460.56 no desktop on resume from suspend to ram

nvidia-bug-report.log.gz (1.0 MB)

With 460.56 I get below errors on Ubuntu 20.10 on Dell Inspiron 7577 laptop with nVIDIA GeForce GTX 1060 with max-Q design (and Intel GPU as well but Prime setting is NVIDIA performance mode):
grep DMA /var/log/syslog
Mar 7 15:50:02 Dell7577Linux kernel: [25809.607959] nvidia-modeset: ERROR: GPU:0: Failed to allocate a DMA context
Mar 7 15:50:02 Dell7577Linux kernel: [25809.608116] nvidia-modeset: ERROR: GPU:0: Notifier DMA allocation failed
Mar 7 15:50:02 Dell7577Linux kernel: [25809.608261] nvidia-modeset: ERROR: GPU:0: Failed to allocate display engine core DMA push buffer
I get no desktop any more after resume from suspend to RAM, only above errors in text display.
My laptop display is 4K 3840x2160 if that would have any influence.

4 Likes

I have the same problem after suspend I have a black screen. No problem before updating to nvidia-460.

1 Like

Same problema here with Arch Linux.

2 Likes

460.67 same behaviour.nvidia-bug-report.log.gz (1001.8 KB)

2 Likes

@ygcaicn @mario156090
Could you please share nvidia bug report fetched in repro state.
Would like to know if dell system is affected or other OEM systems as well.

1 Like

Exact the same Error on popular linux cinemon and Lenovo P51.
“Linux Mint 20”| GM206GLM [Quadro M2200 Mobile]| NVIDIA-SMI 460.67
Thank you a lot for all the great LINUX community work in the last years !!!

1 Like

Here is a bug report from a Lenovo P50 with a frozen state (I ssh’ed in remotely)

nvidia-bug-report.log.gz (228 KB)

1 Like

I have this error on resume from suspend on a hp pavilion 15-cx0999na, running LM19.3 and driver 460.67 (and on previous 460 line versions before upgrading to latest on the PPA):
[ 9961.069146] pcieport 0000:00:1b.3: PCIe Bus Error: severity=Corrected, type=Physical Layer, id=00db(Transmitter ID)
[ 9961.069147] pcieport 0000:00:1b.3: device [8086:a343] error status/mask=00001001/000020000
[ 9961.069148] pcieport 0000:00:1b.3: [ 0] Receiver Error (First)
[ 9961.069149] pcieport 0000:00:1b.3: Replay Timer Timeout
[10060.664837] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])
[10060.665048] nvidia-modeset: ERROR: GPU:0: Failed to allocate display core DMA push buffer
[10064.664773] nvidia-modeset: ERROR: GPU:0: Display engine push buffer channel allocation failed: 0x65 (Call timed out [NV_ERR_TIMEOUT])
[10064.664968] nvidia-modeset: ERROR: GPU:0: Failed to allocate display core DMA push buffer
When this appears on screen after several seconds of being blank with just “_”, the cooling fan starts spinning up to maximum as if it’s stuck in a loop on something and I just end up hard-resetting.

If I switch back to 450.102.04, resume from suspend works on the 450 line, but then OBS Studio will not start recording on that driver version, throwing up an error:
Failed to open NVENC codec: Function not implemented

Please check your video drivers are up to date.

@ me if you need more info. I’ll need pointers on how to find the relevant error log.

2 Likes

With 465.24.02 I get the same errors on Ubuntu 20.10 on Dell Inspiron 7577 laptop with nVIDIA GeForce GTX 1060 with max-Q design (and Intel GPU as well but Prime setting is NVIDIA performance mode)
nvidia-bug-report.log.gz (1.0 MB)

1 Like

I got the this problem when my pc go to sleep mode and come back.
It doesn’t wake up and just shows messages below. I’m using Lenovo P50 and Ubuntu 20.04 LTS

2 Likes

My Dell Inspiron 7577 laptop with nVIDIA GeForce GTX 1060 with max-Q design (and Intel GPU as well but Prime setting is NVIDIA performance mode) now runs Ubuntu 21.04 with Gnome on Wayland with driver 465.24.02. Previously it was using Xorg. Wayland runs ok as long as you do not need OpenGL hardware acceleration. And the great news is resume from suspend to RAM just WORKS. Though it takes 1 minute 45 seconds from power button to lock screen.

1 Like

As described in my other post https://forums.developer.nvidia.com/t/465-24-02-ubuntu-21-04-wayland-no-external-monitor/176747/4 I came to the conclusion that my integrated Intel GPU is driving the display and probably that is why resume from suspend to RAM works now. Although the answer of “prime-select query” is “nvidia” so very confusing…

1 Like

The same since Ubuntu 20.10, no variation of driver/kernel/options helps.
nvidia-bug-report.log.gz (422.3 KB)

1 Like

Same issue with Ubuntu 20.04 on ThinkPad P50.

1 Like

This worked for me. I changed the BIOS setting from
“hybrid graphics” to “discrete graphics”.

Linux Mint 20.04,
Lenovo P51,
Nvidia Quatro M2200
Driver: 465.27

Im using pm-suspend for suspend mode.

I hope that helps you too.

I have the exact same hardware, Linux Mint installation and driver version as you, Tom, and suspend does not work for me.

Where is Nvidia support? Surely M2200 is not the only chipset effected? I am pretty ticked off that I cannot return to the original 450 series drivers. I just downgraded to the 390 driver and my suspend works again. Once again, propriety drivers screw me over.

Was also having this on anything using 460.56 or greater. Rolling back to 460.39 fixed it.

Mint 20.1 (Ubuntu 20.04)
GTX 980

I noticed that even though my displays would not work I could get in via SSH.

Perhaps this bit might help?

[ 288.661581] WARNING: CPU: 5 PID: 10278 at /var/lib/dkms/nvidia/465.27/build/nvidia/nv.c:3909 nv_restore_user_channels+0xce/0xe0 [nvidia]

Suspend was working for me in driver version 460.80, but after update to version 460.84 the above error started appearing again. Can anyone else confirm this difference in fix/regression?

A bunch of other packages were updated at the same time, as I assumed the driver was fixed and didn’t think to separate it out. If I can find an easy enough way to roll it back individually then I will.