After resume from suspend to RAM dmesg on my HP 2510p notebook shows: irq 19: nobody cared (try booting with the "irqpoll" option) Pid: 0, comm: swapper Not tainted 2.6.28-rc7 #67 Call Trace: [] ? ohci_irq_handler+0x60/0x7e9 [ohci1394] [] __report_bad_irq+0x38/0x87 [] note_interrupt+0x10e/0x174 [] handle_fasteoi_irq+0xa7/0xd1 [] do_IRQ+0x73/0xe4 [] ret_from_intr+0x0/0xa [] ? acpi_idle_enter_bm+0x26b/0x2b2 [processor] [] ? acpi_idle_enter_bm+0x261/0x2b2 [processor] [] ? notifier_call_chain+0x33/0x5b [] ? cpuidle_idle_call+0x8c/0xc4 [] ? cpu_idle+0x4a/0x9a [] ? rest_init+0x5c/0x5e handlers: [] (ohci_irq_handler+0x0/0x7e9 [ohci1394]) Disabling IRQ #19 Attached a dmesg that includes one suspend/resume cycle. I only get the error once on resume, not at any other time. While looking at the suspend and resume functions I noticed that the suspend function calls free_irq before saving the PCI config, but there is no corresponding request_irq in the resume function. So I naively tried the following patch: diff --git a/drivers/firewire/fw-ohci.c b/drivers/firewire/fw-ohci.c index 46610b0..3f283df 100644 --- a/drivers/firewire/fw-ohci.c +++ b/drivers/firewire/fw-ohci.c @@ -2570,6 +2570,14 @@ static int pci_resume(struct pci_dev *dev) fw_error("pci_enable_device failed\n"); return err; } + if (request_irq(dev->irq, irq_handler, + IRQF_SHARED, ohci_driver_name, ohci)) { + fw_error("Failed to allocate shared interrupt %d.\n", + dev->irq); + dma_free_coherent(ohci->card.device, CONFIG_ROM_SIZE, + ohci->config_rom, ohci->config_rom_bus); + return -EIO; + } return ohci_enable(&ohci->card, NULL, 0); } And, hé presto, the error disappeared, and that's the only relevant diff with a kernel without the patch. Only after testing that did I notice that I had copied the call to request_irq from ohci_enable, but that function is already getting called immediately afterwards! So it seems that the real problem is a timing issue and request_irq is somehow being called too late during resume? Could it be that the device is left in a half-active state (or that its PCI config is saved while in a half-active state), resulting in IRQs getting through before request_irq is executed in ohci_enable? Note that I don't actually use firewire. Note also that both with and without the error IRQ 19 is listed the same in /proc/interrupts (and even with rather a high count for a device that is not in use; this is shortly after a clean reboot + suspend/resume). Cheers, FJP $ cat /proc/interrupts CPU0 CPU1 0: 18744 10269 IO-APIC-edge timer 1: 115 41 IO-APIC-edge i8042 8: 1 0 IO-APIC-edge rtc0 9: 454 152 IO-APIC-fasteoi acpi 12: 342 139 IO-APIC-edge i8042 14: 4243 3517 IO-APIC-edge ata_piix 15: 0 0 IO-APIC-edge ata_piix 16: 46 11 IO-APIC-fasteoi uhci_hcd:usb1, i915@pci:0000:00:02.0 17: 1780 1583 IO-APIC-fasteoi uhci_hcd:usb3, HDA Intel 18: 2 2 IO-APIC-fasteoi ehci_hcd:usb2, uhci_hcd:usb7, yenta 19: 99968 33 IO-APIC-fasteoi ohci1394 20: 33 12 IO-APIC-fasteoi ehci_hcd:usb4, uhci_hcd:usb5, mmc0 22: 0 0 IO-APIC-fasteoi uhci_hcd:usb6 508: 1335 1039 PCI-MSI-edge iwlagn NMI: 0 0 Non-maskable interrupts LOC: 10382 14256 Local timer interrupts RES: 10016 9596 Rescheduling interrupts CAL: 173 36 Function call interrupts TLB: 41 23 TLB shootdowns TRM: 0 0 Thermal event interrupts THR: 0 0 Threshold APIC interrupts SPU: 0 0 Spurious interrupts ERR: 1 MIS: 0