Summary: | [Regression]Multiple monitors via DP TB dock detection/resolution/layout failure after Skylake->Coffeelake upgrade | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | DRI | Reporter: | Arcadiy Ivanov <arcadiy> | ||||||
Component: | DRM/Intel | Assignee: | Intel GFX Bugs mailing list <intel-gfx-bugs> | ||||||
Status: | RESOLVED MOVED | QA Contact: | Intel GFX Bugs mailing list <intel-gfx-bugs> | ||||||
Severity: | major | ||||||||
Priority: | high | CC: | intel-gfx-bugs | ||||||
Version: | unspecified | Keywords: | regression | ||||||
Hardware: | x86-64 (AMD64) | ||||||||
OS: | Linux (All) | ||||||||
Whiteboard: | Triaged | ||||||||
i915 platform: | CFL | i915 features: | display/DP, display/eDP | ||||||
Attachments: |
|
Description
Arcadiy Ivanov
2019-07-20 21:58:33 UTC
BIOS Information Vendor: Dell Inc. Version: 1.1.1 Release Date: 05/31/2019 System Information Manufacturer: Dell Inc. Product Name: Precision 5540 SKU Number: 0906 Family: Precision Base Board Information Manufacturer: Dell Inc. Product Name: 0V030K Version: A00 Fedora KDE 30 Kernel 5.1.18-300.fc30.x86_64 i915 KMS Driver 5.1.18-300.fc30.x86_64 i915 KMS Driver version 1.6.0 20190207 Reporter, can you please attach the coffeelake dmesg from boot with kernel parameters drm.debug=0x1e log_buf_len=4M? Have you verified the issue with drm-tip (https://cgit.freedesktop.org/drm-tip)? I will post the debug log and no, I haven't tried with drm-tip. Created attachment 144841 [details]
5.1.18 debug log
Please let me know if this is not enough and you actually need me to try drm-tip.
(In reply to Arcadiy Ivanov from comment #5) > I will post the debug log and no, I haven't tried with drm-tip. Can you verify with drmtip and give the feedback? This will be helpful during investigation. Created attachment 144857 [details] debug dmesg of drm-tip 2019y-07m-23d-12h-27m-47s UTC drm-tip fails in exactly the same way [ 4647.409] (II) intel(0): resizing framebuffer to 6000x3840 [ 4649.997] (II) intel(0): switch to mode 3840x2160@60.0 on DP1-2 using pipe 2, position (2160, 600), rotation normal, reflection none [ 4650.008] (EE) intel(0): failed to set mode: No space left on device [28] Please let me know if you need anything else. I think the main issues is that your 4x5.4Gbps DP link doesn't have enough bandwidth to support two 4k@60 displays: [ 81.717461] [drm:drm_dp_mst_atomic_check [drm_kms_helper]] [MST PORT:000000006465ad61] requires 47 vcpi slots [ 81.717472] [drm:drm_dp_mst_atomic_check [drm_kms_helper]] [MST PORT:00000000fb13674f] requires 47 vcpi slots [ 81.717483] [drm:drm_dp_mst_atomic_check [drm_kms_helper]] [MST PORT:00000000fb13674f] not enough VCPI slots in mst state 000000006830e28e (avail=16) [ 81.717516] [drm:drm_atomic_check_only [drm]] atomic driver check for 000000004efe656f failed: -28 There is just no way to make that work. The only explanation I have why the SKL might have worked is that its TB implementation uses multiple DP links for the TB dock whereas the CFL uses just one link for whatever reason. The TB hardware supposedly is different between the two platforms. Our TB guru suggested that you should perhaps try to upgrade the TB firmware on the machine and see if that helps. Thanks for looking at this. Both Dell Precision 7510 (SKL) and Dell Precision 5540 (CFL) have the same TB capabilities, TB3 ***HBR2***. It's true, with HBR2 they don't support 2x4k@60Hz(**), but the point of filing this bug was somewhat different - there was a change in behavior in monitor detection and resolution/refresh rate setting. SKL was supporting 2x4k with auto-detection, which, given hardware limitations, means that it was settling on 2x4k@30Hz mode all by itself. When an upgrade was made to CFL, this auto-detection stopped working properly - as you can see it tried to configure 2x4K@60Hz and failed, without fallback to 2x4K@30Hz. The change in behavior cannot be blamed on changes in Xorg, xrandr etc - the entire SSD was moved. Dock is similarly blameless - it was the same dock. The only change was laptop hardware. Given that monitors could be detected and EDID was read the expected fallback should've been onto the 2x4K@30Hz on CFL same as on SKL. It wasn't. All firmware is up-to-date. ===== ** there is a way with WD19TB for 2x4k@60Hz with DP pass-through, but this wasn't the setup for this bug (In reply to Arcadiy Ivanov from comment #11) > Thanks for looking at this. > > Both Dell Precision 7510 (SKL) and Dell Precision 5540 (CFL) have the same > TB capabilities, TB3 ***HBR2***. It's true, with HBR2 they don't support > 2x4k@60Hz(**), but the point of filing this bug was somewhat different - > there was a change in behavior in monitor detection and resolution/refresh > rate setting. > > SKL was supporting 2x4k with auto-detection, which, given hardware > limitations, means that it was settling on 2x4k@30Hz mode all by itself. > When an upgrade was made to CFL, this auto-detection stopped working > properly - as you can see it tried to configure 2x4K@60Hz and failed, > without fallback to 2x4K@30Hz. Are you sure SKL used 30Hz? If it did then that was being done by some userspace component. The kernel driver never does stuff like that (it just more or less blindly tries to do what userspace asks). -- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/drm/intel/issues/341. |
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.