Bug 87531 - do_IRQ: 0.166 No irq handler for vector (irq -1)
Summary: do_IRQ: 0.166 No irq handler for vector (irq -1)
Status: NEW
Alias: None
Product: Platform Specific/Hardware
Classification: Unclassified
Component: i386 (show other bugs)
Hardware: i386 Linux
: P1 blocking
Assignee: platform_i386
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-11-02 10:03 UTC by Elmar Stellnberger
Modified: 2016-02-18 12:33 UTC (History)
4 users (show)

See Also:
Kernel Version: 3.17.1-1-ARCH #1 SMP PREEMPT Wed Oct 15 15:36:07 CEST 2014
Subsystem:
Regression: No
Bisected commit-id:


Attachments
dmesg (49.47 KB, text/plain)
2014-11-02 10:04 UTC, Elmar Stellnberger
Details
/proc/interrupts (1.18 KB, text/plain)
2014-11-02 10:04 UTC, Elmar Stellnberger
Details
/proc/irq/* (3.97 KB, text/plain)
2014-11-02 10:05 UTC, Elmar Stellnberger
Details
lspci (1.26 KB, text/plain)
2014-11-02 10:05 UTC, Elmar Stellnberger
Details
cpuinfo: 2xPIII800 (1002 bytes, text/plain)
2014-11-10 15:14 UTC, Elmar Stellnberger
Details
/proc/interrupts (2xCPU) (1.43 KB, text/plain)
2014-11-10 15:15 UTC, Elmar Stellnberger
Details
/proc/irq/* (2xCPU) (3.45 KB, text/plain)
2014-11-10 15:16 UTC, Elmar Stellnberger
Details
dmesg (2xCPU) (67.18 KB, text/plain)
2014-11-10 15:19 UTC, Elmar Stellnberger
Details
lspci -vvvxx (2xCPU) (22.76 KB, text/plain)
2014-11-10 15:23 UTC, Elmar Stellnberger
Details

Description Elmar Stellnberger 2014-11-02 10:03:54 UTC
My Promise SATA 300 PCI card seems to cause irq problems when run with Linux no matter what other PCI cards I use (tried multiple configurations). The monitor displays "do_IRQ: No IRQ hadnler for vector (irq -1)" from time to time and the keyboard can get blocked (does no more react on key strokes or every character is displayed twice). In order to extract the needed information I have booted several times and connected two keyboards (1USB, 1PS/2 but both can get blocked). I have also tried it with different mainboards but Linux does not seem to like that PCI card.
Comment 1 Elmar Stellnberger 2014-11-02 10:04:23 UTC
Created attachment 156111 [details]
dmesg
Comment 2 Elmar Stellnberger 2014-11-02 10:04:44 UTC
Created attachment 156121 [details]
/proc/interrupts
Comment 3 Elmar Stellnberger 2014-11-02 10:05:02 UTC
Created attachment 156131 [details]
/proc/irq/*
Comment 4 Elmar Stellnberger 2014-11-02 10:05:23 UTC
Created attachment 156141 [details]
lspci
Comment 5 Elmar Stellnberger 2014-11-10 15:14:01 UTC
Created attachment 157141 [details]
cpuinfo: 2xPIII800

  After having happily installed a second CPU, now everything works fine like by a miracle; SATA works well, no more messages 'No IRQ handler for vector (irq -1)', no keyboard that would refuse to work.
Comment 6 Elmar Stellnberger 2014-11-10 15:15:01 UTC
Created attachment 157151 [details]
/proc/interrupts (2xCPU)
Comment 7 Elmar Stellnberger 2014-11-10 15:16:09 UTC
Created attachment 157161 [details]
/proc/irq/* (2xCPU)
Comment 8 Elmar Stellnberger 2014-11-10 15:19:02 UTC
Created attachment 157171 [details]
dmesg (2xCPU)

 Now it even works with a kernel which is a little bit older. However I can remember this bug having already persisted for a long time before I have filed this report; i.e. the observed effects won`t depend on the kernel version.
Comment 9 Elmar Stellnberger 2014-11-10 15:23:09 UTC
Created attachment 157181 [details]
lspci -vvvxx (2xCPU)

The board is an Asus P2B-DS; it requires 2CPUs to work with this card. Other boards supporting only one CPU will not work at all with this Promise SATA300 card under Linux as far as I could test.
Comment 10 Tejun Heo 2015-01-05 16:31:44 UTC
Looks like PCI IRQ assignment somehow gets screwed w/ single CPU. Bjorn, any ideas?

Thanks.

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