David van Hoose reports that interrupts broke in 2.4.27-pre3: interrupts-2.4.27-pre2: CPU0 0: 6956 IO-APIC-edge timer 1: 136 IO-APIC-edge keyboard 2: 0 XT-PIC cascade 7: 2 IO-APIC-edge parport0 8: 1 IO-APIC-edge rtc 14: 25 IO-APIC-edge ide0 15: 683 IO-APIC-edge ide1 16: 11868 IO-APIC-level ft3xx, nvidia 17: 2 IO-APIC-level ohci1394 18: 0 IO-APIC-level SiS SI7012 19: 20 IO-APIC-level eth0 20: 41 IO-APIC-level acpi, usb-ohci 21: 0 IO-APIC-level usb-ohci 22: 72 IO-APIC-level usb-ohci 23: 0 IO-APIC-level ehci_hcd NMI: 0 LOC: 6907 ERR: 0 MIS: 0 interrupts-2.4.27-pre3 CPU0 0: 46794 local-APIC-edge timer 1: 1760 IO-APIC-edge keyboard 7: 2 IO-APIC-edge parport0 8: 1 IO-APIC-edge rtc 14: 25 IO-APIC-edge ide0 15: 683 IO-APIC-edge ide1 16: 18090 IO-APIC-level ft3xx, nvidia 17: 3 IO-APIC-level ohci1394 18: 0 IO-APIC-level SiS SI7012 19: 30 IO-APIC-level eth0 20: 0 none acpi, usb-ohci 21: 0 IO-APIC-level usb-ohci 22: 67 IO-APIC-level usb-ohci 23: 0 IO-APIC-level ehci_hcd NMI: 0 LOC: 46744 ERR: 0 MIS: 0 IRQ20 has no handler type, and the devices attached do not respond.
Created attachment 3079 [details] 2.4.27-pre3 test patch The root cause appears to be that 2.4/i386 dynamically allocates mp_irqs[] so the fix to bug 2520 caused us to write into mp_irqs[] before it was allocated and thus the interrupt source over-ride information was lost. This patch simply allocates mp_irqs[] earlier.
shipped in 2.4.27 - closing