Black screen on lid open

sinatosk · 2019-08-25 19:20:02

With kernel 5.2.10 I've noticed this change on arch linux set of patches

https://git.archlinux.org/linux.git/com … 8c3c11a479

and the kernel config has changed "CONFIG_AMD_IOMMU_V2=m" changed too "CONFIG_AMD_IOMMU_V2=y"

let us know if it's improved anything, I'm going to try it soon

Last edited by sinatosk (2019-08-25 19:21:55)

adim · 2019-08-25 19:38:14

Likun Gao wrote:

drm/amdgpu: pin the csb buffer on hw init for gfx v8

I'm pretty sure these APUs are gfx9.

There are however a few other interesting commits, like:

Pierre-Eric Pelloux-Prayer wrote:

drm/amdgpu: fix gfx9 soft recovery

Maybe I'll give it a shot.

sinatosk · 2019-08-25 19:44:30

I think you mean we have gfx 9 gpu's and yes it does look like that

I thought gfx 8 thinking vega 8 when I saw that commit

but looking at gfx 8 it does have in there VEGAM which I think means vega mobile... maybe? we are gfx 8

but then glxinfo does report RAVEN

I dunno, bloody confusing code names

Last edited by sinatosk (2019-08-25 21:12:04)

sinatosk · 2019-08-26 02:40:30

Maybe I'm speaking to soon but I've sleep/resumed over 20 times now in some hours and while I do still see the error about dcn_10_hw_sequencer, My sleep/wake issue looks like it might be gone ( this is with the amd_iommu option removed )

thats with the patch I showed in my earlier post, applied

I've not tested hibernation yet

Last edited by sinatosk (2019-08-26 11:51:00)

adim · 2019-08-26 12:13:52

sinatosk wrote:

I think you mean we have gfx 9 gpu's and yes it does look like that
I thought gfx 8 thinking vega 8 when I saw that commit
but looking at gfx 8 it does have in there VEGAM which I think means vega mobile... maybe? we are gfx 8
but then glxinfo does report RAVEN
I dunno, bloody confusing code names

Vega M refers to the GPU found in certain Intel CPUs with Vega graphics.
https://ark.intel.com/content/www/us/en … 0-ghz.html

sinatosk · 2019-08-26 12:18:59

adim wrote:

sinatosk wrote:
I think you mean we have gfx 9 gpu's and yes it does look like that
I thought gfx 8 thinking vega 8 when I saw that commit
but looking at gfx 8 it does have in there VEGAM which I think means vega mobile... maybe? we are gfx 8
but then glxinfo does report RAVEN
I dunno, bloody confusing code names
Vega M refers to the GPU found in certain Intel CPUs with Vega graphics.
https://ark.intel.com/content/www/us/en … 0-ghz.html

heh I forgot intel licensed that from amd

looking at the source code from the kernel tree at "drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c", looking at the line that says

"{0x1002, 0x15d8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RAVEN|AMD_IS_APU},"

yeah we have RAVEN which is gfx 9

Last edited by sinatosk (2019-08-26 13:07:26)

adim · 2019-09-05 12:54:21

Been using 5.2.11 for a while and I think this is gone for good now.

sinatosk · 2019-09-05 13:25:05

Yeah, things been fine for me since 5.2.10 with that patch and since 5.2.11 no issues

However I have another issue but I'll start another thread for that.

I'm unable to hibernate my system. Hibernate fails because it's unable to reset the GPU as reported by drm amdgpu

And something odd, kernel saying it's unable to find the image even though the swap partition is mounted and I've checked the path and uuid over and over to make sure it's correct.

After seeing the message about GPU failing to reset, laptop powers off, I turn it back on, then boots to try in stead of resuming. There are sometimes it does resume but it's saying it's unable to find image

Preycon · 2019-09-05 16:54:35

This still happens to me on a plasma session.

sinatosk · 2019-09-05 16:58:05

Preycon wrote:

This still happens to me on a plasma session.

