Page MenuHomeFreeBSD

Make it obvious that SR-IOV VF creation fails because of the PCIe bus, not the device driver
ClosedPublic

Authored by val_packett.cool on Jun 10 2019, 7:12 PM.
Tags
Referenced Files
Unknown Object (File)
Wed, Nov 13, 11:16 AM
Unknown Object (File)
Wed, Nov 6, 8:19 PM
Unknown Object (File)
Fri, Nov 1, 2:20 AM
Unknown Object (File)
Thu, Oct 31, 6:44 PM
Unknown Object (File)
Oct 3 2024, 9:45 AM
Unknown Object (File)
Oct 3 2024, 8:26 AM
Unknown Object (File)
Oct 3 2024, 12:01 AM
Unknown Object (File)
Oct 2 2024, 12:37 PM
Subscribers

Details

Summary

"huh, iovctl on my new Intel NIC says 'no space left on device', let's google, okay, that means you're requesting too many VFs.. what, I'm requesting two. let's search ENOSPC in the driver.. here, add debug prints.. no, that's not it, argh.. let's walk up the calls, who calls this sriov init.. ah the iflib one.. who calls that... pci_iov_init OHH HERE IT IS!! so it's my AMD X370 chipset PCIe being weird, not the ixgbe driver being broken!"

No one should go through that debugging process again, let's make it loud and clear when ENOSPC is from the PCIe code and not from the driver code :)

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Not Applicable
Unit
Tests Not Applicable