[MarkT]

Even with 0xf there is still some artefacting with mmal. Though it is much less often.

Yes, by channel zap I ment change channel up or down. I get excessive "Buffering...xx%" messages and then a freeze with MMAL. Not in OMXPlayer.

If you run a "journalctl -f" log tail on the RPi2, do you see continuity error lines such as these?

Code:

Nov 07 20:48:21 RPI2 tvheadend[673]: TS: Kabel Deutschland/114MHz/arte HD: H264 @ #6661 Continuity counter error (total 1)

Followed by video artifacts briefly after a new message pops up? How many per hour do you get? I used to use a Terratec H5 Rev.2 with said Empia USB chip with a Micronas DRX-K tuner. Its isosynchronous USB endpoints still give me nightmares. If possible, try a USB dongle with Cypress chip. Those use bulk endpoints and I think (just imho) those are better tolerated by an RPi2 under SDRAM bandwidth starvation. Using a Cypress stick cut those errors down by a factor 50 or more in my setup. From 3-5 video dropouts per minute with frequent pitch problems due to resampling catching up, to now maybe one every 20 minutes. I am also using dwc_otg.fiq_fsm_mask=0xF as kernel command line on the FAT32 partition. For testing, use a powered USB hub and make sure you do not ever (even during boot) get that colored square top right in the display (low voltage indication).

@Milhouse Props to you for being totally above the kernel patching game with regard to video hardware. I'll keep my eyes open.