Suspension or Hibernation?

sinatosk · 2019-09-06 12:54:23

ok, I've compiled and installed 5.2.13 and both suspension and hibernation works successfuly

the dcn10 warnings/errors I was talking about earlier are gone however they are patches that have not been submitted to the stable queue but a link reffered to by an AMD member

they are here. these patches are what got rid of the dcn10 warnings/errors

however I'm still seeing the error about amdgpu failed to reset on hibernate but resuming from hibernate ( so far ) seems to be working for me

edit: hibernation for me still seems to cause me problems where after hibernation... my desktop will eventually just stop responding and have to force reboot, no errors or warnings in my logs though that I see on my console. I will work on getting those warnings/errors but suspension is all good now

Last edited by sinatosk (2019-09-07 10:48:16)

Ploppz · 2019-09-09 14:48:49

With 5.2.13 I still have what seems like the same problem as before.
But `journalctl -b -1` shows only logs up until suspension - nothing after opening the lid (which was all black and unresponsivE) and forcing it to shut down.

runninghorse · 2019-09-10 08:29:53

I have this same issue. Thanks

dimischiavone · 2019-09-13 15:02:19

Hi everyone, Lenovo Thinkpad t495s aswell here with the latest bios (v1.11) and 5.2.14-arch2-1-ARCH #1 SMP PREEMPT kernel. I have issues resuming from hibernation (goes to black screen after sudden flash becoming unresponsive and needing a force reboot), whereas it resumes fine from suspend. As suggested some posts back with the kernel 5.1.16 it all works but it also worked well with the kernel 5.2.11 (even though reverting to that kernel doesn't fix it, probably due to mesa and other stuff being updated meanwhile). I guess we can wait for a future kernel release that fixes these problems (i was quite impressed that such a new machine had it all working immediately).

Last edited by dimischiavone (2019-09-13 15:03:46)

adim · 2019-09-13 23:13:04

T495s owner chiming in again. It sems like this device also suffers from some strange xhci bug during suspend, similar to the issue described here.

It was working fine on a previous BIOS but v1.11 seems to have broken it again, anways following the advice from the wiki makes that xhci error go away and make suspend work again. I was getting the same xhci errors on previous BIOS versions too but didn't think much of it, maybe it actually had something to do with broken suspends as well.

dmesg before the fix:

xhci_hcd 0000:05:00.4: WARN: xHC CMD_RUN timeout
PM: suspend_common(): xhci_pci_suspend+0x0/0xd0 [xhci_pci] returns -110
PM: pci_pm_suspend(): hcd_pci_suspend+0x0/0x30 returns -110
PM: dpm_run_callback(): pci_pm_suspend+0x0/0x120 returns -110
PM: Device 0000:05:00.4 failed to suspend async: error -110
PM: Some devices failed to suspend, or early wake event detected

Let me know if this fixes things for you @dimischiavone

dimischiavone · 2019-09-14 06:23:42

@adim resuming from suspend works fine for me but after reading the issue described in the link you mentioned i tried to suspend while plugging in a usb flash drive and got a frozen/black screen on resume (in the logs i could see the spurious wakeup mentioned in the wiki). Resuming from hibernate still doesn't work for me, i guess it might have to do with the integrated vega gpu; because on resuming i can briefly see the screen turning on and then it goes black and freezes. Other times the device doesn't even get to shutdown while hibernating, the lid blinks for a while and the screen turns off, then it turns back on with a light black colour shows the cursor and freezes.

Last edited by dimischiavone (2019-09-14 06:24:46)

Charles_Carmichael · 2019-09-15 04:22:07

Hello, HP Envy x360 here

Operating System: Arch Linux 
KDE Plasma Version: 5.16.5
KDE Frameworks Version: 5.61.0
Qt Version: 5.13.1
Kernel Version: 5.2.14-arch2-1-ARCH
OS Type: 64-bit
Processors: 8 × AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx
Memory: 6.8 GiB of RAM

Last time sleep worked for me was kernel 5.1.16. I'm not sure about hibernate as my swapfile does not like to work for hibernate. I have noticed when trying to wake from sleep my keyboard becomes unresponsive, therefore I can't switch to a virtual terminal. When I try to wake from suspend, I believe this is relevant journalctl output except it cycles through all CPU cores.

