Kodi Community Forum
v19 LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Printable Version

+- Kodi Community Forum (https://forum.kodi.tv)
+-- Forum: Support (https://forum.kodi.tv/forumdisplay.php?fid=33)
+--- Forum: General Support (https://forum.kodi.tv/forumdisplay.php?fid=111)
+---- Forum: Linux (https://forum.kodi.tv/forumdisplay.php?fid=52)
+---- Thread: v19 LibreELEC Testbuilds for x86_64 (Kodi 19.0) (/showthread.php?tid=343069)

Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-07

Weekly Linux 5.4-rc2 build #1007x: Generic

Packages disabled/not included:
text:


digital_devices

xf86-video-nvidia-legacy
Known issues:
text:


xf86-video-nvidia may not be working (untested)


RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-07

(2019-10-07, 23:04)Nafi Wrote: @Milhouse

I did some binary-search like investigation.

The issue starts with 0611 build. every build before 0611 is fine.
Quote:mesa: update to mesa-19.1.0

The bump to mesa-19.1.0 is possible, but #0611 also bumps to kernel 5.1.9 in which the Gemini Lake framebuffer compression is disabled, which may be a more likely cause for this issue - the framebuffer compression causes graphical corruption of the top part of the display. Are you disabling/enabling or otherwise modifying the framebuffer compression behaviour on your hardware via the kernel command line (the option is [inline]i915.enable_fbc[inline])? If so, remove any framebuffer kernel command line options and see if that helps - the default should now be fine (ie. disabled).

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-07

@Nafi my only other suggestion at this time would be to try a different HDMI cable, in case your cable is somehow marginal on bandwidth...

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - dapharsyde - 2019-10-08

(2019-10-07, 22:21)Milhouse Wrote: Can you post a "kodi log" from the LE settings addon (using build #1006) so I can see if there's anything useful in the system log? You could also try the various 5.3-rc builds - there are 8 rc builds - as that might help narrow it down (see the 5.3 PR for links). 
While trying to narrow this down, I tried older builds like #0620, #0801, #0901, #0913-#0916, and they all work fine.  The change appears to be between rc-8 and the final build.

Here is a link to the log generated from the #1007 build - http://ix.io/1XXl

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Nafi - 2019-10-08

@Milhouse 

it seems the issue was caused by my xorg.conf. In the early days, i added some modelines because some resolutions/refreshrates were not detected when HTPC was connected to my AVR.

without that custom xorg.conf, the issue seems to be gone. It seems to be useless anyway since all the needed resolutions are available without it.

thanks for guiding me into the right direction. BTW, I also tried that enable_fbc kernel parameter (with 0 and 1). It did not make any difference.

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-08

(2019-10-08, 00:58)Nafi Wrote: @Milhouse 

it seems the issue was caused by my xorg.conf. In the early days, i added some modelines because some resolutions/refreshrates were not detected when HTPC was connected to my AVR.

without that custom xorg.conf, the issue seems to be gone. It seems to be useless anyway since all the needed resolutions are available without it.

Great! Smile

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-08

(2019-10-08, 00:35)dapharsyde Wrote:
(2019-10-07, 22:21)Milhouse Wrote: Can you post a "kodi log" from the LE settings addon (using build #1006) so I can see if there's anything useful in the system log? You could also try the various 5.3-rc builds - there are 8 rc builds - as that might help narrow it down (see the 5.3 PR for links). 
While trying to narrow this down, I tried older builds like #0620, #0801, #0901, #0913-#0916, and they all work fine.  The change appears to be between rc-8 and the final build.

Here is a link to the log generated from the #1007 build - http://ix.io/1XXl

OK so just to be clear - #0909x (5.3-rc8) is good but #0917 (5.3.0) is bad?

