Bug 4843 - irq 20: nobody cared
Summary: irq 20: nobody cared
Status: CLOSED PATCH_ALREADY_AVAILABLE
Alias: None
Product: ACPI
Classification: Unclassified
Component: Config-Interrupts (show other bugs)
Hardware: i386 Linux
: P2 normal
Assignee: Shaohua
URL:
Keywords:
: 4960 (view as bug list)
Depends on:
Blocks:
 
Reported: 2005-07-04 14:27 UTC by Alexey Dobriyan
Modified: 2005-08-08 18:29 UTC (History)
6 users (show)

See Also:
Kernel Version: 2.6.13-rc1
Subsystem:
Regression: ---
Bisected commit-id:


Attachments
/var/log/messages (15.30 KB, text/plain)
2005-07-04 14:28 UTC, Alexey Dobriyan
Details
Patch to fix VIA chipset problems caused by IRQ packing patch (575 bytes, text/plain)
2005-07-06 16:43 UTC, Natalie Protasevich
Details

Description Alexey Dobriyan 2005-07-04 14:27:16 UTC
From: Marko Kohtala <marko.kohtala@gmail.com>
http://marc.theaimsgroup.com/?t=112038967400001&r=1&w=2

2.6.12 worked fine, but soon after it got broken and was still broken just now
[Sun, 3 Jul 2005 14:16:53 +0300 --adobriyan] that I checked git version.

USB Universal Host Controller Interface driver v2.3
<7>Pin 2-12 already programmed
ACPI: PCI Interrupt 0000:00:11.2[D] -> GSI 12 (level, low) -> IRQ 20
PCI: Via IRQ fixup for 0000:00:11.2, from 12 to 4
uhci_hcd 0000:00:11.2: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller
uhci_hcd 0000:00:11.2: new USB bus registered, assigned bus number 4
uhci_hcd 0000:00:11.2: irq 20, io base 0x0000cc00
hub 4-0:1.0: USB hub found
hub 4-0:1.0: 2 ports detected
usb 3-1: new low speed USB device using ohci_hcd and address 2
irq 20: nobody cared (try booting with the "irqpoll" option)

Working kernel uses IRQ 12.
Comment 1 Alexey Dobriyan 2005-07-04 14:28:27 UTC
Created attachment 5264 [details]
/var/log/messages
Comment 2 Alexey Dobriyan 2005-07-04 14:31:19 UTC
From: Marko Kohtala <marko.kohtala@gmail.com>

Motherboard is MSI-6380 ver 2, i.e. K7T266 Pro2-RU 
Comment 3 Shaohua 2005-07-04 18:00:36 UTC
How about revert below patch?
http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-
2.6.git;a=commit;h=c434b7a6aedfe428ad17cd61b21b125a7b7a29ce
Comment 4 Marko Kohtala 2005-07-06 00:35:16 UTC
Yes, reverting this patch removed the problem.
http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=c434b7a6aedfe428ad17cd61b21b125a7b7a29ce

Thank you.

I'll be out of reach for few weeks now, so I will not be able to test this patch
in fixed form for you. After that then.
Comment 5 Shaohua 2005-07-06 00:49:23 UTC
The via irq quirk is done after mp_register_gsi, so the above patch breaks it. 
I'll ping Natilie.
Comment 6 Natalie Protasevich 2005-07-06 16:43:11 UTC
Created attachment 5292 [details]
Patch to fix VIA chipset problems caused by IRQ packing patch
Comment 7 Natalie Protasevich 2005-07-06 16:50:32 UTC
Recently, I had another complaint and the patch attached fixed it for the other 
case. The VIA chipset uses 4-bit register for its internal routing, which 
limits it to IRQs below 16. This is a usual ISA range, but IRQs 9,10, 11, and 
13 are not used by traditional ISA devices, and that's what VIA relies upon. My 
patch starts distributing PCI IRQs starting 16, and this breaks VIA.
Marko, please test the patch and let me know if it worked for you.
Thanks,
--Natalie
Comment 8 Marko Kohtala 2005-07-20 04:40:27 UTC
Seems to work with the patch to fix VIA.

Thank you Natalie.
Comment 9 Andrew Morton 2005-07-28 23:20:11 UTC
Natalie's patch is still not present in 2.6.13-rc4!

Could someone please get a copy over to me asap?
Comment 10 Shaohua 2005-07-29 00:47:06 UTC
*** Bug 4960 has been marked as a duplicate of this bug. ***
Comment 11 Natalie Protasevich 2005-07-29 07:02:37 UTC
Andrew, you already have the patches in the -mm tree, they are:
x86-avoid-wasting-irqs-patch-update.patch
x86_64-avoid-wasting-irqs-patch-update.patch

Thanks,
--Natalie
Comment 12 Andrew Morton 2005-07-29 11:36:17 UTC
Ah, thanks.  I'll merge the x86 one and I've asked Andi to ack the x86_64
fix for 2.6.13 as well.

Comment 13 Adrian Bunk 2005-08-08 18:29:31 UTC
The patches are in 2.6.13-rc6.

Note You need to log in before you can comment on or make changes to this bug.