Sep 14 23:00:05 archlinux kernel: ------------[ cut here ]------------
Sep 14 23:00:05 archlinux kernel: WARNING: CPU: 7 PID: 1170 at drivers/gpu/drm/amd/amdgpu/../display/dc/dcn10/dcn10_hw_sequencer.c:854 dcn10_verify_allow_pstate_change_high.cold+0xc/0x229 [amdgpu]
Sep 14 23:00:05 archlinux kernel: Modules linked in: rfcomm ccm bnep hid_logitech_hidpp mousedev joydev amdgpu edac_mce_amd kvm_amd ccp arc4 kvm hid_multitouch hid_logitech_dj rtwpci rtw88 btusb uvcvideo btrtl btbcm irqbypass btintel vi>
Sep 14 23:00:05 archlinux kernel:  mac_hid pcc_cpufreq acpi_cpufreq vboxnetflt(OE) vboxnetadp(OE) vboxpci(OE) vboxdrv(OE) crypto_user ip_tables x_tables ext4 crc32c_generic crc16 mbcache jbd2 rtsx_pci_sdmmc mmc_core serio_raw atkbd libp>
Sep 14 23:00:05 archlinux kernel: CPU: 7 PID: 1170 Comm: kworker/u32:46 Tainted: G        W  OE     5.2.14-arch2-1-ARCH #1
Sep 14 23:00:05 archlinux kernel: Hardware name: HP HP ENVY x360 Convertible 15-cp0xxx/8497, BIOS F.19 08/20/2018
Sep 14 23:00:05 archlinux kernel: Workqueue: events_unbound commit_work [drm_kms_helper]
Sep 14 23:00:05 archlinux kernel: RIP: 0010:dcn10_verify_allow_pstate_change_high.cold+0xc/0x229 [amdgpu]
Sep 14 23:00:05 archlinux kernel: Code: 83 c8 ff e9 08 80 f9 ff 48 c7 c7 a8 d0 2e c1 e8 3b a7 2a f8 0f 0b 83 c8 ff e9 f2 7f f9 ff 48 c7 c7 a8 d0 2e c1 e8 25 a7 2a f8 <0f> 0b 80 bb 93 01 00 00 00 75 05 e9 dc a5 f9 ff 48 8b 83 80 02 00
Sep 14 23:00:05 archlinux kernel: RSP: 0018:ffff9d468378fb18 EFLAGS: 00010246
Sep 14 23:00:05 archlinux kernel: RAX: 0000000000000024 RBX: ffff97910e1dc000 RCX: 0000000000000000
Sep 14 23:00:05 archlinux kernel: RDX: 0000000000000000 RSI: 0000000000000092 RDI: 00000000ffffffff
Sep 14 23:00:05 archlinux kernel: RBP: ffff9d468378fba0 R08: 000000000002389f R09: 0000000000000001
Sep 14 23:00:05 archlinux kernel: R10: 0000000000000000 R11: 0000000000000001 R12: ffff97910e1dc000
Sep 14 23:00:05 archlinux kernel: R13: 0000000000000000 R14: ffff979114be81b8 R15: ffff979114be9318
Sep 14 23:00:05 archlinux kernel: FS:  0000000000000000(0000) GS:ffff979117bc0000(0000) knlGS:0000000000000000
Sep 14 23:00:05 archlinux kernel: CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Sep 14 23:00:05 archlinux kernel: CR2: 00007f8c5ebd6008 CR3: 0000000201674000 CR4: 00000000003406e0
Sep 14 23:00:05 archlinux kernel: Call Trace:
Sep 14 23:00:05 archlinux kernel:  dc_commit_updates_for_stream+0x91b/0xc20 [amdgpu]
Sep 14 23:00:05 archlinux kernel:  amdgpu_dm_atomic_commit_tail+0xabe/0x19a0 [amdgpu]
Sep 14 23:00:05 archlinux kernel:  ? commit_tail+0x3c/0x70 [drm_kms_helper]
Sep 14 23:00:05 archlinux kernel:  commit_tail+0x3c/0x70 [drm_kms_helper]
Sep 14 23:00:05 archlinux kernel:  process_one_work+0x1d1/0x3e0
Sep 14 23:00:05 archlinux kernel:  worker_thread+0x4a/0x3d0
Sep 14 23:00:05 archlinux kernel:  kthread+0xfb/0x130
Sep 14 23:00:05 archlinux kernel:  ? process_one_work+0x3e0/0x3e0
Sep 14 23:00:05 archlinux kernel:  ? kthread_park+0x80/0x80
Sep 14 23:00:05 archlinux kernel:  ret_from_fork+0x22/0x40
Sep 14 23:00:05 archlinux kernel: ---[ end trace f9a7093f9e5a0b3d ]---
Sep 14 23:00:05 archlinux kernel: [drm] pstate TEST_DEBUG_DATA: 0x36F60000
Sep 14 23:00:05 archlinux kernel: ------------[ cut here ]------------