This look a bit suspicious (from your #1007 log):
text:


Oct 07 15:14:15 KODI02 bluetoothd[402]: Can't get HIDP connection info

Oct 07 15:14:15 KODI02 kernel: Bluetooth: HIDP (Human Interface Emulation) ver 1.2

Oct 07 15:14:15 KODI02 kernel: Bluetooth: HIDP socket layer initialized

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: unknown main item tag 0x0

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: DualShock 4 calibration report's CRC check failed, received crc 0x1cffffa1 != 0x54313c2e

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: Retrying DualShock 4 get calibration report request

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: DualShock 4 calibration report's CRC check failed, received crc 0x1cffffa1 != 0x54313c2e

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: Retrying DualShock 4 get calibration report request

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: DualShock 4 calibration report's CRC check failed, received crc 0x1cffffa1 != 0x54313c2e

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: Failed to get calibration data from Dualshock 4

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: hidraw3: BLUETOOTH HID v81.00 Gamepad [Wireless Controller] on 00:15:83:0c:bf:eb

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: failed to claim input

Oct 07 15:14:15 KODI02 systemd-udevd[1408]: hidraw3: Failed to process device, ignoring: No such file or directory

Could you post the same log, but from #0916 (ie. working PS4 controller) - or from #0909x, assuming that works - for comparison purposes?

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - dapharsyde - 2019-10-08

(2019-10-08, 01:35)Milhouse Wrote:
(2019-10-08, 00:35)dapharsyde Wrote:
(2019-10-07, 22:21)Milhouse Wrote: Can you post a "kodi log" from the LE settings addon (using build #1006) so I can see if there's anything useful in the system log? You could also try the various 5.3-rc builds - there are 8 rc builds - as that might help narrow it down (see the 5.3 PR for links). 
While trying to narrow this down, I tried older builds like #0620, #0801, #0901, #0913-#0916, and they all work fine.  The change appears to be between rc-8 and the final build.

Here is a link to the log generated from the #1007 build - http://ix.io/1XXl 

OK so just to be clear - #0909x (5.3-rc8) is good but #0917 (5.3.0) is bad?

This look a bit suspicious (from your #1007 log):
text:


Oct 07 15:14:15 KODI02 bluetoothd[402]: Can't get HIDP connection info

Oct 07 15:14:15 KODI02 kernel: Bluetooth: HIDP (Human Interface Emulation) ver 1.2

Oct 07 15:14:15 KODI02 kernel: Bluetooth: HIDP socket layer initialized

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: unknown main item tag 0x0

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: DualShock 4 calibration report's CRC check failed, received crc 0x1cffffa1 != 0x54313c2e

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: Retrying DualShock 4 get calibration report request

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: DualShock 4 calibration report's CRC check failed, received crc 0x1cffffa1 != 0x54313c2e

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: Retrying DualShock 4 get calibration report request

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: DualShock 4 calibration report's CRC check failed, received crc 0x1cffffa1 != 0x54313c2e

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: Failed to get calibration data from Dualshock 4

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: hidraw3: BLUETOOTH HID v81.00 Gamepad [Wireless Controller] on 00:15:83:0c:bf:eb

Oct 07 15:14:15 KODI02 kernel: sony 0005:054C:05C4.0004: failed to claim input

Oct 07 15:14:15 KODI02 systemd-udevd[1408]: hidraw3: Failed to process device, ignoring: No such file or directory

Could you post the same log, but from #0916 (ie. working PS4 controller) - or from #0909x, assuming that works - for comparison purposes? 

I had a feeling you would ask for that Smile  Here is a log from #0916 - I had to remove a bunch of clutter from the emby plugin, but the section with pairing the controller is included - https://pastebin.com/69GNU13B

What I don't understand is why the debug messages look nothing alike...#0916 shows messages from the Joystick addon, whereas #1007 shows messages from the kernel.

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-08

(2019-10-08, 03:10)dapharsyde Wrote: I had a feeling you would ask for that Smile  Here is a log from #0916 - I had to remove a bunch of clutter from the emby plugin, but the section with pairing the controller is included - https://pastebin.com/69GNU13B

What I don't understand is why the debug messages look nothing alike...#0916 shows messages from the Joystick addon, whereas #1007 shows messages from the kernel.

Thanks, but can you upload the #0916 "kodi log" using the LE Settings add-on, as you did before with #1007? This provides a combined kodi.log & system log and in this case it's the system log that I'm most interested in. If it's too large to upload (due to emby spam) then try with debug logging disabled (the kodi log isn't so relevant here) or just upload the system log with journalctl -a | pastebinit.

Edit: also can you confirm:
(2019-10-08, 01:35)Milhouse Wrote: OK so just to be clear - #0909x (5.3-rc8) is good but #0917 (5.3.0) is bad?


RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - dapharsyde - 2019-10-08

(2019-10-08, 03:30)Milhouse Wrote:
(2019-10-08, 03:10)dapharsyde Wrote:  had a feeling you would ask for that Smile  Here is a log from #0916 - I had to remove a bunch of clutter from the emby plugin, but the section with pairing the controller is included - https://pastebin.com/69GNU13B

What I don't understand is why the debug messages look nothing alike...#0916 shows messages from the Joystick addon, whereas #1007 shows messages from the kernel.

Thanks, but can you upload the #0916 "kodi log" using the LE Settings add-on, as you did before with #1007? This provides a combined kodi.log & system log and in this case it's the system log that I'm most interested in. If it's too large to upload (due to emby spam) then try with debug logging disabled (the kodi log isn't so relevant here) or just upload the system log with journalctl -a | pastebinit.

Edit: also can you confirm:
(2019-10-08, 01:35)Milhouse Wrote: OK so just to be clear - #0909x (5.3-rc8) is good but #0917 (5.3.0) is bad?
Hi Milhouse - 

Thanks for the pointers.  I've disabled some plugins so the kodi log is shorter and able to be uploaded to ix.io.  I am now understanding that you prepared special "x" builds for these 5.3-rc's.  The tests I had done previously did not include these "x" builds.  Now that I understand, I have tested them and have discovered the break occurs between 0821x (rc-5) and 0827x (rc-6)

0722x - works - http://ix.io/1XYb
0814x - works - http://ix.io/1XYd
0821x - works - http://ix.io/1XYv
0827x - does not work - after pairing, does not register as controller - http://ix.io/1XYh

Edit 1: Bluetooth adapter worked in 0821x after re-plugging

Edit 2: I also tested regular #0827, and it working in this build - http://ix.io/1XZH

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - redfred - 2019-10-08

wow audio on linux 5.4 sounds so nice.

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-08

(2019-10-08, 04:22)dapharsyde Wrote: Hi Milhouse - 

Thanks for the pointers.  I've disabled some plugins so the kodi log is shorter and able to be uploaded to ix.io.  I am now understanding that you prepared special "x" builds for these 5.3-rc's.  The tests I had done previously did not include these "x" builds.  Now that I understand, I have tested them and have discovered the break occurs between 0821x (rc-5) and 0827x (rc-6)

0722x - works - http://ix.io/1XYb
0814x - works - http://ix.io/1XYd
0821x - works - http://ix.io/1XYv
0827x - does not work - after pairing, does not register as controller - http://ix.io/1XYh

Edit 1: Bluetooth adapter worked in 0821x after re-plugging

Edit 2: I also tested regular #0827, and it working in this build - http://ix.io/1XZH

Can you test #1007x which is based on 5.4-rc2 - it's a bit rough (early days) with a few issues due to namespace/depmod changes - but it does include quite a few HID and Bluetooth changes that are not currently in 5.3.y.

I've had a look at the commits bettween 5.3-rc5 and 5.3-rc6 and while there are several Bluetooth and HID changes, there's nothing Sony/PSx related, so I'd need to keep digging. Fingers crossed the issue is already fixed in 5.4-rc. If not, if you're willing I could try bisecting the kernel...

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - dapharsyde - 2019-10-08

(2019-10-08, 21:05)Milhouse Wrote:
(2019-10-08, 04:22)dapharsyde Wrote: Hi Milhouse - 

Thanks for the pointers.  I've disabled some plugins so the kodi log is shorter and able to be uploaded to ix.io.  I am now understanding that you prepared special "x" builds for these 5.3-rc's.  The tests I had done previously did not include these "x" builds.  Now that I understand, I have tested them and have discovered the break occurs between 0821x (rc-5) and 0827x (rc-6)

0722x - works - http://ix.io/1XYb
0814x - works - http://ix.io/1XYd
0821x - works - http://ix.io/1XYv
0827x - does not work - after pairing, does not register as controller - http://ix.io/1XYh

Edit 1: Bluetooth adapter worked in 0821x after re-plugging

Edit 2: I also tested regular #0827, and it working in this build - http://ix.io/1XZH

Can you test #1007x which is based on 5.4-rc2 - it's a bit rough (early days) with a few issues due to namespace/depmod changes - but it does include quite a few HID and Bluetooth changes that are not currently in 5.3.y.

