Bug 90987

sigh ... don't know how this got fouled up - this sentence 

the only difference with this kernel command option (radeon.dpm=0) is that the messages do not appear; the symptoms stay - the black xorg TTY, killing X from another TTY.  



should be attached to this section: 

when this started one thing I tried was changing the kernel command line

GRUB_CMDLINE_LINUX_DEFAULT=" md_mod.start_ro=1 radeon.audio=1 radeon.dpm=0 "

from here:

https://bugs.freedesktop.org/show_bug.cgi?id=90340

I tried the xrandr script, both from a regular terminal and from a kmscon.

Also tried changing the resolution to several other values.

It seemed to have no effect - the blank X still stayed blank; after running these xrandr commands and switching to that TTY the monitor reports "no signal"

Seems you've faced infamous DP Link training issue. Try to boot with recent kernel like 4.3-RC3 or so and check if your problem fixed. I can see bunch of patches dealing with similar conditions.

Can confirm, I see this, too:

kernel: [drm:radeon_dp_link_train] *ERROR* displayport link status failed
kernel: [drm:radeon_dp_link_train] *ERROR* clock recovery failed
kernel: [drm:radeon_dp_link_train] *ERROR* displayport link status failed
kernel: [drm:radeon_dp_link_train] *ERROR* clock recovery failed

The screen is an Eizo EV2455. GPU is an AMD Kaveri on Linux 4.4.0.

What I noticed is, that I only get this, if I set
Option      "TearFree" "Off"

and maybe (didn't test explicitely, but it was set at the time):
Option      "DRI" "3"

Should note: The screen is and stays black, but I can switch to VT and back, just like mentioned in the first comment.

(In reply to Bernd Steinhauser from comment #4)
> 
> What I noticed is, that I only get this, if I set
> Option      "TearFree" "Off"

TearFree is disabled by default. Do you mean that enabling it avoids the problem? If so, does Option "EnablePageFlip" "off" avoid the problem as well?

(In reply to Michel Dänzer from comment #6)
> (In reply to Bernd Steinhauser from comment #4)
> > 
> > What I noticed is, that I only get this, if I set
> > Option      "TearFree" "Off"
> 
> TearFree is disabled by default. Do you mean that enabling it avoids the
> problem? If so, does Option "EnablePageFlip" "off" avoid the problem as well?

That works as well, so since TearFree does implicate that, it's probably EnablePageFlip, that fixes it.

Please attach the full Xorg log and output of dmesg captured after reproducing the problem.

Created attachment 121360 [details]
dmesg output

Created attachment 121361 [details]
Xorg log

I think I was able to reproduce this bug with linux 4.2.5, but the behavior was slightly different.
The screens went black for about 2s (pointer was still visible), just like described, but then things returned to normal.
This was with the pageflip option set to on. With off, I wasn't able to reproduce this.

I can confirm the same effect for R9 270X + Dell U3415W connected via DisplayPort.

The other monitor (Dell 2007WFP) connected via DVI works fine.

Latest kernel tested: 4.5.0-rc3-00071-g17a5cd3

Don't know if its related but I also see this every boot:

radeon 0000:01:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x8b00

[drm:si_dpm_set_power_state [radeon]] *ERROR* si_restrict_performance_levels_before_switch failed

Also note that this is the same card as in #76490.

(In reply to Daniel Exner from comment #12)
> I can confirm the same effect for R9 270X + Dell U3415W connected via
> DisplayPort.
My Dell U2415 has a setting "DP 1.2" in the display submenu. So maybe yours has that, too.
If so, do you have that activated and does the effect occur if you disable that?

Switching DP 1.2 off in the menue helps in the cases where the monitor went into DPMS indeed.

But it doesn't help for the "warm reboot" cases.

Still get this:

