Page MenuHomeFreeBSD

pci: Only re-route IRQs based on firmware on x86
ClosedPublic

Authored by cperciva on Mar 28 2025, 6:12 PM.
Tags
None
Referenced Files
Unknown Object (File)
Tue, Apr 7, 4:26 PM
Unknown Object (File)
Sun, Apr 5, 2:08 PM
Unknown Object (File)
Sun, Apr 5, 6:45 AM
Unknown Object (File)
Sat, Apr 4, 12:25 AM
Unknown Object (File)
Thu, Apr 2, 4:00 AM
Unknown Object (File)
Mon, Mar 30, 1:04 AM
Unknown Object (File)
Sun, Mar 29, 5:38 AM
Unknown Object (File)
Sat, Mar 28, 9:05 PM

Details

Summary

There is a (very historical) call to pci_assign_interrupt for the
purpose of routing IRQs which may have been set up wrong by x86 BIOS
or firmware. On non-x86 systems, this is unnecessary; and on INTRNG
systems it results in a (synthetic) IRQ leak and ultimately a kernel
panic after many hotplug/unplug cycles.

Suggested by: jhb
MFC after: 2 weeks
Sponsored by: Amazon

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Skipped
Unit
Tests Skipped
Build Status
Buildable 63186
Build 60070: arc lint + arc unit

Event Timeline

cperciva created this revision.
This revision is now accepted and ready to land.Mar 28 2025, 6:43 PM

Could it bring a new flow of messages into the kernel message buffer? After today's upgrade of my desktop PC, I see plenty of "Can't enable IRQ/MSI because no handler is installed," which didn't appear in the past.

It was triggered by D49543 (likely) and seems to be harmless so far.