Upgraded python to 3.13 w/ emerge @world and now kodi fails

Message

thegrind · Post by **thegrind** » Mon Jul 07, 2025 4:21 am

I upgraded my system to python 3.13 from 3.12 using emerge @world. Now when it tried to emerge kodi git (22), I get a failure after about 45 minutes.

Here are the log files

https://pastebin.com/gjH3hC74
https://pastebin.com/yupcDPSe
https://pastebin.com/XQ572D7N

Where am I going wrong? Is this my issue?

m_hasBlurayPlaylist = HasBlurayPlaylist

Post by Hu » Mon Jul 07, 2025 2:05 pm

You're using a live ebuild of Kodi from an overlay. Live ebuilds can break at any time due to upstream commits. This breakage may not be related to your Python update at all. If you rebuild the same commit you used the last time this worked, does it still fail now?

thegrind · Post by **thegrind** » Mon Jul 07, 2025 4:28 pm

Thanks for the help. The gentoo ebuild tries to pulli in Kodi 21 even though I have specified kodi-9999.

I ended up getting by the faile by installing media-libs/bluray but now when I try to start I get the following error.

/usr/bin/pulseaudio: /usr/lib64/libdbus-1.so.3: no version information available (required by /usr/bin/pulseaudio)
/usr/bin/pulseaudio: /usr/lib64/libdbus-1.so.3: no version information available (required by /usr/lib64/pulseaudio/libpulsecore-17.0.so)
/usr/bin/pulseaudio: symbol lookup error: /usr/lib64/pulseaudio/libpulsecore-17.0.so: undefined symbol: pa_dbus_wrap_connection_free
/usr/bin/kodi: line 216: 31797 Segmentation fault (core dumped) ${KODI_BINARY} ${ENV_ARGS} $SAVED_ARGS
Crash report available at /home/kodi/kodi_crashlog-20250707_105822.log

I am emerging now with pipewrire now to see if this works.

thegrind · Post by **thegrind** » Tue Jul 08, 2025 4:46 pm

I reposted in the multimedia forum as I am no longer using the kodi-overlay as I am using the gentoo repo but the post got locked.

Post by Hu » Tue Jul 08, 2025 5:14 pm

Cross-posting is prohibited, so I am not surprised that the thread was moderated. Upgrade Gentoo and now kodi 22 segfaults says you are using "kodi git", which to me means you are still using a live ebuild, albeit one now curated by Gentoo instead of from an overlay. Please show the output of emerge --info --verbose media-tv/kodi. Problems with projects from live ebuilds still get sent here, Unsupported Software, even when the live ebuild is curated by Gentoo, because the live code is not curated. In some cases, merging the thread would have been used rather than locking it, but in this case, little had happened in that thread, so I think linking to it here is sufficient. In that thread, you wrote:

[post=8866189]thegrind[/post] wrote:I am using the gentoo repo for kodi git. I had issues with getting it built but I resolved this with using -optical as a flag. The problem I have now is that when I start Kodi I get a segfault and a core dump.

I am not familiar with a compiler flag -optical, though I see that media-tv/kodi recognizes a USE flag optical. According to your pastebin, kodi is crashing in gallium. Please rebuild gallium with debug symbols and reproduce the crash. It may also be useful to show info locals and info args in the faulting thread.

thegrind · Post by **thegrind** » Wed Jul 09, 2025 1:09 am

Thanks for the help. I thought it went to unsupported since I wasn't using a gentoo repo which would make sense. Thanks for clarifying.

It shows one of the flags as optical for kodi. I get a bluray compile error is I don't add -optical. I am not sure why.

I setup debug and it shows in mesa where it throws the error. I figured the issue was with mesa so I tried all ebuilds thinking it either got fixed or added in one of the revs of mesa. The final one was mesa git.

https://pastebin.com/0usHx5KF

Code: Select all

(gdb) info locals
i = 0
hex_digits = "0123456789abcdef"
i = <optimized out>
(gdb) info args
len = 20
binary = 0x10 <error: Cannot access memory at address 0x10>
buf = 0x7fffffffd968 "\200\326AWUU"

thegrind · Post by **thegrind** » Wed Jul 09, 2025 3:29 am

I made a backup of my gentoo before I did the upgrade and it doesn't even have a libgallium file in the /usr/{lib64/lib}

Post by Hu » Wed Jul 09, 2025 3:18 pm

The pastebin is missing or deleted, so I cannot see the backtrace from it. I want to see the equivalent of the backtrace in your composite pastebin from the other thread, but built with debug symbols, which you did not have when that backtrace was created. Their absence is why gallium has lines like:

Code: Select all

#7 0x00007ffef9a31f14 in ?? () from /usr/lib64/libgallium-25.0.7.so

Instead of the slightly better data from gbm:

Code: Select all

#12 0x00007ffefedf1920 in gbm_create_device () from /usr/lib64/libgbm.so.1

or, even better, lines with a source filename and line number.

thegrind · Post by **thegrind** » Wed Jul 09, 2025 3:57 pm

https://pastecode.io/s/0hwi2rhm

I'm not sure why pastebin deleted my paste anyhow I have made another one.

thegrind · Post by **thegrind** » Wed Jul 09, 2025 9:46 pm

Since I compiled mesa with debugging this is all I get when I run kodi-gbm through gdb. I no longer see the libgallium threads.

Code: Select all

The program is not being run.
Temporary breakpoint 1 at 0x538798
Starting program: /usr/lib64/kodi/kodi-gbm
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib64/libthread_db.so.1".

