SIGABRT in copy_image when ArtCNN GLSL user shader engages (video.c:1419 assertion) on 0.41.0 / NVIDIA / OpenGL
Maintainers usually reply within 1 day
Nobody has claimed this yet.
Assessment
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Newbie friendliness
- 48/100
- Issue type
- Bug
- Clarity
- Mostly clear
- Activity status
- Active
- Tech stack
- c
- Domain
- computer-graphics
Research direction
Start at video/out/gpu/video.c:1419 and reproduce with the provided mpv command using ArtCNN_C4F32_DS.glsl, focusing on the copy_image assertion when the //!WHEN guard engages. Compare the failing ArtCNN pass structure with the clean CuNNy-4x16-DS case; done means the shader engages without an abort under the reported scaling conditions.
Written by the indexing model from the issue text.
Description
Summary
mpv 0.41.0 aborts the moment the ArtCNN_C4F32_DS user shader engages (output ≥1.3× the video size, i.e. when its //!WHEN guard fires):
video/out/gpu/video.c:1419: copy_image: Assertion '*offset + count < sizeof(dst)' failed.
The shader compiles cleanly — this is a runtime abort, not a GLSL error. Windowed at 1.0× (guard not firing) playback idles cleanly; the abort happens exactly on engagement (2× window scale and on fullscreen 4K both reproduce).
Repro
mpv --no-config --vo=gpu --gpu-api=opengl --glsl-shaders=ArtCNN_C4F32_DS.glsl --force-window=yes --window-scale=2.0 <any 640x360 test clip>
ArtCNN_C4F32_DS.glsl from https://github.com/Artoriuz/ArtCNN @ 7d69551 (with the #extension line hoisted to line 1 for NVIDIA compat — Artoriuz/ArtCNN#48).
Environment
- mpv 0.41.0 (vo=gpu AND vo=gpu-next),
--gpu-api=opengl - NVIDIA 595.84 and 595.91.07, RTX 3070
- Ubuntu (kernel 6.x), GNOME/mutter wayland + XWayland
- The abort reproduces identically headless (Xvfb, llvmpipe) — so it is not driver-specific rendering, it is the copy_image path's bounds assumption on the doubled texture
Notes
- The sibling doubler CuNNy-4x16-DS engages and plays cleanly under identical conditions, so this is specific to ArtCNN's pass structure (it uses
//!SAVEtextures at 2× width / 4× height ratios). - A backtrace from the core shows the abort in copy_image during the second render pass after the WHEN guard engages.
Happy to provide the full log, the core backtrace, or test a patch.
- Dominant language
- C
- Stars
- 38.2k
- Forks
- 3.7k
- Avg merge
- 1d 13h
- Merged PRs (30d)
- 76
Getting set up
- No Dockerfile or Docker Compose file
- Has a pull request template
- No contributing guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from mpv-player/mpv
-
meta:feature-request
Difficulty 2/5 1-3 hours Newbie friendliness 66/100
mpv-player/mpv#18855 · 2 comments ·
Maintainers usually reply within 1 day
-
meta:feature-request
Difficulty 2/5 1-3 hours Newbie friendliness 70/100
mpv-player/mpv#18683 · 4 comments ·
Maintainers usually reply within 1 day
-
Subtitle delay value displays "-0ms"May be free again A pull request for this issue was closed without being merged. Openos:linux
Difficulty 2/5 1-3 hours Newbie friendliness 66/100
mpv-player/mpv#18649 · 1 comment ·
Maintainers usually reply within 1 day
-
meta:feature-request
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
mpv-player/mpv#18464 ·
Maintainers usually reply within 1 day
-
Keyboard shortcut next and previous secondary subtitle filePossibly taken @ravarnax claimed this 225 days ago. Openmeta:feature-request
Difficulty 2/5 1-3 hours Newbie friendliness 64/100
mpv-player/mpv#17693 · 5 comments · 1 reaction ·
Maintainers usually reply within 1 day
Similar issues
-
`sha256()` returns a wrong hash when libsql-ffi is built with gcc 11 or laterPossibly taken @colinhacks claimed this today. Open
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
tursodatabase/libsql#2333 ·
Maintainers usually reply within 11 days
-
Difficulty 2/5 1-3 hours Newbie friendliness 76/100
-
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
OpenPrinting/cups#1826 ·
Maintainers usually reply within 1 day
-
Backport to 3.0 advertises a suffix that does not existPossibly taken A pull request linked to this issue is open or already merged. Open
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
389ds/389-ds-base#8152 ·
Maintainers usually reply within 2 days
-
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
RetroAchievements/rcheevos#580 · 1 comment ·
Maintainers usually reply within 5 days