HD MiniMyth on DG45FC

[azawacki]

I'm glad it is helpful.

Compile went for a LONG time before the next failure. Probably around 7 hours into it.

Now I think we're going to start getting to the good stuff. I've not upgraded to 21, I'm waiting for 22 as mentioned above... So I'm trying to compile myth 20 w/ fixes. The compile errors out here:

make[7]: Entering directory `/home/minimyth-x86_64/minimyth/gar-minimyth/script/myth-0.20/mythtv/work/main.d/mythtv-0.20-16082/libs/libmythtv'

...[snip]...

x86_64-minimyth-linux-gnu-g++ -c -I/home/minimyth-x86_64/minimyth/gar-minimyth/images/main/usr/include/mysql -L/home/minimyth-x86_64/minimyth/gar-minimyth/images/main/usr/lib/mysql -pipe -march=x86-64 -mtune=generic -O3 -mfpmath=sse -m64 -DUSE_ATI_PROPRIETARY_DRIVER_XVIDEO_HACK -Wall -W -O3 -pipe -march=x86-64 -mtune=generic -O3 -mfpmath=sse -m64 -DUSE_ATI_PROPRIETARY_DRIVER_XVIDEO_HACK -Wall -Wno-switch -fomit-frame-pointer -I/home/minimyth-x86_64/minimyth/gar-minimyth/images/main//usr/include/freetype2 -D_REENTRANT -DPIC -fPIC -DMMX -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr\" -DLIBDIR=\"/usr/lib\" -D_LARGEFILE_SOURCE -DUSING_OSS -DUSING_H264TOOLS -DUSING_XV -DUSING_XVMC -DUSING_XVMC_VLD -DUSING_OPENGL -DUSING_OPENGL_VSYNC -DUSING_FRONTEND -DUSING_V4L -DUSING_IVTV -DUSING_DVB -DUSING_BACKEND -DQT_NO_DEBUG -DQT_THREAD_SUPPORT -DQT_SHARED -I/home/minimyth-x86_64/minimyth/gar-minimyth/images/main/usr/lib/qt3/mkspecs/default -I. -I../../../../../../../../images/main/usr/include -I../../../../../../../../images/main/usr/include -I../../../../../../../../images/main/usr/include -I../.. -I.. -I../libmyth -I../libavcodec -I../libavutil -I../libmythmpeg2 -Idvbdev -Impeg -I../libmythlivemedia/BasicUsageEnvironment/include -I../libmythlivemedia/groupsock/include -I../libmythlivemedia/liveMedia/include -I../libmythlivemedia/UsageEnvironment/include -I../../../../../../../../images/main/usr/include -I../../../../../../../../images/main/usr/lib/qt3/include -I../../../../../../../../images/main/usr/include -I../../../../../../../../images/main/usr/include -o cardutil.o cardutil.cpp
In file included from dvbchannel.h:22,
from cardutil.cpp:13:
dvbtypes.h:29:2: error: #error "DVB driver includes with API version 3 not found!"
dvbtypes.h:42:2: warning: #warning DVB API version < 3.1
dvbtypes.h:43:2: warning: #warning ATSC will not be supported using the Linux DVB drivers
In file included from ../libmyth/mythwidgets.h:20,
from ../libmyth/settings.h:10,
from cardutil.h:5,
from cardutil.cpp:7:
../../../../../../../../images/main/usr/lib/qt3/include/qimage.h: In member function ‘bool QImageTextKeyLang::operator<(const QImageTextKeyLang&) const’:
../../../../../../../../images/main/usr/lib/qt3/include/qimage.h:61: warning: suggest parentheses around && within ||
make[7]: *** [cardutil.o] Error 1

[Pablo]

This appears to be the problem that was fixed by MythTV changeset 19245. Unfortunately, this fixed the problem in trunk only.

For MiniMyth, I backported the fix to MythTV 0.21 and MythTV 0.20. The fix is in MiniMyth changeset 4247.

