Skip to contentSkip to conversation

topic/pipe-reorder for drm-intel-next and possibly v7.1-rc1

GIT PULL Merged Jani Nikula 3 participants 5 messages Apr 15 – Apr 22

lore
Share & tools
lore
#0 lore raw

Hi Dave & Linus -

This is a topic pull request for changing the pipe allocation order for discrete platforms from the usual A,B,C,D to A,C,B,D. The goal is to help pipe joiner configurations that reserve the adjacent pipe as the secondary pipe without the user space knowing. More details in the relevant commit message. The CRTC iteration is also changed to remain in pipe order.

This has been a long time in the making, primarily because IGT basically assumed CRTC index == pipe, and it took hundreds of commits from Ville and me over several months to untangle it, and actually get CI results.

I'm going to merge this to drm-intel-next soon, and at this point in time it would normally target v7.2.

We're fully aware this is beyond feature deadlines for v7.1 and it hasn't seen linux-next. But in the off-chance you might consider also pulling this as a fix for v7.1 (and really, distros might want this to be backported to stable) I'm sending this as a separate topic pull request.

I'm leaving it to your discretion.

BR,
Jani.

topic/pipe-reorder-2026-04-15:
drm/i915/display: change pipe allocation order for discrete platforms

The following changes since commit 4516432284e1b2ad9e70de8067f779c9c1072189:

ttm/pool: track allocated_pages per numa node. (2026-04-08 06:52:48 +1000)

are available in the Git repository at:

https://gitlab.freedesktop.org/drm/i915/kernel.git tags/topic/pipe-reorder-2026-04-15

for you to fetch changes up to 2c683e9b419328da3433a49f7c467da71aaf0469:

drm/i915/display: change pipe allocation order for discrete platforms (2026-04-15 13:30:26 +0300)

----------------------------------------------------------------
drm/i915/display: change pipe allocation order for discrete platforms

----------------------------------------------------------------
Jani Nikula (1):
drm/i915/display: change pipe allocation order for discrete platforms

Ville Syrjälä (2):
drm/i915/joiner: Make joiner "nomodeset" state copy independent of pipe order drm/i915: Walk crtcs in pipe order

 drivers/gpu/drm/i915/display/intel_crtc.c          | 49 +++++++++++-
 drivers/gpu/drm/i915/display/intel_display.c       | 20 ++---
 drivers/gpu/drm/i915/display/intel_display.h       | 90 +++++++++-------------
 drivers/gpu/drm/i915/display/intel_display_core.h  |  3 +
 .../gpu/drm/i915/display/intel_display_driver.c    |  1 +
 drivers/gpu/drm/i915/display/intel_display_types.h |  1 +
 drivers/gpu/drm/xe/display/xe_display.c            |  1 +
 7 files changed, 95 insertions(+), 70 deletions(-)
Signature · 1 line
Jani Nikula, Intel
#1 lore raw
Quoted 6 lines · Jani Nikula

We're fully aware this is beyond feature deadlines for v7.1 and it hasn't seen linux-next. But in the off-chance you might consider also pulling this as a fix for v7.1 (and really, distros might want this to be backported to stable) I'm sending this as a separate topic pull request.

Does this basically obviate the hack that I've been running for several months now for my dual 6k setup?

I guess I'll test it just to verify, but if this fixes that known issue I'm all for getting this fixed sooner rather than later.

Linus

#2 lore raw
Quoted 3 lines · Linus Torvalds

I guess I'll test it just to verify, but if this fixes that known issue I'm all for getting this fixed sooner rather than later.

Well, that was easy enough. I assume this was expected, but I can indeed verify that that branch works for me as a replacement for (two versions) of the local patch I have been running for the last six months or so.

And maybe I'm the only person on the planet that runs this setup, but it gets an ack from me. So I'm ok with a late pull as a fix, but somebody who can better judge the risks for other setups should eventually make the judgement call.

Linus

#3 lore raw
Quoted 10 lines · Linus Torvalds
Quoted 3 lines · Linus Torvalds

I guess I'll test it just to verify, but if this fixes that known issue I'm all for getting this fixed sooner rather than later.

Well, that was easy enough. I assume this was expected, but I can indeed verify that that branch works for me as a replacement for (two versions) of the local patch I have been running for the last six months or so.

Yay, thanks for confirming it still works for you!

Quoted 4 lines

And maybe I'm the only person on the planet that runs this setup, but it gets an ack from me. So I'm ok with a late pull as a fix, but somebody who can better judge the risks for other setups should eventually make the judgement call.

We've tested a similar setup locally, and a bunch of other setups in CI, and haven't seen any issues. As far as real world scenarios go, I think yours is as real as it gets. That said, we'll only really know once this hits more end users... but I'm not sure if there's any benefit in dragging this across another development cycle.

Dave, up to you then, will you take this through drm-next?

BR,
Jani.

Signature · 1 line
Jani Nikula, Intel
merge bot Merged
#4 lore raw
In reply to Jani Nikula
The pull request you sent on Wed, 15 Apr 2026 19:08:52 +0300:

The pull request you sent on Wed, 15 Apr 2026 19:08:52 +0300:

Quoted 1 line

https://gitlab.freedesktop.org/drm/i915/kernel.git tags/topic/pipe-reorder-2026-04-15

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/993eb191400f5d06322989ea7c890fd11965f80b

Thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/prtracker.html