Page MenuHomeFreeBSD

Fix pci-passthru MSI issues with OpenBSD guests.
ClosedPublic

Authored by grehan on May 22 2020, 8:56 AM.
Tags
None
Referenced Files
F163064587: D24951.id72107.diff
Sun, Jul 19, 6:35 PM
F163057396: D24951.id.diff
Sun, Jul 19, 4:56 PM
Unknown Object (File)
Wed, Jul 8, 8:00 AM
Unknown Object (File)
Mon, Jul 6, 5:08 PM
Unknown Object (File)
Mon, Jun 22, 4:59 PM
Unknown Object (File)
May 7 2026, 2:01 AM
Unknown Object (File)
Apr 29 2026, 1:43 AM
Unknown Object (File)
Apr 28 2026, 9:37 PM
Subscribers

Details

Summary
  • Return 2 x 16-bit registers in the correct byte order for a 4-byte read that spands the CMD/STATUS register. This reversal was hiding the capabilities-list, which prevented the MSI capability from being found for XHCI passthru.
  • Reorganize MSI/MSI-x config writes so that a 4-byte write at the capability offset would have the read-only portion skipped. This prevented MSI interrupts from being enabled.

PR 245392

Reported and tested by: Anatoli (me at anatoli dot ws)

Test Plan

Set up an OpenBSD guest with a passthru device.

Diff Detail

Repository
rS FreeBSD src repository - subversion
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

I think this is fine. The other design option would be to have pci_emul_capwrite to always require the explicit offset and id and move the while loop out into a helper function (pci_emul_lookup_cap or some such) that pci_cfgrw would use.

This revision is now accepted and ready to land.May 22 2020, 3:39 PM
This revision was automatically updated to reflect the committed changes.