[azawacki]

Using changeset 4247, everything compiles for pentium-mmx and x86_64 architectures. I'll try for via-c2 later. (I wish I had a single common architecture for all of my frontends, but I'd prefer to have optimzed code rather than simply running the pentium-mmx build everywhere.)

This is good improvement. Minimyth boots, is able to get through the X configuration and is at the main minimyth menu at 1920x1080P. I had to copy and rename G.A.N.T back to G.A.N.T. since 20 w/ fixes uses the old name. These are the current issues that I am debugging and troubleshooting:

1. The menus are fine, but the text on the setup screens, and on the recording lists is REALLY tiny. That might be as easy as increasing the font size, but I don't think I adjusted it that much in the past, I'm thinking that there is still something a little off in the X configuration.

2. There is no audio, either via HDMI nor via analog. I was really hoping that HDMI audio would work in this build, and I see sporatically others mentioning that they get it working with the version of alsa that we're using, just need to spend some time playing around with parameters. I've tried several audio devices, but so far, no luck.

3. Even without audio, playing back a recording is "jerky" in that it plays 1-2 seconds, and then pauses 1/2 second, and repeats that pattern. I want to get audio working before I worry about this, because the audio may have something to do with it.

4. I need to play around a bit more and try to replicate an error that caused X to crash. I tried to play a video, and xine came up with an error that it couldn't find the file (fairly sure it was mounted) but I clicked ok too fast. Attempting to play the same video a second time (so I could double check the path of the video and compare to what was mounted) caused X to crash, and I had to reboot the machine.

I'm a bit concerned that the intel video driver may still not be ready for primetime, dispite the fact that we have the latest released version.

If there are any tips on debugging audio on a minimyth system, please let me know, as most of the tools normally used are not there. The machine is configured dual boot to a Fedora where analog audio works... I'm going to try to see how alsa is configured there and try to replicate the configuration for minimyth. Reports that I've read online are that analog audio stopped working on the latest release of the kernel + alsa but that digital audio started working. Previous release of kernel + alsa had the reverse, so I don't know how much I am going to be able to get out of comparing configurations. At least there, aplay -L should give me a list of device numbers to experiment with.

[azawacki]

I guess I spoke too soon about HDMI audio, maybe just digital audio is working, as I see this: ( http://intellinuxgraphics.org/ )

2009-1-15:xf86-video-intel 2.6.0 released, with DRI2, XvMC for Mpeg2 on i965, HDMI audio support, better TV image quality, basic SDVO LVDS support, together with Intel 2008Q4 graphics package released.

Implying that I'll need to wait for an upgrade to 2.6 to get HDMI audio working. Let me see what I can do about audio using something else for now.

Thanks,
Anthony

[Pablo]

I guess I spoke too soon about HDMI audio, maybe just digital audio is working, as I see this: ( http://intellinuxgraphics.org/ )

2009-1-15:xf86-video-intel 2.6.0 released, with DRI2, XvMC for Mpeg2 on i965, HDMI audio support, better TV image quality, basic SDVO LVDS support, together with Intel 2008Q4 graphics package released.

Implying that I'll need to wait for an upgrade to 2.6 to get HDMI audio working. Let me see what I can do about audio using something else for now.

Thanks,
Anthony

As of SVN commit #4260, MiniMyth contains the xf86-video-intel 2.6.0, libdrm 2.4.4, and the the Linux 2.6.28 patches. It should not need the ALSA patch as MiniMyth uses the out-of-tree ALSA 1.0.18a drivers that appears to contain the patch already. It does not contain either the pre-release Mesa or pre-release Xorg server packages.

[azawacki]

Can we add an additional kernel patch?

http://intellinuxgraphics.org/download/alsa-hdmi.patch

This modifies the kernel to enable HDMI Audio for the Intel Eaglelake chipset.

I do think things are getting better. At the moment, video appears to be much less choppy than in the previous build.

[Pablo]

Can we add an additional kernel patch?

http://intellinuxgraphics.org/download/alsa-hdmi.patch

This modifies the kernel to enable HDMI Audio for the Intel Eaglelake chipset.

I do think things are getting better. At the moment, video appears to be much less choppy than in the previous build.

MiniMyth SVN 4292 updates MiniMyth's ALSA version to 1.0.19, which should include those changes.

[azawacki]

But that's actually a 2.6.28 kernel patch to enable HDMI audio on Intel Eaglelake chipsets... Not an ALSA patch, or am I misunderstanding?

[Pablo]

But that's actually a 2.6.28 kernel patch to enable HDMI audio on Intel Eaglelake chipsets... Not an ALSA patch, or am I misunderstanding?

The kernel contains an in-tree version of the ALSA drivers. The patch updates some of the in-tree ALSA files and no other files.

MiniMyth uses the separate, out-of-tree ALSA because it is generally updated before the in-tree ALSA. It appears the the ALSA 1.0.19 that is now included in MiniMyth already includes these changes.

[azawacki]

Still working on it. I think I'm getting close to making this version of minimyth usable for day-to-day activities. I still don't have HDMI audio working, and no audio in mythtv. But I am able to log in and run xine from the command line and specify analog audio and get sound playing back DVDs. I found it quite helpful to copy alsamixer and aplay over to the minimyth system after booting... Am able to see the devices, and aplay looks like it is doing its job when playing HDMI audio, no warnings of any sort, but no sound.

I'll spend a bit more time trying to get HDMI audio to work, but then shift to using this machine with analog audio. The system is much more stable than my FC9 due to latest drivers. As noted above, xine crashes X under FC9, but here, it's been working well.

Video playback at this point is pretty good, mythtv recordings without sound play nicely, and xine with weave deinterlacer looks pretty good.

While trying to debug, dmesg contains this:
[drm] Initialized drm 1.1.0 20060810
pci 0000:00:02.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
pci 0000:00:02.0: setting latency timer to 64
[drm] Initialized i915 1.6.0 20080730 on minor 0
unionfs: new lower inode mtime (bindex=0, name=ssh)
resource map sanity check conflict: 0xd0000000 0xdfffffff 0xd0000000 0xd1feffff vesafb
------------[ cut here ]------------
WARNING: at arch/x86/mm/ioremap.c:226 __ioremap_caller+0xc8/0x2cd()
Modules linked in: i915 drm coretemp joydev usbhid evdev snd_hda_codec_intelhdmi snd_hda_codec_idt ata_piix psmouse libata scsi_mod i2c_i801 i2c_core snd_hda_intel snd_hda_codec snd_hwdep snd_pcm snd_timer snd soundcore snd_page_alloc intel_agp ehci_hcd uhci_hcd thermal button processor ppdev parport_pc parport loop as_iosched rtc af_packet usbcore unionfs
Pid: 2816, comm: X Not tainted 2.6.28.1 #1
Call Trace:
[<ffffffff8022b2f9>] warn_on_slowpath+0x51/0x75
[<ffffffff80230049>] iomem_map_sanity_check+0x75/0xac
[<ffffffff8021f303>] __ioremap_caller+0xc8/0x2cd
[<ffffffffa01c47da>] i915_gem_entervt_ioctl+0x5a/0x4d4 [i915]
[<ffffffffa01c47da>] i915_gem_entervt_ioctl+0x5a/0x4d4 [i915]
[<ffffffffa01c4780>] i915_gem_entervt_ioctl+0x0/0x4d4 [i915]
[<ffffffffa01a8bcb>] drm_ioctl+0x1c9/0x253 [drm]
[<ffffffff8027e1c7>] vfs_ioctl+0x56/0x6c
[<ffffffff8027e562>] do_vfs_ioctl+0x385/0x3b2
[<ffffffff80273b7d>] vfs_write+0x121/0x136
[<ffffffff8027e5cc>] sys_ioctl+0x3d/0x5c
[<ffffffff8020b19b>] system_call_fastpath+0x16/0x1b
---[ end trace 0105393db7a800a8 ]---
HDMI hot plug event: Presence_Detect=0 ELD_Valid=1


It appears that this is ok, as it sounds like this happens because both xf86-video-intel and the intel alsa driver are accessing the same device...

Will keep hunting. Thanks for all of the help!

[azawacki]

Just an FYI kind of thing.

I still am working on getting HDMI audio to work, and figure out why the font size of the lists are so small...

I noticed that you updated the intel driver to 2.6.1, so I svn updated to svn revision to 4339, did a make clean; make garchive; make build... and found that it compiles a lot more packages that it did in the past.

With svn version 4294, the make build step (by grep'ing for NOW BUILDING) compiles 453 projects.

However, with version 4339, the make build step (by grep'ing for NOW BUILDING) compiles 1137 projects.

For example, previously, it only built mythtv version 0.20 for me. Now I can see it building both version 0.20, 0.21, and trunk. Perhaps I need to wipe out my directory and get the source from svn from scratch, or has something changed in the configuration that I didn't see to update?

I can live with it, but it takes a LONG time to compile now...

Thanks
Anthony

--------Edit--------
Actually, my build fails to boot with a really weird error:

modeprobe: FATAL: Could not load /lib/modules/2.6.27.13/modules.dep: No such file or director
Mount: none has wrong device number or fs type unionfs not supported
mkdir: cannot create directory 'initrd': Read-only file system
pivot_root: No such file or directory
/sbin/init: line 13: /usr/bin/test: not found
...
Kernel panic - not syncing: Attempted to kill init!

I say weird because I should have a 2.6.28 kernel... I think my build is faulty, deleting the entire tree, and rebuliding from scratch...

[Pablo]

Just an FYI kind of thing.

I still am working on getting HDMI audio to work, and figure out why the font size of the lists are so small...

I noticed that you updated the intel driver to 2.6.1, so I svn updated to svn revision to 4339, did a make clean; make garchive; make build... and found that it compiles a lot more packages that it did in the past.

With svn version 4294, the make build step (by grep'ing for NOW BUILDING) compiles 453 projects.

However, with version 4339, the make build step (by grep'ing for NOW BUILDING) compiles 1137 projects.

For example, previously, it only built mythtv version 0.20 for me. Now I can see it building both version 0.20, 0.21, and trunk. Perhaps I need to wipe out my directory and get the source from svn from scratch, or has something changed in the configuration that I didn't see to update?

I can live with it, but it takes a LONG time to compile now...

Given that it is building all versions of mythtv, sounds like you ran 'make build' in gar-minimyth/script rather than gar-minimyth/script/meta/minimyth. This will result in many more packages building as well as an image that would not likely work.

[seth_keith]

Do you guys have an update on the support for G45 based video? I am trying to make a DG45ID work, it has G45 w/ 4500HD acceleration. Is there some package I should try? Currently I am running 21.0.64, is 21.0.66b worth trying? If so, can I just replace the intel drivers? how?

[plato]

HDMI hot plug event: Presence_Detect=0 ELD_Valid=1

Hi, I get exactly this problem on a G45 too. Sound works fine until X starts, then I get the above event and sound cuts out. If I stop X, I get Presence_Detect=1 and sound carries on working. Anyone solved this?

[pmcenery]

HDMI hot plug event: Presence_Detect=0 ELD_Valid=1

I have recently bought a DG45FC and been trying to get HDMI audio working. I am also getting this message and can confirm that sound works until X is started. I also found the following upstream bug which has been reported. Hopefully it gets fixed soon:

https://bugs.freedesktop.org/show_bug.cgi?id=21371