Sep 14 23:00:05 archlinux kernel: WARNING: CPU: 0 PID: 1170 at drivers/gpu/drm/amd/amdgpu/../display/dc/dcn10/dcn10_hw_sequencer.c:854 dcn10_verify_allow_pstate_change_high.cold+0xc/0x229 [amdgpu]
Sep 14 23:00:05 archlinux kernel: Modules linked in: rfcomm ccm bnep hid_logitech_hidpp mousedev joydev amdgpu edac_mce_amd kvm_amd ccp arc4 kvm hid_multitouch hid_logitech_dj rtwpci rtw88 btusb uvcvideo btrtl btbcm irqbypass btintel vi>
Sep 14 23:00:05 archlinux kernel:  mac_hid pcc_cpufreq acpi_cpufreq vboxnetflt(OE) vboxnetadp(OE) vboxpci(OE) vboxdrv(OE) crypto_user ip_tables x_tables ext4 crc32c_generic crc16 mbcache jbd2 rtsx_pci_sdmmc mmc_core serio_raw atkbd libp>
Sep 14 23:00:05 archlinux kernel: CPU: 0 PID: 1170 Comm: kworker/u32:46 Tainted: G        W  OE     5.2.14-arch2-1-ARCH #1
Sep 14 23:00:05 archlinux kernel: Hardware name: HP HP ENVY x360 Convertible 15-cp0xxx/8497, BIOS F.19 08/20/2018
Sep 14 23:00:05 archlinux kernel: Workqueue: events_unbound commit_work [drm_kms_helper]
Sep 14 23:00:05 archlinux kernel: RIP: 0010:dcn10_verify_allow_pstate_change_high.cold+0xc/0x229 [amdgpu]
Sep 14 23:00:05 archlinux kernel: Code: 83 c8 ff e9 08 80 f9 ff 48 c7 c7 a8 d0 2e c1 e8 3b a7 2a f8 0f 0b 83 c8 ff e9 f2 7f f9 ff 48 c7 c7 a8 d0 2e c1 e8 25 a7 2a f8 <0f> 0b 80 bb 93 01 00 00 00 75 05 e9 dc a5 f9 ff 48 8b 83 80 02 00
Sep 14 23:00:05 archlinux kernel: RSP: 0018:ffff9d468378faf8 EFLAGS: 00010246
Sep 14 23:00:05 archlinux kernel: RAX: 0000000000000024 RBX: ffff97910e1dc000 RCX: 0000000000000000
Sep 14 23:00:05 archlinux kernel: RDX: 0000000000000000 RSI: 0000000000000096 RDI: 00000000ffffffff
Sep 14 23:00:05 archlinux kernel: RBP: ffff97910e1dc000 R08: 00000000000238be R09: 0000000000000001
Sep 14 23:00:05 archlinux kernel: R10: 0000000000000000 R11: 0000000000000001 R12: ffff979114be81b8
Sep 14 23:00:05 archlinux kernel: R13: 0000000000000000 R14: ffff979114be81b8 R15: 0000000000000004
Sep 14 23:00:05 archlinux kernel: FS:  0000000000000000(0000) GS:ffff979117a00000(0000) knlGS:0000000000000000
Sep 14 23:00:05 archlinux kernel: CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Sep 14 23:00:05 archlinux kernel: CR2: 00007f1d9f05e000 CR3: 000000021310a000 CR4: 00000000003406f0
Sep 14 23:00:05 archlinux kernel: Call Trace:
Sep 14 23:00:05 archlinux kernel:  dcn10_pipe_control_lock.part.0+0x69/0x70 [amdgpu]
Sep 14 23:00:05 archlinux kernel:  dc_commit_updates_for_stream+0x84c/0xc20 [amdgpu]
Sep 14 23:00:05 archlinux kernel:  amdgpu_dm_atomic_commit_tail+0xabe/0x19a0 [amdgpu]
Sep 14 23:00:05 archlinux kernel:  ? commit_tail+0x3c/0x70 [drm_kms_helper]
Sep 14 23:00:05 archlinux kernel:  commit_tail+0x3c/0x70 [drm_kms_helper]
Sep 14 23:00:05 archlinux kernel:  process_one_work+0x1d1/0x3e0
Sep 14 23:00:05 archlinux kernel:  worker_thread+0x4a/0x3d0
Sep 14 23:00:05 archlinux kernel:  kthread+0xfb/0x130
Sep 14 23:00:05 archlinux kernel:  ? process_one_work+0x3e0/0x3e0
Sep 14 23:00:05 archlinux kernel:  ? kthread_park+0x80/0x80
Sep 14 23:00:05 archlinux kernel:  ret_from_fork+0x22/0x40
Sep 14 23:00:05 archlinux kernel: ---[ end trace f9a7093f9e5a0b3e ]---
Sep 14 23:00:05 archlinux kernel: [drm] pstate TEST_DEBUG_DATA: 0x36F60000
Sep 14 23:00:05 archlinux kernel: ------------[ cut here ]------------

