Bug#366507: Possible workaround
This is probably a BIOS bug. What's happening is that the BIOS is
reporting that a particular interrupt routing is in place, when it
actually isn't.
As PCI interrupts are level rather than edge triggered, if the device
interrupting doesn't get serviced, the interrupt will happen again next
time interrupts are enabled. If the device is claiming it's connected to
IRQ X but actually triggers IRQ Y, the interrupt will not be serviced at
all, and after 10,000 unclaimed interrupts Linux will disable that
interrupt line.
My workaround was to set the Laptop's BIOS so that all the PCI interrupt
links (A, B, C, and D) route to the same ISA IRQ. That forces Linux to
poll all the devices, which works.
--
Paul Martin <pm@debian.org>
Reply to: