test0r
6c50354afb
User has been running the more aggressive 0001 variant (skip the watchDmaBuf wait entirely) downstream and reports Plasma feels measurably snappier — fewer latency spikes under heavy compositor activity. The present 0002 MR has different (correct) wait semantics so the perceived gain can't be directly attributed, but calling it out gives reviewers an honest signal that the patch at least preserves whatever benefit was on the table downstream. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
upstream-submissions/
Drafts of the upstream submission texts for the patch series in
arch/ and kernel/. Each subdirectory pairs a patch series with
the issue / merge-request / mailing-list body that should accompany
it when going upstream.
Layout
upstream-submissions/
├── qt6-base-fourier/ # bugreports.qt.io issue + Gerrit notes
│ └── qt-bug-report.md
├── kwin-fourier/ # invent.kde.org MR body + KDE flow notes
│ └── kde-mr-body.md
└── vb2-dma-resv/ # linux-media cover letter + send-email recipe
└── lkml-submission-notes.md
The patches themselves live in:
arch/qt6-base-fourier/0001..0003-*.patch(Qt 6 patches)arch/kwin-fourier/0002-*.patch(KWin upstream-shape patch;0001-*-bypass-*.patchis the local diagnostic, not for upstream)kernel/vb2-dma-resv-rfc/0001..0003-*.patchplus0000-cover-letter.patch(kernel patches)
Submission status
| Series | Bug filed | Patches submitted | Landed |
|---|---|---|---|
| qt6-base-fourier | not yet | not yet | not yet |
| kwin-fourier (0002) | n/a — KDE goes straight to MR | not yet | not yet |
| vb2-dma-resv | n/a — kernel goes straight to mailing list | not yet | not yet |
Validation gates before submitting
Each subdirectory's notes file lists the validation gates specific to that series. Common to all three:
- Patches must apply cleanly to the latest upstream branch (rebase before sending).
- Build must complete (compile-test for kernel; the others are already validated as building cleanly).
- End-to-end runtime test on ohm should still pass with the patch applied (preferably with the other layers of the campaign also applied, so the patch doesn't get blamed for behavior in another layer).
Submission order
Recommended:
-
vb2-dma-resv to linux-media first. The kernel-side fix is the architectural correction that makes the kwin patch's premise (the wait will signal correctly) actually true. If the kernel patch isn't merged or accepted, the kwin patch submission can still proceed but on weaker rationale.
-
qt6-base-fourier to bugreports.qt.io + Gerrit second. Independent of the others; pure spec-correctness improvement for OpenGL ES 3.x, can land at any time.
-
kwin-fourier (0002) to invent.kde.org last, ideally referencing the merged or in-flight kernel patch as the underlying correctness fix.
Cadence after submitting
- Watch each upstream for review feedback; revise and resubmit.
- After landing, request backports to LTS branches where applicable (Qt 6.5 LTS, Plasma 6.x, kernel stable).
- Once landed in a release shipping to Arch / Debian / etc., the
corresponding
arch/*-fourier/package becomes redundant and can be retired (drop theepoch=1override, archive the PKGBUILD as historical reference).