bhyve: Do not panic on invalid input in HDA emulation

[christos]

Summary

The emulated HDA controller passed values taken from guest registers and
from guest memory straight into assert(), so a guest could abort bhyve
with values the emulation did not expect. Reject them instead.

In case the guest asked to start something and it failed, clear the
corresponding run/enable bit.

PR: 256379, 256381, 256382, 256383, 256384, 256385, 256386, 256498
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks

[bnovkov]

usr.sbin/bhyve/pci_hda.c

812–814

Is it safe to continue executing after this fails?

813

966

There's a bunch of other typos like this.

[christos]

usr.sbin/bhyve/pci_hda.c

812–814

Is it safe to continue executing after this fails?

Yes. This code path is called before any work is being done, and verb failures are not necessarily fatal errors. In snd_hda(4), hdac_send_command() does something similar for instance.

966

I know. I kept the same mistake for uniformity than cluttering the commit with these fixes.

[jhb]

Thanks for fixing these.

usr.sbin/bhyve/pci_hda.c

646–649

Maybe fix this to be ((bdle_sz % HDA_DMA_ACCESS_LEN) != 0)? Or alternatively if it isn't too tortured use __is_aligned() from <sys/cdefs.h> (that might be more readable)

[bnovkov]

usr.sbin/bhyve/pci_hda.c

966

got it, thanks.