Bug 12382
Summary: | Asus M2A-VM shows IRQ nobody cared due to wrongly derived IRQ from PCI bridge | ||
---|---|---|---|
Product: | ACPI | Reporter: | Thomas Renninger (trenn) |
Component: | Config-Interrupts | Assignee: | acpi_config-interrupts |
Status: | REJECTED INVALID | ||
Severity: | normal | CC: | bjorn.helgaas, jbarnes, lenb, shaohua.li, tj |
Priority: | P1 | ||
Hardware: | All | ||
OS: | Linux | ||
Kernel Version: | Probably all which are ACPI capable | Subsystem: | |
Regression: | --- | Bisected commit-id: | |
Attachments: |
Original DSDT.dsl of m2a-vm BIOS
Diff between m2a-vm and m2a-vm-hdmi mainboard's DSDTs which fixes the issue Conservative patch fixing the issue for the ASUS m2a-vmi and providing a boot param for others General fix and Windows adoption. |
Description
Thomas Renninger
2009-01-07 14:56:43 UTC
Created attachment 19705 [details]
Original DSDT.dsl of m2a-vm BIOS
Created attachment 19706 [details]
Diff between m2a-vm and m2a-vm-hdmi mainboard's DSDTs which fixes the issue
Defines an IRQ for the ohci1394 device.
Created attachment 19707 [details]
Conservative patch fixing the issue for the ASUS m2a-vmi and providing a boot param for others
Could you please review this one also please.
If this is more generally solved, I still like to add this one for SLE11 for now.
Based on 2.6.27.10
Created attachment 19708 [details]
General fix and Windows adoption.
irq derive is a must for pcmcia card, as pcmcia card hasn't _PRT entry and has the same irq as its pcmcia bridge. PCI spec actually defines how to do irq derive, and linux follows the spec. If you want to have a generic solution, I thought you can figure out a debug patch (not remove derive for pcmcia card) and saw if it breaks anything. Since the box has no firewire connector, is there any reason to load the ohci1394 module? If we don't load ohci1394, I assume there's no "IRQ: nobody cared" problem. Maybe we can just prevent ohci1394 from binding to the device. I'm not sure how best to do that. One slightly hacky way would be an early PCI quirk that checks for M2A-VM DMI info and sets dev->hdr_type to something invalid so pci_setup_device() would ignore the device. What a confusion...: This is a fixed BIOS bug. We have a m2a-vm-hdmi board, not a m2a-vm. m2a-vm-hdmi 1501 BIOS had the firewire port not configured, thus the firewire driver was using the wrong IRQ -> ahci IRQ got disabled. This got fixed between BIOS 1501 and 1701. The m2a-vm-hdmi dmi info states m2a-vm at one place, thus we thought we have a m2a-vm board. The next BIOS bug was that we could update our -hdmi board with the wrong m2a-vm BIOS (got fixed with 2201 BIOS :) ). Because the m2a-vm board does not have firewire, the wrong BIOSes again did not configure the firewire IRQ and we still saw the exactly same error -> firewire driver used wrong IRQ -> ahci breaks. Solution: Update to the correct latest BIOS and things work -> bug is invalid. Sorry for the confusion... |