Mailing List Archive

On 29/11/2019 11:35, Mark Kendall wrote:
> So after getting on for 600 commits and 11 months, I think the render
> branch is ready for merging.
>
> I've put a little more flesh on the bones of the wiki page (but it
> still needs feeding):-
>
> https://www.mythtv.org/wiki/2019-render
>
> and the full gory detail is obviously available on github:-
>
> https://github.com/MythTV/mythtv/tree/devel/2019-render
>
> The great irony is that, after all those commits, a lot of people
> won't notice any difference:)
>
> Regardless, the new code should be leaner, more portable, more
> extensible and more feature rich. It might even be faster.
>
> There are still a few tweaks needed and I have a couple of clean up
> commits to push but otherwise the database schema update required (for
> video display profile settings) is already committed and ready to be
> enabled (i.e. via a bump to the schema version).
>
> Unless anyone has any serious concerns/objections, I will look to
> merge it into master on Sunday. As Stuart has pointed out, we need to
> give this a few months of stress testing before the next release.
>
> Happy to answer any questions you may have.
>

I'd like to see this in ASAP as we need to release before the dev freeze
for Ubuntu 20.04 (the next LTS release)


Regards
Stuart
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 29/11/2019 11:35, Mark Kendall wrote:
> So after getting on for 600 commits and 11 months, I think the render
> branch is ready for merging.
>
> I've put a little more flesh on the bones of the wiki page (but it
> still needs feeding):-
>
> https://www.mythtv.org/wiki/2019-render
>
> and the full gory detail is obviously available on github:-
>
> https://github.com/MythTV/mythtv/tree/devel/2019-render
>
> The great irony is that, after all those commits, a lot of people
> won't notice any difference:)
>
> Regardless, the new code should be leaner, more portable, more
> extensible and more feature rich. It might even be faster.
>
> There are still a few tweaks needed and I have a couple of clean up
> commits to push but otherwise the database schema update required (for
> video display profile settings) is already committed and ready to be
> enabled (i.e. via a bump to the schema version).
>
> Unless anyone has any serious concerns/objections, I will look to
> merge it into master on Sunday. As Stuart has pointed out, we need to
> give this a few months of stress testing before the next release.
>
> Happy to answer any questions you may have.
>
> Thanks all and regards
> Mark

Mark,

Is there anything special for building and/or running on Raspberry Pi ?

For building is the normal ./configure (no extra parameters) enough ?

I am assuming build depends setup using current mythtv ansible.

I have Pi2, Pi3 and Pi4 (4GB).

Note Pi4 defaults to fkms is this ok ?


Mike


_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 11/29/2019 10:44 PM, Stuart Auchterlonie wrote:
> On 29/11/2019 11:35, Mark Kendall wrote:
>> So after getting on for 600 commits and 11 months, I think the render
>> branch is ready for merging.
>>
>> I've put a little more flesh on the bones of the wiki page (but it
>> still needs feeding):-
>>
>> https://www.mythtv.org/wiki/2019-render
>>
>> and the full gory detail is obviously available on github:-
>>
>> https://github.com/MythTV/mythtv/tree/devel/2019-render
>>
>> The great irony is that, after all those commits, a lot of people
>> won't notice any difference:)
>>
>> Regardless, the new code should be leaner, more portable, more
>> extensible and more feature rich. It might even be faster.
>>
>> There are still a few tweaks needed and I have a couple of clean up
>> commits to push but otherwise the database schema update required (for
>> video display profile settings) is already committed and ready to be
>> enabled (i.e. via a bump to the schema version).
>>
>> Unless anyone has any serious concerns/objections, I will look to
>> merge it into master on Sunday. As Stuart has pointed out, we need to
>> give this a few months of stress testing before the next release.
>>
>> Happy to answer any questions you may have.
>>
> I'd like to see this in ASAP as we need to release before the dev freeze
> for Ubuntu 20.04 (the next LTS release)
>
I have no problem with this too. I've been looking forward to it.
Hopefully better on lower end android boxes.

Mark (S)



_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Fri, 29 Nov 2019 at 12:49, Mike Bibbings <mike.bibbings@gmail.com> wrote:
> Is there anything special for building and/or running on Raspberry Pi ?
>
> For building is the normal ./configure (no extra parameters) enough ?
>
> I am assuming build depends setup using current mythtv ansible.
>
> I have Pi2, Pi3 and Pi4 (4GB).
>
> Note Pi4 defaults to fkms is this ok ?

Mike

Off the top of my head, the regular ansible playlists should be fine
but you'll want to make sure the libdrm-dev package is installed.

Otherwise check configure shows you have DRM, Video4Linux (with DRM),
MMAL, OpenGL, OpenGL ES and EGL.

I've used a Pi3 with stock raspbian, open source driver and full kms
for testing. Don't expect miracles on the Pi3 - I can't get much more
than about 35fps for HD material due to a limitation in the driver
somewhere (load is only just double digits). Pi4 should arrive on
Monday for testing - just picked up a black Friday deal:)

Thanks and regards,
Mark
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Fri, Nov 29, 2019 at 11:36 AM Mark Kendall <mark.kendall@gmail.com> wrote:

> Happy to answer any questions you may have.

Thanks for the work. Refactor and cleanup
is often not recognized as being as important
as it really is for the future of the code base.

If I remember the details correctly, I suspect you
might want to emphasize(*) in the notes that
the removal of the mythtv specific filters means
that the files in {libdir}/mythtv/filters/* are no
longer built/installed, and packagers may
want/need to revise their packaging (and/or
add cleanup), and self-builders may want to
nuke that directory.




(*) You do mention the filters removal, just
not the explicit implication for the directory.
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 29/11/2019 15:56, Mark Kendall wrote:
> On Fri, 29 Nov 2019 at 12:49, Mike Bibbings <mike.bibbings@gmail.com> wrote:
>> Is there anything special for building and/or running on Raspberry Pi ?
>>
>> For building is the normal ./configure (no extra parameters) enough ?
>>
>> I am assuming build depends setup using current mythtv ansible.
>>
>> I have Pi2, Pi3 and Pi4 (4GB).
>>
>> Note Pi4 defaults to fkms is this ok ?
> Mike
>
> Off the top of my head, the regular ansible playlists should be fine
> but you'll want to make sure the libdrm-dev package is installed.
>
> Otherwise check configure shows you have DRM, Video4Linux (with DRM),
> MMAL, OpenGL, OpenGL ES and EGL.
>
> I've used a Pi3 with stock raspbian, open source driver and full kms
> for testing. Don't expect miracles on the Pi3 - I can't get much more
> than about 35fps for HD material due to a limitation in the driver
> somewhere (load is only just double digits). Pi4 should arrive on
> Monday for testing - just picked up a black Friday deal:)
>
> Thanks and regards,
> Mark
> _______________________________________________

Mark,

The following maybe helpful on Raspberry Pi 4 with Raspbian Buster

Known Issues:

1. USB SATA converters for SSD/Hardisks, terrible performance or disk
corruption

  See https://forum.mythtv.org/viewtopic.php?f=46&t=3221&start=15#p15760

2. WIFI Keyboards limited range

  There is a known issue with USB3 interference on some WIFI keyboards
which limits the range to a few inches rather than many feet.

  A solution is to use USB extender cable and plug keyboard receiver
into that.

  Note this issue is not just Raspberry Pi4, other systems have been
known to have this interference problem.


Helper Scripts (also work on Pi2 or Pi3)

1. Move Raspbian Buster from microSD card to SSD/Harddisk

  See https://forum.mythtv.org/viewtopic.php?f=46&t=3232&start=15#p15884

2. Setup full mythtv backend

  See https://forum.mythtv.org/viewtopic.php?f=46&t=3232#p15773

  Note: assumes install location of mythbackend is /usr/

  If building locally default ./configure will use /usr/local/, unless
--prefix=/usr is specified.

  mythtv packaging deb-light uses --prefix=/usr

  If install location is /usr/local the systemd mythtv-backend.service
file needs adjustment.


Mike


_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 29/11/2019 11:35, Mark Kendall wrote:
> So after getting on for 600 commits and 11 months, I think the render
> branch is ready for merging.
>
> I've put a little more flesh on the bones of the wiki page (but it
> still needs feeding):-
>
> https://www.mythtv.org/wiki/2019-render
>
> and the full gory detail is obviously available on github:-
>
> https://github.com/MythTV/mythtv/tree/devel/2019-render
>
> The great irony is that, after all those commits, a lot of people
> won't notice any difference:)
>
> Regardless, the new code should be leaner, more portable, more
> extensible and more feature rich. It might even be faster.
>
> There are still a few tweaks needed and I have a couple of clean up
> commits to push but otherwise the database schema update required (for
> video display profile settings) is already committed and ready to be
> enabled (i.e. via a bump to the schema version).
>
> Unless anyone has any serious concerns/objections, I will look to
> merge it into master on Sunday. As Stuart has pointed out, we need to
> give this a few months of stress testing before the next release.
>
> Happy to answer any questions you may have.
>
> Thanks all and regards
> Mark

253d4df44 built and running for el7. mythtv-setup segfaulted on exit
after a no-op visit, and the frontend gave the same 'Widget does not
have a window' warning on startup but seems to be running ok.

Cheers,

John

{{{
2019-12-01 11:37:53.443914 I Binding to UDP [::]:6948
2019-12-01 11:37:53.474531 I Display: New main widget
2019-12-01 11:37:53.474546 W Display: Widget does not have a window!
2019-12-01 11:37:53.474571 I UI Screen Resolution: 1024 x 576
2019-12-01 11:37:53.699713 I OpenGL: OpenGL vendor : VMware, Inc.
2019-12-01 11:37:53.699727 I OpenGL: OpenGL renderer : llvmpipe
(LLVM 7.0, 128 bits)
2019-12-01 11:37:53.699734 I OpenGL: OpenGL version : 2.1 Mesa 18.3.4
2019-12-01 11:37:53.699746 I OpenGL: Qt platform : xcb
2019-12-01 11:37:53.699806 I OpenGL: EGL display : No
2019-12-01 11:37:53.699812 I OpenGL: EGL images : No
2019-12-01 11:37:53.699818 I OpenGL: Qt OpenGL format : OpenGL 2.1
2019-12-01 11:37:53.699823 I OpenGL: Qt OpenGL surface : RGBA: 8880
Depth: 0 Stencil: 0
2019-12-01 11:37:53.699829 I OpenGL: Max texture size : 8192
2019-12-01 11:37:53.699835 I OpenGL: Max texture units : 96
2019-12-01 11:37:53.699841 I OpenGL: Shaders : Yes
2019-12-01 11:37:53.699846 I OpenGL: NPOT textures : Yes
2019-12-01 11:37:53.699851 I OpenGL: Multitexturing : Yes
2019-12-01 11:37:53.699856 I OpenGL: Rectangular textures : Yes
2019-12-01 11:37:53.699861 I OpenGL: Buffer mapping : Yes
2019-12-01 11:37:53.699866 I OpenGL: Framebuffer objects : Yes
2019-12-01 11:37:53.699872 I OpenGL: Unpack Subimage : Yes
2019-12-01 11:37:53.722397 I OpenGL: Initialised MythRenderOpenGL
2019-12-01 11:37:53.722407 I OpenGL: Using full range output
2019-12-01 11:37:53.723710 I MythUIHelper: Pruning cache directory:
/home/john/.mythtv/cache/remotecache
2019-12-01 11:37:53.727621 I MythUIHelper: Kept 424 files, deleted 0
files, stat error on 0 files
2019-12-01 11:37:53.727764 I MythUIHelper: Pruning cache directory:
/home/john/.mythtv/cache/thumbnails
2019-12-01 11:37:53.731640 I MythUIHelper: Kept 392 files, deleted 0
files, stat error on 0 files
2019-12-01 11:37:53.944309 I MythUIWebBrowser: Loading css from -
file:///usr/share/mythtv/themes/default/htmls/mythbrowser.css
2019-12-01 11:37:53.946771 I MythUIWebBrowser: disabling plugins
2019-12-01 11:37:54.024025 I Current MythTV Schema Version
(DBSchemaVer): 1357
2019-12-01 11:38:16.816601 W MythPainter: 137 images not yet de-allocated.
2019-12-01 11:38:16.816717 I Clearing OpenGL painter cache.
2019-12-01 11:38:16.818946 I OpenGL: MythRenderOpenGL closing
2019-12-01 11:38:16.830133 I Waiting for threads to exit.
Handling Segmentation fault
Segmentation fault
[john@HP_Box ~]$

}}}
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 01/12/2019 11:58, John Pilkington wrote:

> 253d4df44 built and running for el7.  mythtv-setup segfaulted on exit
> after a no-op visit, and the frontend gave the same 'Widget does not
> have a window' warning on startup but seems to be running ok.
>
> Cheers,
>
> John

And mythpreviewgen trouble again. None are being autogenerated for SD
or redio recordings, the only ones created by this box, and running
mythpreviewgen from the keyboard segfaults after printing the Buzzoff
protocol version.
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 01/12/2019 15:41, John Pilkington wrote:
> On 01/12/2019 11:58, John Pilkington wrote:
>
>> 253d4df44 built and running for el7.  mythtv-setup segfaulted on exit
>> after a no-op visit, and the frontend gave the same 'Widget does not
>> have a window' warning on startup but seems to be running ok.
>>
>> Cheers,
>>
>> John
>
> And mythpreviewgen trouble again.  None are being autogenerated for SD
> or redio recordings, the only ones created by this box, and running
> mythpreviewgen from the keyboard segfaults after printing the Buzzoff
> protocol version.

I'm seeing the same two commits later, in F30 with nvidia.

The 'edit cursor' problem that I reported earlier, also on both boxes,
is still present and almost certainly from c794391ade, to which I added
a typo. I had been trying to understand the workings of
mythtv/libs/libmythtv/deletemap.cpp and its relationship to
~/.mythtv/themes/MythCenter-wide/osd.xml,
while the commit changed osd.cpp

_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Sun, 1 Dec 2019 at 15:42, John Pilkington <johnpilk222@gmail.com> wrote:
>
> On 01/12/2019 11:58, John Pilkington wrote:
>
> > 253d4df44 built and running for el7. mythtv-setup segfaulted on exit
> > after a no-op visit, and the frontend gave the same 'Widget does not
> > have a window' warning on startup but seems to be running ok.
> >
> > Cheers,
> >
> > John
>
> And mythpreviewgen trouble again. None are being autogenerated for SD
> or redio recordings, the only ones created by this box, and running
> mythpreviewgen from the keyboard segfaults after printing the Buzzoff
> protocol version.

Thanks John,

Should all be fixed now.

Regards
Mark
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Sun, 1 Dec 2019 at 17:42, John Pilkington <johnpilk222@gmail.com> wrote:
> The 'edit cursor' problem that I reported earlier, also on both boxes,
> is still present and almost certainly from c794391ade, to which I added
> a typo. I had been trying to understand the workings of
> mythtv/libs/libmythtv/deletemap.cpp and its relationship to
> ~/.mythtv/themes/MythCenter-wide/osd.xml,
> while the commit changed osd.cpp

John

Are you saying there is a problem with that commit or a problem with
your local changes?

Regards
Mark
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 02/12/2019 08:31, Mark Kendall wrote:
> On Sun, 1 Dec 2019 at 17:42, John Pilkington <johnpilk222@gmail.com> wrote:
>> The 'edit cursor' problem that I reported earlier, also on both boxes,
>> is still present and almost certainly from c794391ade, to which I added
>> a typo. I had been trying to understand the workings of
>> mythtv/libs/libmythtv/deletemap.cpp and its relationship to
>> ~/.mythtv/themes/MythCenter-wide/osd.xml,
>> while the commit changed osd.cpp
>
> John
>
> Are you saying there is a problem with that commit or a problem with
> your local changes?
>
> Regards
> Mark

Hi Mark: ..that the change in usability was caused by that commit or
related ones immediately before it.

In the early days of qt5 the 'current position' marker in edit mode
disappeared. A workaround was created but not committed, and upstream
fixed it. Now the marker is visible but flicks between start and end of
the recording; it's bi-stable. The workaround might be helpful in
identifying relevant bits of code - but it doesn't restore proper
operation now.

http://www.gossamer-threads.com/lists/mythtv/users/588659#588659

> For the cutlist editor, this would be applied in the theme's osd.xml in
> the editbar section, specifically the position, cuttoright, cuttoleft,
> keeptoright, and keeptoleft elements.

Thanks,

John
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 02/12/2019 08:30, Mark Kendall wrote:

> Should all be fixed now.
>
> Regards
> Mark

OK, thanks. I have 4fbabd9 running on the F30+nVidia+TV box with no
segfaults or new functionality problems seen so far.

5.3.13-200.fc30.x86_64, nVidia 440.36, GeForce GT 710/PCIe/SSE2

nvdec came up in the second launch of the frontend, and the widget still
says it does not have a window, but... Now for el7.

Great,

John

_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 02/12/2019 13:39, John Pilkington wrote:
> On 02/12/2019 08:30, Mark Kendall wrote:
>
>> Should all be fixed now.
>>
>> Regards
>> Mark
>
> OK, thanks.  I have 4fbabd9 running on the F30+nVidia+TV box with no
> segfaults or new functionality problems seen so far.
>
> 5.3.13-200.fc30.x86_64, nVidia 440.36, GeForce GT 710/PCIe/SSE2
>
> nvdec came up in the second launch of the frontend, and the widget still
> says it does not have a window, but...   Now for el7.

Looking good on el7 too, at 4fbabd9. No vaapi/vdpau/nvdec active, just a
similar old Intel(R) Core(TM)2 Duo CPU E7400 @ 2.80GHz, Mesa DRI Intel
Q33 renderer under 4.4.205-1.el7.elrepo.x86_64

he edit-bar problem is still there and I haven't run for long enough to
test the mild lipsync drift, but otherwise, as you remarked ruefully,
there's little obvious change :-)

One more thing:

In a mildly tweaked version of dvbradioexport.pl (from the wiki)

Database schema 1358 not supported.
Bindings support schema version 1357

>
> Great,
>
> John
>

_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

Hi Mark

I am having a few problems compiling master for android.

In ./libs/libmythui/platforms/mythdisplayandroid.cpp

latforms/mythdisplayandroid.cpp: In member function 'virtual DisplayInfo
MythDisplayAndroid::GetDisplayInfo(int)':
platforms/mythdisplayandroid.cpp:37:9: error: 'VB_GENERAL' was not
declared in this scope
     LOG(VB_GENERAL, LOG_INFO, LOC +
         ^
platforms/mythdisplayandroid.cpp:37:21: error: 'LOG_INFO' was not
declared in this scope
     LOG(VB_GENERAL, LOG_INFO, LOC +
                     ^
platforms/mythdisplayandroid.cpp:37:31: error: 'LOC' was not declared in
this scope
     LOG(VB_GENERAL, LOG_INFO, LOC +
                               ^
platforms/mythdisplayandroid.cpp:41:9: error: 'LOG' was not declared in
this scope
         );
         ^

Also, I had to remove some configure options -  no longer valid - in
mythbuild.sh
--disable-crystalhd
--enable-opengl-video
--disable-openmax

Am I missing something? Should I add an include for the missing header
file to mythdisplayandroid.cpp ? I will update the mythbuild.sh. Are
there any other settings I should have for android?

Peter

_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Mon, 2 Dec 2019 at 20:50, Peter Bennett <pb.mythtv@gmail.com> wrote:
>
> Hi Mark
>
> I am having a few problems compiling master for android.
>
> In ./libs/libmythui/platforms/mythdisplayandroid.cpp
>
> latforms/mythdisplayandroid.cpp: In member function 'virtual DisplayInfo
> MythDisplayAndroid::GetDisplayInfo(int)':
> platforms/mythdisplayandroid.cpp:37:9: error: 'VB_GENERAL' was not
> declared in this scope
> LOG(VB_GENERAL, LOG_INFO, LOC +
> ^
> platforms/mythdisplayandroid.cpp:37:21: error: 'LOG_INFO' was not
> declared in this scope
> LOG(VB_GENERAL, LOG_INFO, LOC +
> ^
> platforms/mythdisplayandroid.cpp:37:31: error: 'LOC' was not declared in
> this scope
> LOG(VB_GENERAL, LOG_INFO, LOC +
> ^
> platforms/mythdisplayandroid.cpp:41:9: error: 'LOG' was not declared in
> this scope
> );
> ^

Hi Peter

I'm just about to push an update for MythDisplay so I'll add the
missing include here as well - apologies - haven't had a chance to try
the android build for a few weeks.

> Also, I had to remove some configure options - no longer valid - in
> mythbuild.sh
> --disable-crystalhd
> --enable-opengl-video
> --disable-openmax

Ah - I forgot about those - already removed in the render branch of
the packaging repo:)

> Am I missing something? Should I add an include for the missing header
> file to mythdisplayandroid.cpp ? I will update the mythbuild.sh.

Thanks

>Are there any other settings I should have for android?

Nothing springs to mind - but don't hold me to that:)

Thanks and regards
Mark
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

Hi,

I have tried to run some basic tests for devel/2019-render on two
systems ...
On both systems I'm using DVB-S2 cards with Astra 19.2E channels (SD
mpeg2 544x576, 720x576, HD H264 1280x720 German FTA, Astra UHD demo
channel HEVC Main10 3840x2160).

First I had to change libs/libmythbase/mythversion.h: #define
MYTH_DATABASE_VERSION "1357" to "1358", otherwise BE/FE would not start
because schema version would be newer than expected after the schema
updates.

On the first system (my office PC, Intel Skylake, HD 530, additional
Nvidia GT1030 card with HDMI 2.0 output, OpenSuSE Tumbleweed, 1920x1080
monitor):
    When using the Intel graphics using playback profile "VAAPI normal"
SD and HD LiveTV works fine but UHD stutters (this is to be expected
because only HEVC Main profile is supported by HD 530).
    When using the NVidia graphics card (driver version 430.26 first,
now 440.31) with playback profile "NVDEC Standard":
        SD LiveTV and UHD Live LiveTV works fine but channel change to
HD channels and sometimes to UHD channel results in
        SEGV in cuda_buffer_free (backtrace something like ...
mythfrontend.TVMenuCallback -> mythfrontend.startTVNormal -> libmythtv.* ->
        libmythavutil.buffer_replace, buffer_pool_free,
cuda_buffer_free, will provide more details as soon as possible)

On the second system (living room, Intel Coffee Lake, UHD 630, HDMI 1.4
output - limits output to 30Hz@3840x2160 - connected to UHD TV, OpenSuSE
Leap 42.3 with updates for vaapi, minimal X server without display
manager, playback profile "VAAPI normal"):
    LiveTV on SD, HD and UHD channels (oncluding channel switch) works
fine, but I'm not able to >30Hz on this system.
    In the end I would like to install the NVidia GT1030 card on this
system to get full UHD support.

Until now I was not able to run mythfronend without X (eglfs) on both
systems using the Intel graphics (did not even try NVidia), but
currently the SEGV for channel switching with the NVidia GT1030 is the
bigger problem for me ...

Br,
Andreas

_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Mon, 2 Dec 2019 at 21:29, Andreas Mayer <and.mayer@aon.at> wrote:
>
> Hi,
>
> I have tried to run some basic tests for devel/2019-render on two
> systems ...
> On both systems I'm using DVB-S2 cards with Astra 19.2E channels (SD
> mpeg2 544x576, 720x576, HD H264 1280x720 German FTA, Astra UHD demo
> channel HEVC Main10 3840x2160).
>
> First I had to change libs/libmythbase/mythversion.h: #define
> MYTH_DATABASE_VERSION "1357" to "1358", otherwise BE/FE would not start
> because schema version would be newer than expected after the schema
> updates.

The branch was merged into master yesterday - so master is the place
to test now. I'm assuming you have a timing issue on which versions of
master/render branch you used - and hence the schema version mismatch.

> On the first system (my office PC, Intel Skylake, HD 530, additional
> Nvidia GT1030 card with HDMI 2.0 output, OpenSuSE Tumbleweed, 1920x1080
> monitor):
> When using the Intel graphics using playback profile "VAAPI normal"
> SD and HD LiveTV works fine but UHD stutters (this is to be expected
> because only HEVC Main profile is supported by HD 530).

Yes - that will be using software decoding.

> When using the NVidia graphics card (driver version 430.26 first,
> now 440.31) with playback profile "NVDEC Standard":
> SD LiveTV and UHD Live LiveTV works fine but channel change to
> HD channels and sometimes to UHD channel results in
> SEGV in cuda_buffer_free (backtrace something like ...
> mythfrontend.TVMenuCallback -> mythfrontend.startTVNormal -> libmythtv.* ->
> libmythavutil.buffer_replace, buffer_pool_free,
> cuda_buffer_free, will provide more details as soon as possible)

That's not good. A full backtrace would be useful. I'll try and
reproduce - but I don't have any tuners that are HD/UHD capable at the
moment - so will have to fabricate a dummy tuner.

> On the second system (living room, Intel Coffee Lake, UHD 630, HDMI 1.4
> output - limits output to 30Hz@3840x2160 - connected to UHD TV, OpenSuSE
> Leap 42.3 with updates for vaapi, minimal X server without display
> manager, playback profile "VAAPI normal"):
> LiveTV on SD, HD and UHD channels (oncluding channel switch) works
> fine, but I'm not able to >30Hz on this system.
> In the end I would like to install the NVidia GT1030 card on this
> system to get full UHD support.

I've just been caught out by the CoffeeLake 30Hz UHD limit - very annoying:)

> Until now I was not able to run mythfronend without X (eglfs) on both
> systems using the Intel graphics (did not even try NVidia)

I can't get eglfs to work with Nvidia at all and I'm not entirely sure
whether CUDA/NVDEC will work with EGL without some modification.

For Intel, "QT_QPA_PLATFORM=eglfs mythfrontend" works for me when
there is no X. You may want to make sure you have the latest Qt and/or
mesa packages.

If you still have issues, Qt may give you a hint about the problem
with extra logging enabled:

export QT_LOGGING_RULES=qt.qpa.*=true

You may also need to change the permissions for your input devices to
ensure keyboards/mice work.

Thanks for the feedback,
Mark
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Fri, 29 Nov 2019 at 15:56, Mark Kendall <mark.kendall@gmail.com> wrote:
> I've used a Pi3 with stock raspbian, open source driver and full kms
> for testing. Don't expect miracles on the Pi3 - I can't get much more
> than about 35fps for HD material due to a limitation in the driver
> somewhere (load is only just double digits). Pi4 should arrive on
> Monday for testing - just picked up a black Friday deal:)

Just a heads up for anyone testing on the PI - I pushed a fix last
night that greatly improves Pi3 performance (may need to be applied
for Pi4 as well).

Playback is now respectable - haven't concluded whether it is as good
as it can be.

Regards
Mark
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Mon, 2 Dec 2019 at 21:29, Andreas Mayer <and.mayer@aon.at> wrote:
> When using the NVidia graphics card (driver version 430.26 first,
> now 440.31) with playback profile "NVDEC Standard":
> SD LiveTV and UHD Live LiveTV works fine but channel change to
> HD channels and sometimes to UHD channel results in
> SEGV in cuda_buffer_free (backtrace something like ...
> mythfrontend.TVMenuCallback -> mythfrontend.startTVNormal -> libmythtv.* ->
> libmythavutil.buffer_replace, buffer_pool_free,
> cuda_buffer_free, will provide more details as soon as possible)

Andreas

Don't worry about backtrace/more details - I've found the issue. Not
just a CUDA/NVDEC problem - bit of on an issue with live tv channel
switching/transitions. Now to try and fix it.

Regards
Mark
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On Tue, 3 Dec 2019 at 11:49, Mark Kendall <mark.kendall@gmail.com> wrote:
>
> On Mon, 2 Dec 2019 at 21:29, Andreas Mayer <and.mayer@aon.at> wrote:
> > When using the NVidia graphics card (driver version 430.26 first,
> > now 440.31) with playback profile "NVDEC Standard":
> > SD LiveTV and UHD Live LiveTV works fine but channel change to
> > HD channels and sometimes to UHD channel results in
> > SEGV in cuda_buffer_free (backtrace something like ...
> > mythfrontend.TVMenuCallback -> mythfrontend.startTVNormal -> libmythtv.* ->
> > libmythavutil.buffer_replace, buffer_pool_free,
> > cuda_buffer_free, will provide more details as soon as possible)
>
> Andreas
>
> Don't worry about backtrace/more details - I've found the issue. Not
> just a CUDA/NVDEC problem - bit of on an issue with live tv channel
> switching/transitions. Now to try and fix it.

OK - should be fixed now.
Regards
Mark
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

On 12/2/19 4:04 PM, Mark Kendall wrote:
> I'm just about to push an update for MythDisplay so I'll add the
> missing include here as well - apologies - haven't had a chance to try
> the android build for a few weeks.

Hi Mark
There are new errors today, in the same module

platforms/mythdisplayandroid.cpp: In member function 'virtual
DisplayInfo MythDisplayAndroid::GetDisplayInfo(int)':
platforms/mythdisplayandroid.cpp:38:31: error: 'LOC' was not declared in
this scope
     LOG(VB_GENERAL, LOG_INFO, LOC +
                               ^
/home/peter/proj/github.com/MythTV/packaging/android/mythinstall/qt/include/QtCore/qglobal.h:710:38:
note: in definition of macro 'qPrintable'
 #  define qPrintable(string) QString(string).toLocal8Bit().constData()
                                      ^
platforms/mythdisplayandroid.cpp:38:5: note: in expansion of macro 'LOG'
     LOG(VB_GENERAL, LOG_INFO, LOC +
     ^
Regards
Peter
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org

>> When using the NVidia graphics card (driver version 430.26 first,
>> now 440.31) with playback profile "NVDEC Standard":
>> SD LiveTV and UHD Live LiveTV works fine but channel change to
>> HD channels and sometimes to UHD channel results in
>> SEGV in cuda_buffer_free (backtrace something like ...
>> mythfrontend.TVMenuCallback -> mythfrontend.startTVNormal -> libmythtv.* ->
>> libmythavutil.buffer_replace, buffer_pool_free,
>> cuda_buffer_free, will provide more details as soon as possible)

>Don't worry about backtrace/more details - I've found the issue. Not
>just a CUDA/NVDEC problem - bit of on an issue with live tv channel
>switching/transitions. Now to try and fix it.

Switched to master today, no SEGV anymore (and TV db schema version OK), thank you Mark!

>> Until now I was not able to run mythfronend without X (eglfs) on both
>> systems using the Intel graphics (did not even try NVidia)
> If you still have issues, Qt may give you a hint about the problem
> with extra logging enabled:
> export QT_LOGGING_RULES=qt.qpa.*=true
The problem on the Intel system with the extra Nvidia card was that
eglfs.kms selected the Nvidia card by default (/dev/dri/card0):
qt.qpa.eglfs.kms: Found the following video devices: ("/dev/dri/card0", "/dev/dri/card1")
qt.qpa.eglfs.kms: Using "/dev/dri/card0"
drmModeGetResources failed (Operation not supported)
no screens available, assuming 24-bit color

So I had to force the use of the Intel card by using QT_QPA_EGLFS_KMS_CONFIG
and a configuration file specifing /dev/dri/card1.

The next problem was that the (default) Qt style gtk2 was trying to use a X display:
(mythfrontend:2481): Gtk-WARNING **: 21:20:35.470: cannot open display:

After setting QT_STYLE_OVERRIDE=Fusion or Windows mythfrontend was able to start,
first LiveTV tests were successful.

On the other system (with the old OS) I'm getting:
qt.qpa.egldeviceintegration: EGL device integration plugin keys: ("eglfs_kms_egldevice", "eglfs_x11")
qt.qpa.egldeviceintegration: EGL device integration plugin keys (sorted): ("eglfs_kms_egldevice", "eglfs_x11")
qt.qpa.egldeviceintegration: Trying to load device EGL integration "eglfs_kms_egldevice"
qt.qpa.eglfs.kms: New DRM/KMS on EGLDevice integration created
qt.qpa.egldeviceintegration: Using EGL device integration "eglfs_kms_egldevice"
EGL_EXT_device_base missing
So I guess I need to update the OS here first ...

My next steps will be:
Find out why eglfs does not work for the Nvidia card.
Finally find a configuration capable to be used with my UHD TV with 3840x2160@50/60Hz and HDR(BT2020).
Testing with RPi4B (has to arrive first).

Thank you!

Regards,
Andreas

_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://lists.mythtv.org/mailman/listinfo/mythtv-dev
http://wiki.mythtv.org/Mailing_List_etiquette
MythTV Forums: https://forum.mythtv.org