Bug 24618

Summary:	Digital TV tuning failure for Silicon Labs devices in mga7beta3 and probably Cauldron
Product:	Mageia	Reporter:	Len Lawrence <tarazed25>
Component:	RPM Packages	Assignee:	Kernel and Drivers maintainers <kernel>
Status:	RESOLVED INVALID	QA Contact:
Severity:	normal
Priority:	Normal	CC:	marja11
Version:	7
Target Milestone:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Source RPM:	unknown	CVE:
Status comment:
Attachments:	Various diagnostics for mga6 for comparison with mga7 journal and lsmod output and firmware list

Description Len Lawrence 2019-04-04 18:24:17 CEST

Description of problem:
A regression has been found in the handling of DVB T/T2 tuning in mga7, specifically mga7beta3 with at least two Hauppauge TV tuners. The one currently plugged in is the USB WinTV SoloHD (Bus 001 Device 003: ID 2040:0264 Hauppauge) containing two Silicon Labs chips, the Si2157-A30 tuner and Si2168-B40 demodulator.

In mga6 this combination works fine, for kaffeine and vlc, but in mga7 the system identifies the chips but fails to download the firmware so no TV tuning is possible.

No idea what component to select - the Bugzilla menu does not quite cover this case. Also, I believe that the bug is not specific to mga7beta3.

Version-Release number of selected component (if applicable):

mga7beta3 tested.

How reproducible:

So far always, but note that mga6 is not affected.

Steps to Reproduce:
1. Connect, for instance, a Hauppauge DVB-T2 tuner to an antenna TV feed.
Devices tested are WinTV dualHD and soloHD.
2. Copy the necessary firmware to /lib/firmware. Source is
https://github.com/OpenELEC/dvb-firmware/blob/master/firmware/
3. Try to set up TV channels in kaffeine, or for vlc run
$ w_scan -ft -c GB -L > channels_new.xspf
(change country code to suit)
($ vlc channels_new.xspf
writes the channels to a playlist)
Observe that neither operation establishes any channels.
4. $ dmesg | grep Si21
# lsmod | grep em28xx
Also look for dvb and em28xx messages in the journal. The device should be
recognized at boot time - one of the last messages expected if it is usable
is "dvb set to isoc mode". When unusable expect to see something like
"Analogue or streaming TV is not supported"

Attached are specimen outputs for mga6 and mga7.
USB WinTV SoloHD (Bus 001 Device 003: ID 2040:0264 Hauppauge)
Which software packages are involved is beyond me but the kernel and/or the
em28xx driver may be involved.

Comment 1 Len Lawrence 2019-04-04 18:28:22 CEST

Created attachment 10904 [details]
Various diagnostics for mga6 for comparison with mga7

Comment 2 Len Lawrence 2019-04-04 18:30:04 CEST

Created attachment 10905 [details]
journal and lsmod output and firmware list

Comment 3 Marja Van Waes 2019-04-06 10:18:00 CEST

(In reply to Len Lawrence from comment #0)
> Description of problem:
> A regression has been found in the handling of DVB T/T2 tuning in mga7,
> specifically mga7beta3 with at least two Hauppauge TV tuners.  The one
> currently plugged in is the USB WinTV SoloHD (Bus 001 Device 003: ID
> 2040:0264 Hauppauge) containing two Silicon Labs chips, the Si2157-A30 tuner
> and Si2168-B40 demodulator.
> 
> In mga6 this combination works fine, for kaffeine and vlc, but in mga7 the
> system identifies the chips but fails to download the firmware so no TV
> tuning is possible.
> 
> No idea what component to select - the Bugzilla menu does not quite cover
> this case.  Also, I believe that the bug is not specific to mga7beta3.
> 
> 
<snip>
>     Which software packages are involved is beyond me but the kernel and/or
> the
>     em28xx driver may be involved.

That's my first guess too, I can't now read your attachments and don't know whether I'd understand them well enough, anyway.

Assigning to the kernel and drivers maintainers. Please reassign if I'm wrong.

CC: (none) => marja11
Assignee: bugsquad => kernel

Comment 4 Len Lawrence 2019-05-23 17:28:21 CEST

Solved, finally.  While checking TV operation in Mageia7 RC I noted that some of the firmware files differed from those I had been using in mga6.  Obtained versions from Git and tried again with w_scan2.  No problem at all.  So the problem has been PEBCAK all the time.  Sorry for the noise.

Status: NEW => RESOLVED
Resolution: (none) => INVALID