dimischiavone · 2019-09-15 08:16:47

@Charles_Carmichael i have the same setup as you on my t495s but resuming from suspend works fine with this kernel. I noticed that i have the same output as you for hibernating though.
Edit: I replaced sddm + kde plasma with gdm + gnome on the same kernel and now i can succesfully hibernate and suspend with no problems.

Last edited by dimischiavone (2019-09-16 21:35:29)

adim · 2019-09-17 21:37:01

Probably best to report this to the amdgpu kernel bugtracker where AMD devs can actually see this, I already made a report on https://bugs.freedesktop.org/show_bug.cgi?id=111244 but there's a few other possibly related ones.

adim · 2019-09-20 15:37:02

Seems like there's a commit in drm-fixes-5.4 that might fix this: https://cgit.freedesktop.org/~agd5f/lin … a3767710f7

Will try 5.4 as soon as the first mainline release comes out, if someone else here has the time then you may try a kernel containing this patch.
This will *hopefully* fully fix this issue for me since i still rarely happens and when it happens dmesg shows sdma0 errors.

ambra · 2019-10-01 12:15:47

I have (probably) the same issue.
Here's what I've found out that might help.

I have problem with waking up only when I enable sddm service. If I disable it, there are no problems with waking up. So, I guess it has something to do with graphic drivers.

I noticed this problem as soon as I got this laptop (ACER Aspire 3, A315-41G-R0ZC with AMD Ryzen 5 2500U and AMD Radeon 535).

The Wiki suggested using LTS kernel, and I gave it a try. With LTS kernel, the problem was gone, just until last weekend update when the kernel was updated from 4.19.73 to 4.19.75. I switched back to 4.19.73 (pacman -U), but the problem remained. So, it's not only the kernel, but some combination of packages.
I also tried reverting back to systemd-243.0, but it didn't help.

I'll try suggested kernel parameter.

EDIT:
amd_iommu=off doesn't help in my case. I tried it with LTS kernel and 5.3.1.

