Skip to content

x86_64 irq fixes

Eric W. Biederman lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest3 patches 2 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/3i386/x86_64: FIX pci_enable_irq to set dev->irq to the irq numberarch/i386/pci/irq.c+0 -40 replies
2/3i386/x86_64: Remove global IO_APIC_VECTORarch/i386/kernel/io_apic.c arch/x86_64/kernel/io_apic.c include/asm-i386/hw_irq.h +1 more+10 -150 replies
3/3x86_64 irq: Allocate a vector across all cpus for genapic_flat.arch/x86_64/kernel/genapic_cluster.c arch/x86_64/kernel/genapic_flat.c arch/x86_64/kernel/io_apic.c +2 more+117 -481 reply

Reviews trailers per patch

1 of 3 patches has a Reviewed-by or Acked-by of their own; the series has one for all of them.

PatchMuli Ben-Yehuda
SeriesA-b
1/3
2/3
3/3A-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/3] x86_64 irq fixes
Quoted 5 lines · Muli Ben-Yehuda
Quoted 2 lines · Eric W. Biederman

Can you try CONFIG_CPU_HOTPLUG? That will force genapic to be set to genapic_physflat instead of genapic_flat.

Quoted 7 lines

Yep, it boots with CONFIG_CPU_HOTPLUG!

Quoted 2 lines

If genapic_physflat works we will have to decide what to do about genapic_flat.

I'm happy to test any follow-on patches to make it work without CONFIG_CPU_HOTPLUG.

Ok. I have found a fairly clean way to structure the code that should restore the previous behavior of the genapic_flat allowing lowest priority interrupt delivery to work, and getting lucky and avoiding your hardware that does not do what the software tells it to :)

I still need to dig in and remove the BUG_ON in the interrupt reception path, but that is a separate problem.

I also found another small bug in the pci_enable_irq because of some code I failed to remove earlier, and the patches overlap so I have made this a small patch series.

I have tested the code as best I can, and confirmation that this fixes the original problem would be great. But I don't see how it could fail to fix the problem, as it restores genapic_flat to global vector allocation.

Eric