pcihp: Fix race between config_intrhook and detaching
Needs ReviewPublic
Actions

Authored by imp on May 6 2021, 7:27 PM.

Details

Reviewers

Summary

Make sure the boot configuration interrupt hooks have all run before enabling
interrupts and processing hot plug events. Otherwise, the interrupt hook
processing at boot could race interrupt hooks that may still be running.

Diff Detail

Repository

rS FreeBSD src repository - subversion

Lint

Lint Passed

Unit

No Test Coverage

Build Status

Buildable 39058
Build 35947: arc lint + arc unit

Event Timeline

imp requested review of this revision.May 6 2021, 7:27 PM

imp created this revision.

Harbormaster completed remote builds in B39058: Diff 88749.May 6 2021, 7:27 PM

imp added a reviewer: jhb.May 6 2021, 7:28 PM

jhb added inline comments.May 11 2021, 9:18 PM

sys/dev/pci/pci_pci.c
1425–1427
1429	Doesn't this need to save a tag somewhere or some such? I guess this just leaks the memory? We should probably save the tag and deregister on detach. Also, what about hot plug post-boot that adds a nested HP bridge? We will never enable hotplug events in that case?

I'll look into these issues...

sys/dev/pci/pci_pci.c
1429	Doesn't this need to save a tag somewhere or some such? Which tag? I guess this just leaks the memory? We should probably save the tag and deregister on detach. Ah, the event handler... Also, what about hot plug post-boot that adds a nested HP bridge? We will never enable hotplug events in that case? Hmmm. That's a good question