[mythtv-users] Kernel 4.18 regression with some tuners

[Devin Heitmueller]

> I recently upgraded from kernel 4.17.1 to 4.18.6 and found out that I could no longer tune to any channels.  The channel scan in the mythtv-setup program would give me a non-sense error when trying to tune (it'd read "Cannot measure Signal Strength" and the error message was unknown error 254).  I was convinced that somehow all 3 of my PC HDTV 5500 PCI tuners had finally died, or something goofy was happening on my PCI controller with interrupts.  After playing around long enough, I decided to downgrade to an older but still supported kernel (4.17.19).  After that, I was able to tune to channels like before.  What tipped me off was this bug report filed from another consume of media devices in the kernel:
> https://tvheadend.org/issues/5183

The patch mentioned in this tvheadend issue is entirely an internal
API change (how the core interacts with the demodulators) - it should
have no effect on the userland API.  Further, even if the units were
off, it wouldn't result in a 254 error code.  I think the user was
just taking a stab in the dark based on looking at the changelog, not
based on actually bisecting the kernel.

> That seems to indicate the exact git revision where the regression occurred.  Has anyone else had issues with this kernel and know if and what details should be included in a post to LKML for this regression?

Any such issue should be reported to the linux-media mailing list, not
linux-kernel:

linux-media at vger.kernel.org

Devin

-- 
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com