Page MenuHomeFreeBSD

pci: Don't try to read cfg registers of non-existing devices

Authored by kd on Nov 19 2021, 7:54 AM.
Referenced Files
Unknown Object (File)
Thu, May 25, 8:18 AM
Unknown Object (File)
Sat, May 6, 9:51 AM
Unknown Object (File)
Mar 5 2023, 7:55 AM
Unknown Object (File)
Mar 3 2023, 10:36 AM
Unknown Object (File)
Feb 16 2023, 11:58 AM
Unknown Object (File)
Feb 14 2023, 12:53 PM
Unknown Object (File)
Jan 12 2023, 1:27 PM
Unknown Object (File)
Nov 28 2022, 9:59 AM



Instead of returning 0xffs some controllers, such as Layerscape generate an external exception when someone attempts to read any register of config space of a non-existing device other than PCIR_VENDOR.
This causes a kernel panic.
Fix it by bailing during device enumeration if a device vendor register returns invalid value. (0xffff)
Use this opportunity to replace some hardcoded values with a macro.

I believe that this change won't have any unintended side-effects since it is safe to assume that if vendor == 0xffff -> hdr_type == 0xffff,
which means that we won't accidentally skip enumerating a slot.

Diff Detail

rG FreeBSD src repository
Lint Not Applicable
Tests Not Applicable

Event Timeline

kd requested review of this revision.Nov 19 2021, 7:54 AM
kd created this revision.

Huh, interesting that the newer code (rescan) DTRT.


While this is fine, we should arguably assert that this never happens as I think we should be catching this case in all the callers? Hmm, I guess we expose this for use by CardBus, so perhaps it is best left as you have changed it. I would be happy if we could check all the callers and assert here instead though, perhaps as a followup.


Just add a blank line before the new comment.

This revision is now accepted and ready to land.Nov 22 2021, 5:58 PM