Feb 15 20:50:13 Joshua kernel:  [<ffffffffa08cc01a>] ? radeon_connector_hotplug+0xea/0xf0 [radeon]
Feb 15 20:50:14 Joshua kernel:  [<ffffffffa08d9d82>] ? radeon_dp_work_func+0x32/0x50 [radeon]
Feb 15 20:50:14 Joshua kernel:  [<ffffffff8106a6ac>] ? process_one_work+0x11c/0x3b0
Feb 15 20:50:14 Joshua kernel:  [<ffffffff8106a982>] ? worker_thread+0x42/0x4c0
Feb 15 20:50:14 Joshua kernel:  [<ffffffff8106a940>] ? process_one_work+0x3b0/0x3b0
Feb 15 20:50:14 Joshua kernel:  [<ffffffff8106fd08>] ? kthread+0xb8/0xd0
Feb 15 20:50:14 Joshua kernel:  [<ffffffff8106fc50>] ? kthread_worker_fn+0x170/0x170
Feb 15 20:50:14 Joshua kernel:  [<ffffffff814770cf>] ? ret_from_fork+0x3f/0x70
Feb 15 20:50:14 Joshua kernel:  [<ffffffff8106fc50>] ? kthread_worker_fn+0x170/0x170
Feb 15 20:50:14 Joshua kernel: ---[ end trace 446f7a700841052b ]---
Feb 15 20:50:14 Joshua kernel: ------------[ cut here ]------------
Feb 15 20:50:14 Joshua kernel: WARNING: CPU: 5 PID: 119 at include/drm/drm_crtc.h:2555 drm_helper_choose_crtc_dpms+0x87/0x90 [drm_kms_helper]()
Feb 15 20:50:14 Joshua kernel: Modules linked in: af_packet rfcomm bnep nf_conntrack_ipv4 nf_defrag_ipv4 xt_tcpudp xt_limit xt_conntrack nf_conntrack xt_multiport iptable_
Feb 15 20:50:14 Joshua kernel:  snd_hda_codec_hdmi sp5100_tco sysfillrect ac97_bus snd_hda_intel sysimgblt i2c_piix4 snd_hda_codec i2c_core acpi_cpufreq tpm_tis snd_hda_co
Feb 15 20:50:14 Joshua kernel: CPU: 5 PID: 119 Comm: kworker/5:1 Tainted: G        W       4.5.0-rc4-00031-g1ce457e-dirty #10
Feb 15 20:50:14 Joshua kernel: Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./970 Extreme3 R2.0, BIOS P2.20 08/05/2015
Feb 15 20:50:14 Joshua kernel: Workqueue: events radeon_dp_work_func [radeon]
Feb 15 20:50:14 Joshua kernel:  0000000000000000 ffffffff8123c413 0000000000000000 ffffffffa02f6540
Feb 15 20:50:15 Joshua kernel:  ffffffff81056153 ffff8800bda98000 ffff880234da5000 ffff880234da5000
Feb 15 20:50:15 Joshua kernel:  0000000000000003 0000000000000000 ffffffffa02e61b7 ffff88021a103a00
Feb 15 20:50:15 Joshua kernel: Call Trace:
Feb 15 20:50:15 Joshua kernel:  [<ffffffff8123c413>] ? dump_stack+0x5c/0x79
Feb 15 20:50:15 Joshua kernel:  [<ffffffff81056153>] ? warn_slowpath_common+0x73/0xa0
Feb 15 20:50:15 Joshua kernel:  [<ffffffffa02e61b7>] ? drm_helper_choose_crtc_dpms+0x87/0x90 [drm_kms_helper]
Feb 15 20:50:15 Joshua kernel:  [<ffffffffa02e6517>] ? drm_helper_connector_dpms+0x67/0xe0 [drm_kms_helper]
Feb 15 20:50:15 Joshua kernel:  [<ffffffffa08cc01a>] ? radeon_connector_hotplug+0xea/0xf0 [radeon]
Feb 15 20:50:15 Joshua kernel:  [<ffffffffa08d9d82>] ? radeon_dp_work_func+0x32/0x50 [radeon]
Feb 15 20:50:15 Joshua kernel:  [<ffffffff8106a6ac>] ? process_one_work+0x11c/0x3b0
Feb 15 20:50:15 Joshua kernel:  [<ffffffff8106a982>] ? worker_thread+0x42/0x4c0
Feb 15 20:50:15 Joshua kernel:  [<ffffffff8106a940>] ? process_one_work+0x3b0/0x3b0
Feb 15 20:50:15 Joshua kernel:  [<ffffffff8106fd08>] ? kthread+0xb8/0xd0
Feb 15 20:50:15 Joshua kernel:  [<ffffffff8106fc50>] ? kthread_worker_fn+0x170/0x170
Feb 15 20:50:16 Joshua kernel:  [<ffffffff814770cf>] ? ret_from_fork+0x3f/0x70
Feb 15 20:50:16 Joshua kernel:  [<ffffffff8106fc50>] ? kthread_worker_fn+0x170/0x170

I don't know when it got fixed but I have not seen it for a while.

I can turn off the monitor and have it restart the display fine. 

Same with restarting the xorg display from blanked screensaver.

(In reply to Sanjeev Kumar Sharma from comment #15)
> I don't know when it got fixed but I have not seen it for a while.

Resolving as fixed, thanks for the followup. Feel free to reopen this report if it happens again.

Bernd/Daniel, if you're still seeing issues, please file your own reports.