Last edited by ambra (2019-10-01 12:47:32)

adim · 2019-10-04 13:43:26

adim wrote:

Seems like there's a commit in drm-fixes-5.4 that might fix this: https://cgit.freedesktop.org/~agd5f/lin … a3767710f7
Will try 5.4 as soon as the first mainline release comes out, if someone else here has the time then you may try a kernel containing this patch.
This will *hopefully* fully fix this issue for me since i still rarely happens and when it happens dmesg shows sdma0 errors.

Been running 5.4rc1 for the past few days and it's been running fine, cannot reproduce this anymore. Seems like this commit did indeed fix it for me!

Preycon · 2019-10-08 01:12:30

dimischiavone wrote:

@Charles_Carmichael i have the same setup as you on my t495s but resuming from suspend works fine with this kernel. I noticed that i have the same output as you for hibernating though.
Edit: I replaced sddm + kde plasma with gdm + gnome on the same kernel and now i can succesfully hibernate and suspend with no problems.

Yes, I did this too, anyone can suspend AND hibernate ok with Plasma Desktop?

sinatosk · 2019-10-12 14:47:33

Just a reminder, any of you try the patches I've mentioned at a previous post. these patches stopped my suspension issue on 5.2 and now on 5.3 and they staged to be released in 5.4 and had no suspension issues whats so ever

I've still got hibernation issues but have noticed a bunch of patch related to the amdgpu reset mode1... apparently the APU's don't support mode1 reset and staged in 5.4 are patches that implement mode2 reset which is supported by our APU's but I'm unable to work out the only need patches for the mode2 reset.

I compile my own kernels

nnscr · 2019-11-13 18:20:53

Hello,

I've had the exact same problem on my Thinkpad A485 (AMD Ryzen 7 PRO 2700U w/ Radeon Vega Mobile Gfx) and found a solution which allows me to run X, the latest stock kernels and suspending works like a charm.

I simply forced DRI 2 instead of DRI 3, which seems to be buggy with Raven Ridge APUs and Xorg (Wayland worked fine). It also fixed some other small graphic related problems that I used to have.

Simply paste this into your xorg config:

Section "Device" 
	Identifier "AMDGPU" 
	Driver "amdgpu" 
	Option "DRI" "2" 
EndSection

My command line flags are these, but some of them might not be needed anymore:

idle=nomwait amd_iommu=fullflush amdgpu.gpu_recovery=1

Hope this helps someone!

Arch Linux

#26 2019-08-25 19:20:02

Re: Black screen on lid open

#27 2019-08-25 19:38:14

Re: Black screen on lid open

#28 2019-08-25 19:44:30

Re: Black screen on lid open

#29 2019-08-26 02:40:30

Re: Black screen on lid open

#30 2019-08-26 12:13:52

Re: Black screen on lid open

#31 2019-08-26 12:18:59

Re: Black screen on lid open

#32 2019-09-05 12:54:21

Re: Black screen on lid open

#33 2019-09-05 13:25:05

Re: Black screen on lid open

#34 2019-09-05 16:54:35

Re: Black screen on lid open

#35 2019-09-05 16:58:05

Re: Black screen on lid open

#36 2019-09-06 12:54:23

Re: Black screen on lid open

#37 2019-09-09 14:48:49

Re: Black screen on lid open

#38 2019-09-10 08:29:53

Re: Black screen on lid open

#39 2019-09-13 15:02:19

Re: Black screen on lid open

#40 2019-09-13 23:13:04

Re: Black screen on lid open

#41 2019-09-14 06:23:42

Re: Black screen on lid open

#42 2019-09-15 04:22:07

Re: Black screen on lid open

#43 2019-09-15 08:16:47

Re: Black screen on lid open

#44 2019-09-17 21:37:01

Re: Black screen on lid open

#45 2019-09-20 15:37:02

Re: Black screen on lid open

#46 2019-10-01 12:15:47

Re: Black screen on lid open

#47 2019-10-04 13:43:26

Re: Black screen on lid open

#48 2019-10-08 01:12:30

Re: Black screen on lid open

#49 2019-10-12 14:47:33

Re: Black screen on lid open

#50 2019-11-13 18:20:53

Re: Black screen on lid open

Board footer