Created attachment 115096 [details] xrandr --verbose I have a ThinkPad T550 with Philips Brilliance 288P monitor, connected to mini displayport. I'm trying to use 3840x2160 mode at 60hz refresh, which should work through MST. But i'm not getting any DP1-1 DP1-2 devices for MST virtual display panels. I'm running archlinux with 4.0.0-rc7-drm_intel_nightly_20150411 kernel (i've tried 3.19.3-3-ARCH, and 4.0.0-1 mainline kernel) xf86-video-intel-git 2.99.917.265.g37ce4dd-1 libdrm-git 2.4.60.31.g6f90b77-1 I'm attaching xrandr --verbose output and dmesg with drm.debug after doing xrandr --output DP1 --off to motivate monitor to switch to MST mode. DisplayPort 1.2 in monitor properties is enabled.
Created attachment 115097 [details] dmesg output after xrandr --output DP1 --off
Hmm, I thought it was a composite panel. What does the drm.debug dmesg say when you try to enable the 60Hz 4k mode?
Created attachment 115098 [details] xrandr --output DP1 --mode 3840x2160 --rate 60 This is logged in dmesg when I try to use 60hz. The whole system can hang occasionally at this point (but not all the time, one in ten attempts)
My earlier statement was completely wrong, it definitely is acting as a single 4k@60Hz stream. The bandwidth is right at the edge of the available clocks, and it would seem we are not quite generous enough with our fifo allocation.
Any chance this will be fixed anytime soon? I'm willing to run any tests necessary.
Does i915.enable_ips=0 module parameter help?
On second thoughts, that may be a bogus suggestion, but no harm in trying...
(In reply to Jani Nikula from comment #6) > Does i915.enable_ips=0 module parameter help? Thanks for the suggestion, but I'm actually using this parameter right now, without it the monitor won't even work properly. Should have mentioned it on the bug report..
Seems to be fixed now, in intel nightly kernel. Linux 4.1.0-rc6-drm_intel_nightly_20150605 DP1 connected primary 3840x2160+2880+0 (normal left inverted right x axis y axis) 621mm x 341mm 3840x2160 60.00*+ 30.00
The monitor works, but it still requires i915.enable_ips=0 flag, and a GPU hang occurs occasionally (and was unable to catch the dump). Should this bug remain open? Or should I file a new one when I get the dump from hanged GPU?
(In reply to Marat Bakeev from comment #10) > The monitor works, but it still requires i915.enable_ips=0 flag, and a GPU > hang occurs occasionally (and was unable to catch the dump). > > Should this bug remain open? Or should I file a new one when I get the dump > from hanged GPU? The following two patches were merged to drm-intel-nightly after that above comment was made. Could you check if you can remove the i915.enable_ips=0 flag with those? commit 066cf55b9ce35f1f90dde9fcec01431a9243a949 Author: Rodrigo Vivi <rodrigo.vivi@intel.com> Date: Fri Jun 26 13:55:54 2015 -0700 drm/i915: Fix IPS related flicker commit 8cfb340774744438dea08a32072bea4a162dd132 Author: Ville Syrjälä <ville.syrjala@linux.intel.com> Date: Wed Jun 3 15:45:11 2015 +0300 drm/i915: Don't enable IPS when pixel rate exceeds 95%
(In reply to Ander Conselvan de Oliveira from comment #11) > The following two patches were merged to drm-intel-nightly after that above > comment was made. Could you check if you can remove the i915.enable_ips=0 > flag with those? > With linux-drm-intel-nightly 20150810 (Linux 4.2.0-1-drm-intel-nightly) it is possible to boot and use the monitor without the i915.enable_ips=0 flag. But the monitor now defaults to 30 hz refresh rate, although it is still possible to switch it to 60hz through xrandr (but it's a small annoyance).
With linux-drm-intel-nightly-docs 20150810 the kernel log is filled with messages: [drm:gen8_irq_handler [i915]] *ERROR* The master control interrupt lied (SDE)!
(In reply to Marat Bakeev from comment #13) > With linux-drm-intel-nightly-docs 20150810 the kernel log is filled with > messages: [drm:gen8_irq_handler [i915]] *ERROR* The master control interrupt > lied (SDE)! That's bug 92084. (In reply to Marat Bakeev from comment #12) > With linux-drm-intel-nightly 20150810 (Linux 4.2.0-1-drm-intel-nightly) it > is possible to boot and use the monitor without the i915.enable_ips=0 flag. This part is fixed. > But the monitor now defaults to 30 hz refresh rate, although it is still > possible to switch it to 60hz through xrandr (but it's a small annoyance). Is this the only remaining problem? Please file a new bug about that to not conflate this one. Thanks.
Bug scrub: Marat Bakeev, do you confirm the bug is fixed?
Yes, the issue I reported is definitely fixed now. Thanks!
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.