idle+0x45/0x7a
[<ffffffff816d1c6f>] ? start_kernel+0x357/0x362
[<ffffffff816d1381>] ? x86_64_start_kernel+0xe5/0xe9
handlers:
[<ffffffffa019a8f8>] (ahci_interrupt+0x0/0x5ca [ahci])
Disabling IRQ #16



I also get issues on the IDE disks as follows...

hda: ide_dma_sff_timer_expiry: DMA status (0x61)
hda: DMA timeout error
hda: dma timeout error: status=0x50 { DriveReady SeekComplete }
hda: possibly failed opcode: 0x25

these had been configured automatically to UDMA/100

hda: UDMA/100 mode selected
hdb: UDMA/25 mode selected
hdc: UDMA/100 mode selected

Yet SCSI seems to be rock solid.

Help, what is causing this...

 

[mark@baggywrinkle.co.uk]

Having done more checking.. it seems the issue is related to support for Multiple Processors in the Virtual Machine. Took Parallels back to release 11882, built a new VM and set number of processors to 1 and no problems; Set it to 2 and the problems came back. The setting of enable adaptive hypervisor has no impact on the behaviour.


Have also tested the setup with VMware Fusion and with dual CPU defined in a virtual machine, no issues. Will say that VMware only supports SCSI disks and not SATA.

Seems that there is something within Parallels that is an issue with interrupts and DMA when multiple CPU's are defined.