KMS capture always streams the portrait monitor even though the correct connector is selected

未关闭
  • 哪里不对？
#6,002 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
4/5
预计耗时
1-2 天
新手友好度
45/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
cpp, linux
领域
backend, operating-systems

调研方向

Start by tracing the KMS capture path that uses the selected connector and cursor plane, then compare which CRTC supplies the captured image. The report identifies connectors 80 and 88, with the same portrait monitor appearing for both; use the provided output names and logs to reproduce. Done when selecting output_name 1 streams the ultrawide on DP-2 while output_name 0 still streams the portrait monitor.

由索引模型根据 Issue 内容生成。

描述

Is there an existing issue for this?
  • I have searched the existing issues
Is your issue described in the documentation?
  • I have read the documentation
Is your issue present in the latest beta/pre-release?

I'm too lazy to test

Describe the Bug

With capture = kms and two monitors on one AMD GPU, the stream always shows the same monitor (the portrait one, as displayed) regardless of output_name. The debug log shows Sunshine choosing the correct, different connector for each output_name value, so connector selection appears correct but the captured image does not come from the chosen connector's CRTC. capture = x11 streams the correct monitor.

Possibly related: #3642 (KMS streams wrong monitor, X11 correct).

Layout (X11, Cinnamon):

X output Kernel connector Connector ID Mode Rotation X CRTC
DisplayPort-1 (primary) card2-DP-2 80 3440x1440+1200+159 normal 0
DisplayPort-2 card2-DP-3 88 1200x1920+0+0 left 1

KMS display list order from the debug log: DP-3 DP-2.

output_name Log: connector / cursor plane Streamed image
0 88 / 64 portrait monitor (as displayed)
1 80 / 61 portrait monitor (as displayed); expected: ultrawide

The machine also has an Intel UHD 630 iGPU (card1, i915) with nothing connected; all monitors are on the RX 580 (card2).

Expected Behavior

output_name = 1 (connector 80, DP-2) should stream the 3440x1440 ultrawide.

Additional Context

The streamed image is always the content of the rotated portrait monitor (CRTC 1), whichever connector is selected. The rotation may be relevant, but I haven't tested without it. Happy to run more diagnostics (e.g. modetest -p output) if useful.

Config (~/.config/sunshine/sunshine.conf):

gamepad = xone
capture = kms
output_name = 1
sunshine_name = Magrathea
min_log_level = debug

Host Operating System

Linux

Operating System Version

Linux Mint 22 (Ubuntu 24.04 base), Cinnamon on X11

Architecture

amd64/x86_64

Package

Linux - deb

GPU Type

AMD

GPU Model

AMD Radeon RX 580 (Polaris 10)

GPU Driver/Mesa Version

Mesa 25.2.8-0ubuntu0.24.04.2 (radeonsi, polaris10, ACO, DRM 3.64), kernel 7.0.0-38-generic, amdgpu with Display Core enabled

Capture Method

KMS (Linux)

Apps
Log output
Info: Detected display: DisplayPort-1 (id: 0)DisplayPort-1 connected: true
Info: Detected display: DisplayPort-2 (id: 1)DisplayPort-2 connected: true
Debug: Final KMS display_names return list: DP-3 DP-2
Info: Screencasting with KMS
Info: /dev/dri/card2 -> amdgpu
Info: Found monitor for DRM screencasting
Info: Found connector ID [80]
Info: Found cursor plane [61]

With output_name = 0 the same lines show "Found connector ID [88]" and "Found cursor plane [64]", but the streamed image is identical.
Online logs

No response

主要语言
C++
星标
43k
派生
2.4k
平均合并
1 天 13 小时
30 天内合并 PR
111

环境准备

  • 没有 Dockerfile 或 Docker Compose 文件
  • 没有 Pull Request 模板
  • 阅读贡献指南

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

LizardByte/Sunshine 的其他 Issue

查看 LizardByte/Sunshine 的全部 Issue

相似的 Issue

更多 C++ Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。