[SOLVED] MTP Upload Session in dmesg

eDio · 2020-01-10 19:35:40

From time to time on my machine happen GPU hangs (everything freezes on your screen, sometimes reset is needed, sometime machine recovers). I believe those are caused by qutebrowser (not sure how this is relevant).

Today, after such GPU hang has happened I checked dmesg output and found following lines

[14927.396998] i915 0000:00:02.0: GPU HANG: ecode 9:1:0x00000000, hang on rcs0
[14927.397002] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[14927.397004] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[14927.397005] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[14927.397007] The GPU crash dump is required to analyze GPU hangs, so please always attach it.
[14927.397009] GPU crash dump saved to /sys/class/drm/card0/error
[14927.398021] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[14927.398806] [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}
[14927.399040] i915 0000:00:02.0: Resetting chip for hang on rcs0
[14927.400834] [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}
[14927.401610] [drm:gen8_reset_engines [i915]] *ERROR* rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}
[14935.504514] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[14943.397822] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[14951.503507] i915 0000:00:02.0: Resetting rcs0 for stuck wait on rcs0
[14959.401304] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[14959.804200] audit: type=1701 audit(1578683919.011:177): auid=1000 uid=1000 gid=985 ses=2 pid=13477 comm=4A6176614658204170706C69636174 exe="/home/edio/.local/share/JetBrains/Toolbox/apps/IDEA-U/ch-0/193.5233.102/jbr/bin/java" sig=6 res=1
[14959.842996] audit: type=1701 audit(1578683919.047:178): auid=1000 uid=1000 gid=985 ses=2 pid=16950 comm="spotify" exe="/opt/spotify/spotify" sig=6 res=1
[14959.860157] audit: type=1130 audit(1578683919.067:179): pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=systemd-coredump@0-44724-0 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
[14959.885222] audit: type=1130 audit(1578683919.091:180): pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=systemd-coredump@1-44726-0 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
[14959.914224] MTP Main Sessio[11216]: segfault at 30 ip 00007f5d81d4e5f6 sp 00007f5d4232e3a0 error 4 in libQt5Core.so.5.14.0[7f5d81b3a000+2f8000]
[14959.914240] Code: 53 48 83 ec 28 8b 05 e1 36 29 00 83 f8 ff 7c 16 0f b6 05 7d 36 29 00 84 c0 0f 84 cd 00 00 00 4c 8d 35 8e 36 29 00 4d 8d 7e 30 <41> 8b 2f 41 89 ec 41 89 ed 41 81 e4 ff ff ff 00 41 81 e5 c0 ff ff
[14959.914269] audit: type=1701 audit(1578683919.121:181): auid=1000 uid=1000 gid=985 ses=2 pid=11199 comm=4D5450204D61696E2053657373696F exe="/usr/bin/telegram-desktop" sig=11 res=1
[14959.914831] MTP Upload Sess[11267]: segfault at 30 ip 00007f5d81d4e5f6 sp 00007f5d4132c3a0 error 4 in libQt5Core.so.5.14.0[7f5d81b3a000+2f8000]
[14959.914836] Code: 53 48 83 ec 28 8b 05 e1 36 29 00 83 f8 ff 7c 16 0f b6 05 7d 36 29 00 84 c0 0f 84 cd 00 00 00 4c 8d 35 8e 36 29 00 4d 8d 7e 30 <41> 8b 2f 41 89 ec 41 89 ed 41 81 e4 ff ff ff 00 41 81 e5 c0 ff ff

The beginning is a preceding GPU hang. However the last 5 lines are something strange I can't explain: MTP Main Session and MTP Upload Session.

I had no MTP devices connected to my machine when it happenned. Moreover, _00007f5d81d4e5f6_ is a hex representation of an IP 129.212.229.246.

I checked logs, and this is the only occurence of the message since Aug 10.

Do you have any ideas, explanations, suggestions?

Thanks in advance for help!

Last edited by eDio (2020-01-10 21:18:22)

seth · 2020-01-10 21:10:02

"ip" is for "instruction pointer" - it's not an "IP" (would be fujitsu, btw)

A lot of stuff crashed at that point, probably for losing the display server after the GPU hang.

No idea what those MTP processes are, but chances are, they're running with your session or some program you ran at the time.
"ps aux | grep MTP" might show them up running again and then you can inspect the process, the binary should tell you what package they belong to.

eDio · 2020-01-10 21:18:06

> "ip" is for "instruction pointer" - it's not an "IP" (would be fujitsu, btw)

That's a huge relief.
I got suspicious when I saw something I did not expect to see, online converter I used apparently just took lower 4 bytes of that string (the string too long to be an IPv4 but I realized it only now after your response). But when conversion to IP from the hex string succeeded I got into full panic mode.

Thank you very much for explanations!

Arch Linux

#1 2020-01-10 19:35:40

[SOLVED] MTP Upload Session in dmesg

#2 2020-01-10 21:10:02

Re: [SOLVED] MTP Upload Session in dmesg

#3 2020-01-10 21:18:06

Re: [SOLVED] MTP Upload Session in dmesg

Board footer