[PR #29] [MERGED] Raise Wayland display max buffer size #33

New issue

Closed

opened 2026-10-15 12:47:43 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-15 12:47:43 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/games-on-whales/gst-wayland-display/pull/29
Author: @JBailes
Created: 5/11/2026
Status: ✅ Merged
Merged: 5/11/2026
Merged by: @ABeltramo

Base: stable ← Head: fix/wayland-max-buffer-size

📝 Commits (1)

09dba8a Raise Wayland display max buffer size

📊 Changes

3 files changed (+19 additions, -20 deletions)

View changed files

📝 Cargo.lock (+13 -12)
📝 wayland-display-core/Cargo.toml (+3 -2)
📝 wayland-display-core/src/comp/mod.rs (+3 -6)

📄 Description

Summary

Update Wayland crates and enable wayland-server with libwayland_1_23.
Set the compositor display default max buffer size to 10 MiB before state initialization.
Mirrors the relevant behavior from github.com/linuxserver/pixelflux@ffc2d4daef.

Verification

Fresh CTs were created on the Proxmox host for verification. CT120 was not used.

CT900 Intel RPL-S, host render node /dev/dri/renderD129: test_dmabuf passed.
CT900 Intel: cargo test --locked -p wayland-display-core --lib -- --skip test_dmabuf passed, 12 passed.
CT901 AMD Radeon RX 7900 XTX, host render node /dev/dri/renderD130: test_dmabuf passed.
CT902 NVIDIA GeForce RTX 5080, host render node /dev/dri/renderD128: installed matching NVIDIA 595.71.05 userspace-only from the NVIDIA runfile, mapped /dev/nvidia* and /dev/nvidia-caps, and verified nvidia-smi plus eglinfo -B report NVIDIA EGL on 595.71.05.
CT902 NVIDIA: non-dmabuf library suite passed with XDG_RUNTIME_DIR=/tmp/gwd-xdg, 12 passed.
CT900: cargo check --workspace --all-targets passed, and feature tree confirmed wayland-server/libwayland_1_23 enables wayland-backend/libwayland_server_1_23.

NVIDIA AB24 Linear note

utils::allocator::tests::test_dmabuf still fails on NVIDIA after the CT was converted into a valid NVIDIA EGL environment. The failure is:

DrmFourcc(AB24), Modifier: Linear
GL_INVALID_OPERATION ... EGLImage not supported
panic at wayland-display-core/src/utils/allocator/mod.rs:682, assertion failed: bind_result.is_ok()

This was also reproduced against a clean HEAD baseline before this change, using the same CT902 NVIDIA environment and GBM_BACKEND=nvidia-drm. Therefore the NVIDIA AB24 Linear import issue predates this PR and is not introduced by the Wayland max-buffer change.

CT900, CT901, and CT902 are left running for inspection.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/games-on-whales/gst-wayland-display/pull/29 **Author:** [@JBailes](https://github.com/JBailes) **Created:** 5/11/2026 **Status:** ✅ Merged **Merged:** 5/11/2026 **Merged by:** [@ABeltramo](https://github.com/ABeltramo) **Base:** `stable` ← **Head:** `fix/wayland-max-buffer-size` --- ### 📝 Commits (1) - [`09dba8a`](https://github.com/games-on-whales/gst-wayland-display/commit/09dba8a84f36b5363cb61ffece91220a0a92288e) Raise Wayland display max buffer size ### 📊 Changes **3 files changed** (+19 additions, -20 deletions) <details> <summary>View changed files</summary> 📝 `Cargo.lock` (+13 -12) 📝 `wayland-display-core/Cargo.toml` (+3 -2) 📝 `wayland-display-core/src/comp/mod.rs` (+3 -6) </details> ### 📄 Description ## Summary - Update Wayland crates and enable `wayland-server` with `libwayland_1_23`. - Set the compositor display default max buffer size to 10 MiB before state initialization. - Mirrors the relevant behavior from https://github.com/linuxserver/pixelflux/commit/ffc2d4daef02726857174fd3a6b158a53ac09c5c. ## Verification Fresh CTs were created on the Proxmox host for verification. CT120 was not used. - CT900 Intel RPL-S, host render node `/dev/dri/renderD129`: `test_dmabuf` passed. - CT900 Intel: `cargo test --locked -p wayland-display-core --lib -- --skip test_dmabuf` passed, 12 passed. - CT901 AMD Radeon RX 7900 XTX, host render node `/dev/dri/renderD130`: `test_dmabuf` passed. - CT902 NVIDIA GeForce RTX 5080, host render node `/dev/dri/renderD128`: installed matching NVIDIA 595.71.05 userspace-only from the NVIDIA runfile, mapped `/dev/nvidia*` and `/dev/nvidia-caps`, and verified `nvidia-smi` plus `eglinfo -B` report NVIDIA EGL on 595.71.05. - CT902 NVIDIA: non-dmabuf library suite passed with `XDG_RUNTIME_DIR=/tmp/gwd-xdg`, 12 passed. - CT900: `cargo check --workspace --all-targets` passed, and feature tree confirmed `wayland-server/libwayland_1_23` enables `wayland-backend/libwayland_server_1_23`. ## NVIDIA AB24 Linear note `utils::allocator::tests::test_dmabuf` still fails on NVIDIA after the CT was converted into a valid NVIDIA EGL environment. The failure is: - `DrmFourcc(AB24), Modifier: Linear` - `GL_INVALID_OPERATION ... EGLImage not supported` - panic at `wayland-display-core/src/utils/allocator/mod.rs:682`, `assertion failed: bind_result.is_ok()` This was also reproduced against a clean `HEAD` baseline before this change, using the same CT902 NVIDIA environment and `GBM_BACKEND=nvidia-drm`. Therefore the NVIDIA AB24 Linear import issue predates this PR and is not introduced by the Wayland max-buffer change. CT900, CT901, and CT902 are left running for inspection. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-15 12:47:43 +00:00