GPU HANG: ecode 9:1:85dffffb in Xorg

GPU crashes multiple times per day when an external display is connected (HDMI or DP). I haven't observed similar hangs when the external display is not connected.

OS:

• Fedora 33
• kernel 5.11.0-0.rc6.141.vanilla.1.fc33.x86_64

Reproducible also on Fedora 32, and F33 with the default kernel (v5.10).

Hardware:

Manufacturer: LENOVO
Product Name: 20EQS64N0G
Version: ThinkPad P50

00:02.0 VGA compatible controller: Intel Corporation HD Graphics 530 (rev 06)

[ 4087.986312] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 4087.986331] i915 0000:00:02.0: [drm] Xorg[1505] context reset due to GPU hang
[ 4088.003106] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1505]

card_error

See also #3042 (closed), as that looks identical (f33, v5.10.11).

Do you have any older kernels to test?

See also #3042 (closed), as that looks identical (f33, v5.10.11).

Yes, it looks identical. There are also two other reports which look very similar (Lenovo ThinkPad P50 + external display) #2739 #2668

Do you have any older kernels to test?

This is not a new issue, my GPU hanged for the first time probably after upgrade to Fedora 31 (kernel v5.3). The same problem existed on Fedora 32 (kernel v5.6).

Yesterday I upgraded to 5.11.0-0.rc6 and this bug is still reproducible, but it looks that it is not so frequent as on v5.10.

If you want I can downgrade the kernel and I can check if this is reproducible with a given kernel version. Let me know what you need.

mentioned in issue #3042 (closed)

GPU hanged again, kernel 5.11.0-0.rc6.141.vanilla.1.fc33.x86_64:

[Fri Feb  5 16:48:02 2021] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[Fri Feb  5 16:48:02 2021] i915 0000:00:02.0: [drm] Xorg[1541] context reset due to GPU hang
[Fri Feb  5 16:48:02 2021] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1541]

error_2

One more hang, 5.11.0-0.rc6.20210204git61556703b610.144.vanilla.1.fc33.x86_64:

[ 2260.139518] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 2260.139536] i915 0000:00:02.0: [drm] Xorg[1432] context reset due to GPU hang
[ 2260.146412] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1432]

error_3

One more hang. I was wrong about v5.11-r6, it as unstable as v5.10.

[ 6705.619228] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 6705.619236] i915 0000:00:02.0: [drm] Xorg[1462] context reset due to GPU hang
[ 6705.624746] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1462]

error_4

One more hang:

[ 1962.663568] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 1962.663576] i915 0000:00:02.0: [drm] Xorg[1572] context reset due to GPU hang
[ 1962.667720] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1572]

error_5

One more:

[ 2521.637602] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 2521.637610] i915 0000:00:02.0: [drm] Xorg[4350] context reset due to GPU hang
[ 2521.643954] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [4350]
[ 2521.643956] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[ 2521.643957] Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/intel/issues/new.
[ 2521.643957] Please see https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs for details.
[ 2521.643957] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[ 2521.643958] The GPU crash dump is required to analyze GPU hangs, so please always attach it.
[ 2521.643958] GPU crash dump saved to /sys/class/drm/card1/error
[ 2521.817814] fbcon: Taking over console
[ 2521.818474] Console: switching to colour frame buffer device 240x67
[ 2534.052433] Asynchronous wait on fence 0000:00:02.0:kwin_x11[5992]:128c4 timed out (hint:intel_atomic_commit_ready [i915])
[ 2536.613556] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 2536.613564] i915 0000:00:02.0: [drm] Xorg[4350] context reset due to GPU hang
[ 2536.620088] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dfbfff, in Xorg [4350]

error_6

Yet another one:

[ 2580.645392] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 2580.645400] i915 0000:00:02.0: [drm] Xorg[1535] context reset due to GPU hang
[ 2580.649365] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1535]

error_7

One more hang:

[ 6626.579743] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 6626.579755] i915 0000:00:02.0: [drm] Xorg[1584] context reset due to GPU hang
[ 6626.589290] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1584]
[ 6626.589294] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[ 6626.589295] Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/intel/issues/new.
[ 6626.589296] Please see https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs for details.
[ 6626.589297] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[ 6626.589298] The GPU crash dump is required to analyze GPU hangs, so please always attach it.
[ 6626.589299] GPU crash dump saved to /sys/class/drm/card0/error
[ 6626.759667] fbcon: Taking over console
[ 6626.770171] Console: switching to colour frame buffer device 240x67
[ 6630.610731] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 6630.610739] i915 0000:00:02.0: [drm] Xorg[1584] context reset due to GPU hang
[ 6630.615348] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dfbfff, in Xorg [1584]
[ 6690.450482] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 6690.450491] i915 0000:00:02.0: [drm] Xorg[1584] context reset due to GPU hang
[ 6690.454791] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1584]

error_8

Can you try this:

#3042 (comment 791540)

It's not really a "fix" but a workaround. But wondering if you have luck. So far I havnt had an issue, but im measuring only in hours, while before I was seeing the issue every 3 to 5 minutes. Wondering if it helps you too.

I switched to NVIDIA GPU and so far it works fine. I haven't tried this before because 2 years ago nouveau was broken with my GPU, but now it works surprisingly well, even with an external display. Thanks.

Cool. Yea, I'm not doing anything too graphics intensive on this machine, so I was hoping to use "Discrete Only" with wayland/nouveau. However, I did end up installing Nvidia drivers to see if there is a noticable difference, and it does feel a bit smoother, so I'm going to stick with xorg/nvidia on this P50 for now.

Seems fixed on Fedora 34 with 5.11

I tried 5.11.0-155.vanilla.1.fc33 on Fedora 33 and this hang was still reproducible.

I'm on a i7-4770S with F34 using the IGD with kernel 5.11.0-156.fc34.x86_64 and Wayland, it's been working. With the kernel-5.10.17-200.fc33 on F34 works as well. But with the kernel-5.11.1-300.fc34 graphic jitters everywhere.

I am experiencing the same. It happens when the HDMI is attached and used. Doing compute-intensive stuff such as compiling, gaming, or watching 4k videos seems to trigger it. It happens both with Wayland and Xorg.

error.bz2

# dmesg
...
Mar 18 17:52:11 monolith kernel: i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
Mar 18 17:52:11 monolith kernel: i915 0000:00:02.0: [drm] Xorg[1691] context reset due to GPU hang
Mar 18 17:52:11 monolith kernel: i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in Xorg [1691]
...

I am running Arch GNU/Linux.

# uname -a
Linux monolith 5.11.6-arch1-1 #1 SMP PREEMPT Thu, 11 Mar 2021 13:48:23 +0000 x86_64 GNU/Linux

My graphic cards:

# lspci -k | grep -A 2 -E "(VGA|3D)"
00:02.0 VGA compatible controller: Intel Corporation HD Graphics 530 (rev 06)
	Subsystem: CLEVO/KAPOK Computer Device 1501
	Kernel driver in use: i915
--
01:00.0 VGA compatible controller: NVIDIA Corporation GM206M [GeForce GTX 965M] (rev a1)
	Subsystem: CLEVO/KAPOK Computer Device 1501
	Kernel driver in use: nvidia

My motherboard

# dmidecode -t 2
# dmidecode 3.3
Getting SMBIOS data from sysfs.
SMBIOS 3.0.0 present.

Handle 0x0002, DMI type 2, 15 bytes
Base Board Information
	Manufacturer: Notebook
	Product Name: N15_N17RF1
	Version: Not Applicable
	Serial Number: Not Applicable
	Asset Tag: Tag 12345
	Features:
		Board is a hosting board
		Board is replaceable
	Location In Chassis: Not Applicable
	Chassis Handle: 0x0003
	Type: Motherboard
	Contained Object Handles: 0

My BIOS

# dmidecode -t bios
# dmidecode 3.3
Getting SMBIOS data from sysfs.
SMBIOS 3.0.0 present.

Handle 0x0000, DMI type 0, 24 bytes
BIOS Information
	Vendor: American Megatrends Inc.
	Version: 1.05.04
	Release Date: 06/08/2016
	Address: 0xF0000
	Runtime Size: 64 kB
	ROM Size: 5 MB
	Characteristics:
		PCI is supported
		BIOS is upgradeable
		BIOS shadowing is allowed
		Boot from CD is supported
		Selectable boot is supported
		BIOS ROM is socketed
		EDD is supported
		Print screen service is supported (int 5h)
		8042 keyboard services are supported (int 9h)
		Printer services are supported (int 17h)
		ACPI is supported
		USB legacy is supported
		BIOS boot specification is supported
		Targeted content distribution is supported
		UEFI is supported
	BIOS Revision: 5.11

Handle 0x0029, DMI type 13, 22 bytes
BIOS Language Information
	Language Description Format: Long
	Installable Languages: 1
		en|US|iso8859-1
	Currently Installed Language: en|US|iso8859-1

