System Freeze with "Resetting rcs0 for hang on rcs0"

[sanette]

* The problem started with linux kernel 5.3.0-46

* When the screen (GPU?) is used intensively, I can hear coil noise from my laptop (Dell XPS 13)
and then the whole system will freeze during 5-6 seconds. During freeze I can move the mouse cursor, but nothing else is responding. During freeze, the noise stops. Then the systems comes back to normal for a few seconds (randomly from 10sec to 1 minute), and then the freeze repeats again and again.

* According to my experience, I would strongly bet that the coil noise is due to graphics.

* The hang is reported in the dmesg log:

[ 64.369654] i915 0000:00:02.0: GPU HANG: ecode 9:0:0x00000000, hang on rcs0
[ 64.370664] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[ 162.345195] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[ 192.360856] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[ 206.312567] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[ 230.343937] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0
[ 254.343235] i915 0000:00:02.0: Resetting rcs0 for hang on rcs0

* I have a small demo which uses the GPU intensively (via the SDL library) and I can repeat the freeze easily by running it.

* The problem does *not* occur with 5.3.0-45-generic (no noise, no freeze)

[sanette (sanette-linux)]

I first posted in https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1861395
because I thought it was related

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This change was made by a bot.

[pcworld (pcworld)]

I think this is a duplicate of bug #1872001, and probably also the same as bug #1871867. Can you share your test program?

[sanette (sanette-linux)]

you're right, it seems to be exactly the same bug.
I am attaching my test program (binary: you need the SDL library, and maybe more)

./test.exe

it will open a graphics window and run a simulation of virus transmission
(human virus, not computer virus, there is no risk for the computer ;) )