I've had a look at the commits bettween 5.3-rc5 and 5.3-rc6 and while there are several Bluetooth and HID changes, there's nothing Sony/PSx related, so I'd need to keep digging. Fingers crossed the issue is already fixed in 5.4-rc. If not, if you're willing I could try bisecting the kernel... 

Hi Milhouse - 

Well, the bad news is that my testing box relies on the nvidia-legacy driver, so Xorg does not load in #1007x.  The good news, is that I tested this on another box I have which does not use nvidia-legacy, and the PS4 controller works as usual.

I guess I just need to wait for nvidia-legacy to be reenabled in these builds, and then I should be good to go.  Do you have any insight on why it was disabled, and if there are plans to bring back support?

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-08

(2019-10-08, 22:21)dapharsyde Wrote: Hi Milhouse - 

Well, the bad news is that my testing box relies on the nvidia-legacy driver, so Xorg does not load in #1007x.  The good news, is that I tested this on another box I have which does not use nvidia-legacy, and the PS4 controller works as usual.

I guess I just need to wait for nvidia-legacy to be reenabled in these builds, and then I should be good to go.

That at least sounds promising.

(2019-10-08, 22:21)dapharsyde Wrote: Do you have any insight on why it was disabled, and if there are plans to bring back support?

Yes, DRIVER_PRIME is no longer supported by the 5.4 kernel, so the Nvidia drivers need to be updated - the xf86-video-nvidia driver changes were fairly trivial, but the xf86-video-nvidia-legacy driver looks like it will require more effort.

Unfortunately Nvidia drivers (both xf86-video-nvidia-legacy and xf86-video-nvidia) are always a PITA with new major kernels updates - this is a general problem with out-of-tree kernel drivers so not entirely unique to Nvidia (Realtek with their WiFi drivers are much the same) however Nvidia seem to fall afoul of kernel changes more often than anyone else.

I'm not sure when we'll get a working xf86-video-nvidia-legacy driver for 5.4 - it may only be once 5.4.0 is released, but that's not guaranteed as the last xf86-video-nvidia-legacy driver appeared in June 2018 so it's been 16 months and 10 major kernels without an update (4.15, 4.16, 4.17, 4.18, 4.19, 4.20, 5.0, 5.1, 5.2, 5.3 and now soon 5.4...).

It's possible a patch may turn up during the rc cycle - I'll keep looking and testing - but if not then we may have to proceed without xf86-video-nvidia-legacy support (until such time as Nvidia get their arse in gear). All official Nvidia support is due to end during Kodi 19 development (the Nouveau driver will be supported, as this support VAAPI) so eventually this official Nvidia driver nonsense will come to end (no doubt replaced by a different type of nonsense...)

RE: LibreELEC Testbuilds for x86_64 (Kodi 19.0) - Milhouse - 2019-10-08

New LibreELEC.tv Matrix build #1008: Generic
(Supercedes previous build)

SHA256 Checksum: 474e8caf20816a7a13edb0ba9056f3a15d6f15dec9fdb70da8505102c5215533 (Generic)

text:
# uname -a

Linux NUC 5.3.5 #1 SMP Tue Oct 8 21:03:58 BST 2019 x86_64 GNU/Linux



# lsb_release

LibreELEC (Milhouse): devel-20191008210316-#1008-g96737bf [Build #1008]



# Kodi version

Kodi (19.0-ALPHA1 Git:548692d). Platform: Linux x86 64-bit

Based on tip of LibreELEC.tv master (96737bf, changelog) and tip of XBMC master (548692d, changelog) with the following modifications: Build Highlights:
  1. New 5.3.5 kernel
  2. AdvancedSettings: Fix arttype return
  3. WIP - [remote] Allow longpress mod for remote keymaps
Build Details:
  1. XBMC:
    • Show Studio icons (e.g. on folders, episodes or movies) (PR:16298, 2 commits, 1 file changed)
    • AdvancedSettings: Fix arttype return (PR:16736, 1 commit, 1 file changed)
  2. pvr.mythtv:
  3. Additional commits/pull requests/changes not yet merged upstream:
    • Updated: [env] PR:3807 (perma): rpi-eeprom: update to rpi-eeprom-48a2302
    • Updated: [env] PR:3857 (perma): linux (Generic/RPi/Allwinner): update to linux-5.3.5
    • Added: [pkg] PR:16740 (perma): WIP - [remote] Allow longpress mod for remote keymaps