My CPU

# dmidecode -t 4
# dmidecode 3.3
Getting SMBIOS data from sysfs.
SMBIOS 3.0.0 present.

Handle 0x0017, DMI type 4, 48 bytes
Processor Information
	Socket Designation: U3E1
	Type: Central Processor
	Family: Core i7
	Manufacturer: Intel(R) Corporation
	ID: E3 06 05 00 FF FB EB BF
	Signature: Type 0, Family 6, Model 94, Stepping 3
	Flags:
		FPU (Floating-point unit on-chip)
		VME (Virtual mode extension)
		DE (Debugging extension)
		PSE (Page size extension)
		TSC (Time stamp counter)
		MSR (Model specific registers)
		PAE (Physical address extension)
		MCE (Machine check exception)
		CX8 (CMPXCHG8 instruction supported)
		APIC (On-chip APIC hardware supported)
		SEP (Fast system call)
		MTRR (Memory type range registers)
		PGE (Page global enable)
		MCA (Machine check architecture)
		CMOV (Conditional move instruction supported)
		PAT (Page attribute table)
		PSE-36 (36-bit page size extension)
		CLFSH (CLFLUSH instruction supported)
		DS (Debug store)
		ACPI (ACPI supported)
		MMX (MMX technology supported)
		FXSR (FXSAVE and FXSTOR instructions supported)
		SSE (Streaming SIMD extensions)
		SSE2 (Streaming SIMD extensions 2)
		SS (Self-snoop)
		HTT (Multi-threading)
		TM (Thermal monitor supported)
		PBE (Pending break enabled)
	Version: Intel(R) Core(TM) i7-6700HQ CPU @ 2.60GHz
	Voltage: 1.1 V
	External Clock: 100 MHz
	Max Speed: 8300 MHz
	Current Speed: 3100 MHz
	Status: Populated, Enabled
	Upgrade: Other
	L1 Cache Handle: 0x0014
	L2 Cache Handle: 0x0015
	L3 Cache Handle: 0x0016
	Serial Number: To Be Filled By O.E.M.
	Asset Tag: To Be Filled By O.E.M.
	Part Number: To Be Filled By O.E.M.
	Core Count: 4
	Core Enabled: 4
	Thread Count: 8
	Characteristics:
		64-bit capable
		Multi-Core
		Hardware Thread
		Execute Protection
		Enhanced Virtualization
		Power/Performance Control

I thought my Nvidia card and/or drivers was causing it, so I tried both Nvidia and Nouveau but neither fixes it. Also tried i915.enable_rc6=0 cmdline but this doesn't help as well. Unfortunately, I think my BIOS does not have a discrete graphics switch.

If there is other information that you require/assist with debugging let me know. I will happily help fix this annoying bug.

I just encountered this problem as well. The GPU crashes, causing SDDM to completely reload into the login screen.

OS: Pop!_OS 21.04 x86_64 
Kernel: 5.11.0-7620-generic 
Uptime: 23 days, 20 hours, 51 mins 
Shell: bash 5.1.4 
Resolution: 3840x2160 
DE: Plasma 5.21.4 
WM: KWin 
CPU: Intel i9-10900K (20) @ 5.300GHz 
GPU: Intel CometLake-S GT2 [UHD Graphics 630] 
Memory: 64092MiB 

No discrete GPU, I am only using the onboard graphcs that came with the i9-10900K with the i915 driver available in Kernel 5.11.0-7620-generic

The primary purpose of this workstation is cloud engineering and at the time of the crash, minikube, chrome and IntelliJ IDEA were running.

Here's the stack trace:

8/24/21 2:11 AM		Asynchronous wait on fence 0000:00:02.0:kwin_x11[1788004]:1b128e0 timed out (hint:intel_atomic_commit_ready [i915])
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] Resetting rcs0 for preemption time out
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] Xorg[1787042] context reset due to GPU hang
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] GPU HANG: ecode 9:1:84dfbffc, in Xorg [1787042]
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] Resetting rcs0 for preemption time out
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] Xorg[1787042] context reset due to GPU hang
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] GPU HANG: ecode 9:1:85dfffff, in Xorg [1787042]
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] Resetting rcs0 for preemption time out
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] Xorg[1787042] context reset due to GPU hang
8/24/21 2:11 AM	i915 0000	0:02.0: [drm] GPU HANG: ecode 9:1:87cabff2, in Xorg [1787042]

mentioned in issue #4096 (closed)

#4858 just leaving this here.

The issue still exists on 5.16.6-200.fc35.x86_64 kernel, but I've found a workaround:

# echo 400 > /sys/class/drm/card1/gt_boost_freq_mhz
# echo 400 > /sys/class/drm/card1/gt_min_freq_mhz 

After this change the GPU works stable.

mentioned in issue #4858

closed

reopened

marked this issue as a duplicate of #4858