Results 1 to 10 of 10
Hi all,
I've implemented a PCI driver with MSI interrupt.
For the first time I load my module (insmod ...), my driver works fine. I see expected interrupts from my ...
- 11-20-2010 #1Just Joined!
- Join Date
- Nov 2010
- Posts
- 11
message: "No irq handler for vector"
Hi all,
I've implemented a PCI driver with MSI interrupt.
For the first time I load my module (insmod ...), my driver works fine. I see expected interrupts from my card.
However, when I remove module (rmmod) and insert again, dmesg gives me this weird "No irq handler for vector (irq -1) .." error although the module has successfully requested IRQ NUMBER and registered the ISR.
I've encountered this messages since kernel 2.6.32, pretty sure that 2.6.31 didn't give me this thing. I've upgraded to 2.6.48, error is still there.
Some websites suggest disabling MSI by having boot option (pci = nomsi) ... which i don't think it fixes the root cause.
What's really the cause and why is it introduced since 2.6.32?
Thanks, guys.
Jeff
- 11-20-2010 #2Just Joined!
- Join Date
- Apr 2005
- Location
- Romania
- Posts
- 42
Hi jefflieu,
Try to compile a new kernel enabling:
PCI access support
PCI access mode (Direct) --->
- 11-21-2010 #3Linux Guru
- Join Date
- Apr 2009
- Location
- I can be found either 40 miles west of Chicago, or in a galaxy far, far away.
- Posts
- 8,961
When your module is removed, do you release the IRQ it was using? My guess is that you are not properly releasing resources like this when your module is shut down and removed from the kernel.
Sometimes, real fast is almost as good as real time.
Just remember, Semper Gumbi - always be flexible!
- 11-22-2010 #4Just Joined!
- Join Date
- Nov 2010
- Posts
- 11
Hi, thanks for the reply.
I'm not sure i'm doing the right thing. It did work before. The remove sequence in the driver:
- Disable interrupt generation at the card
- Free irq line: disable_irq(irq_line), irq_line was obtained in the init function, after enabling msi by reading: pcidev->irq.
- Disable MSI: pci_disable_msi(pcidev)
By the way, Init function:
- pci_enable_msi(pcidev)
- irq_line = pcidev->irq
- request_irq(irq_line, pcidev_isr_handler,IRQF_SHARED, DRV_NAME, (void*)mypci);
Thanks very much.
Jeff
- 11-22-2010 #5
- 11-22-2010 #6Just Joined!
- Join Date
- Nov 2010
- Posts
- 11
When I re-insert the module, the request_irq returns SUCCESS.
- 11-22-2010 #7Linux Guru
- Join Date
- Apr 2009
- Location
- I can be found either 40 miles west of Chicago, or in a galaxy far, far away.
- Posts
- 8,961
So, did cbogdan's suggestion work? It sounds like it.
Sometimes, real fast is almost as good as real time.
Just remember, Semper Gumbi - always be flexible!
- 11-23-2010 #8Just Joined!
- Join Date
- Nov 2010
- Posts
- 11
- 11-23-2010 #9Just Joined!
- Join Date
- Nov 2010
- Posts
- 11
Since i have the source, any pointers to which particular areas to look at is really appreciated. I'm quite new to kernel "excavation".
Thanks
- 11-24-2010 #10Linux Guru
- Join Date
- Apr 2009
- Location
- I can be found either 40 miles west of Chicago, or in a galaxy far, far away.
- Posts
- 8,961
Well, there were a number of changes in the kernel with the 2.6.32 kernel that could affect this. There should be some documentation on the kernel web site (The Linux Kernel Archives) on the changes you need to make in your driver if the changes impact what you are doing. Sorry, but that's about as far as I can take you right now. I haven't got into that level of detail on the changes that were instituted, other than reading some newsletter articles about it.
Sometimes, real fast is almost as good as real time.
Just remember, Semper Gumbi - always be flexible!


Reply With Quote