Temporary breakpoint 1, 0x0000555555a88798 in main ()
Program not restarted.
Continuing.
[New Thread 0x7ffff029b020 (LWP 2507)]
[New Thread 0x7fffee5df020 (LWP 2508)]

Thread 1 "kodi-gbm" received signal SIGSEGV, Segmentation fault.
0x00007fffeeeb22d4 in mesa_bytes_to_hex (len=20, binary=0x10 <error: Cannot access memory at address 0x10>, buf=0x7fffffffd978 "\200\326AWUU") at ../mesa-99>
40          buf[i] = hex_digits[binary[i >> 1] >> 4];
Continuing.
Unable to fetch general registers: No such process.
[Thread 0x7fffee5df020 (LWP 2508) exited]
[Thread 0x7ffff029b020 (LWP 2507) exited]

Program terminated with signal SIGSEGV, Segmentation fault.
The program no longer exists.
The program is not being run.

thegrind · Post by **thegrind** » Wed Jul 09, 2025 9:49 pm

It still shows up in the crash file though.

Code: Select all

################ SYSTEM INFO ################
 Date: Tue Jul  8 18:36:06 CDT 2025
 Kodi Options:
 Arch: aarch64
 Kernel: Linux 6.12.12-v8-16k+ #1 SMP PREEMPT Sun Feb  9 17:53:04 CST 2025
 Release: Gentoo
############## END SYSTEM INFO ##############

############### STACK TRACE #################
=====>  Core file: /home/jim/core (2025-07-08 18:35:59.973791463 -0500)
        =========================================
[New LWP 2287]
[New LWP 2289]
[New LWP 2288]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib64/libthread_db.so.1".
Core was generated by `/usr/lib64/kodi/kodi-gbm'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x00007ffebb4c22d4 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
[Current thread is 1 (Thread 0x7ffec87adf20 (LWP 2287))]

Thread 3 (Thread 0x7ffec0b1b020 (LWP 2288)):
#0  0x00007ffec3e608ec in ?? () from /usr/lib64/libc.so.6
#1  0x00007ffec3e63404 in pthread_cond_wait () from /usr/lib64/libc.so.6
#2  0x0000555561241ddc in ANNOUNCEMENT::CAnnouncementManager::Process() ()
#3  0x0000555560946e7c in CThread::Action() ()
#4  0x0000555560948708 in ?? ()
#5  0x00007ffec415e6dc in ?? () from /usr/lib/gcc/aarch64-unknown-linux-gnu/14/libstdc++.so.6
#6  0x00007ffec3e6418c in ?? () from /usr/lib64/libc.so.6
#7  0x00007ffec3ec664c in ?? () from /usr/lib64/libc.so.6

Thread 2 (Thread 0x7ffeb2a2f020 (LWP 2289)):
#0  0x00007ffec3ec6964 in epoll_pwait () from /usr/lib64/libc.so.6
#1  0x0000555560534be8 in CLibInputHandler::Process() ()
#2  0x0000555560946e7c in CThread::Action() ()
#3  0x0000555560948708 in ?? ()
#4  0x00007ffec415e6dc in ?? () from /usr/lib/gcc/aarch64-unknown-linux-gnu/14/libstdc++.so.6
#5  0x00007ffec3e6418c in ?? () from /usr/lib64/libc.so.6
#6  0x00007ffec3ec664c in ?? () from /usr/lib64/libc.so.6

Thread 1 (Thread 0x7ffec87adf20 (LWP 2287)):
#0  0x00007ffebb4c22d4 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
#1  0x00007ffebb78ff48 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
#2  0x00007ffebb78dda8 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
#3  0x00007ffebb7813a4 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
#4  0x00007ffebb757710 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
#5  0x00007ffebb757800 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
#6  0x00007ffebb0e02f0 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
#7  0x00007ffebb710ea4 in ?? () from /usr/lib64/libgallium-25.2.0-devel.so
#8  0x00007ffebb0e3f20 in driCreateNewScreen3 () from /usr/lib64/libgallium-25.2.0-devel.so
#9  0x00007ffeb21f302c in ?? () from /usr/lib64/gbm/dri_gbm.so
#10 0x00007ffeb21f38b0 in ?? () from /usr/lib64/gbm/dri_gbm.so
#11 0x00007ffec4df175c in ?? () from /usr/lib64/libgbm.so.1
#12 0x00007ffec4df1920 in gbm_create_device () from /usr/lib64/libgbm.so.1
#13 0x000055556141d3e4 in KODI::WINDOWING::GBM::CGBMUtils::CreateDevice(int) ()
#14 0x000055556141b734 in KODI::WINDOWING::GBM::CWinSystemGbm::InitWindowSystem() ()
#15 0x000055556141e3cc in KODI::WINDOWING::GBM::CWinSystemGbmEGLContext::InitWindowSystemEGL(int, int) ()
#16 0x000055556141ea50 in KODI::WINDOWING::GBM::CWinSystemGbmGLContext::InitWindowSystem() ()
#17 0x0000555560b480a4 in CApplication::CreateGUI() ()
#18 0x00005555609803e8 in XBMC_Run ()
#19 0x0000555560428898 in main ()
############# END STACK TRACE ###############

thegrind · Post by **thegrind** » Thu Jul 10, 2025 4:36 pm

I had a rsync backup. I couldn't figure out the issue. Flags that I used before caused compile failures. I just ended up restoring the backup. Maybe I will try later just with the python upgrade.