Skip to content
GitLab
Projects Groups Snippets
  • /
  • Help
    • Help
    • Support
    • Community forum
    • Submit feedback
    • Contribute to GitLab
  • Register
  • Sign in
  • mesa mesa
  • Project information
    • Project information
    • Activity
    • Labels
    • Members
  • Repository
    • Repository
    • Files
    • Commits
    • Branches
    • Tags
    • Contributors
    • Graph
    • Compare
  • Issues 3,095
    • Issues 3,095
    • List
    • Boards
    • Service Desk
    • Milestones
  • Merge requests 1,003
    • Merge requests 1,003
  • CI/CD
    • CI/CD
    • Pipelines
    • Jobs
    • Schedules
  • Deployments
    • Deployments
    • Releases
  • Packages and registries
    • Packages and registries
    • Container Registry
  • Analytics
    • Analytics
    • Value stream
    • CI/CD
    • Repository
  • Activity
  • Graph
  • Create a new issue
  • Jobs
  • Commits
  • Issue Boards
Collapse sidebar

Due to an influx of spam, we have had to impose restrictions on new accounts. Please see this wiki page for instructions on how to get full permissions. Sorry for the inconvenience.

  • MesaMesa
  • mesamesa
  • Issues
  • #3090
Closed
Open
Issue created Jun 05, 2020 by Lukas F. Hartmann@mntmn

etnaviv/GC7000L: discard does not work correctly with depth buffer

System information

  • OS: Debian GNU/Linux bullseye/sid
  • GPU: GC7000L in NXP i.MX8MQ
  • Kernel version: Linux reform 5.7.0-rc6+ #9 (closed) SMP PREEMPT Fri Jun 5 01:09:27 CEST 2020 aarch64 GNU/Linux
  • Mesa version: 2.1 Mesa 20.2.0-devel (git-c448f42e178)
  • Xserver version (if applicable): 1.20.99.1
  • Desktop manager and compositor: sway version 1.4-8386655

Description

When discard is used in a shader, the fragment that is "behind" the current fragment should be resolved according to correct depth information. It looks like, on GC7000L, discarded pixels either overwrite the depth buffer anyway, or early-Z is not switched off correctly.

Here is a simple test case that I wrote to illustrate the problem:

https://github.com/mntmn/etna-test2

Here is a new, prettier test case for wayland by rasterman+daniels:

earlyz-bug-xdg-shell.tar.gz

This bug causes accelerated rendering in KiCAD to fail and makes some games like Minetest (a sophisticated open source Minecraft clone) unplayable. I attached some screenshot of the problems below.

Bounty

There was originally a bounty of EUR 1000.00 on this bug, but I happened to find the solution myself, so there is no more bounty (see comments).

Screenshots

These are the outputs from the new test case.

Test case (wrong output with etnaviv):

zbug-1-etnaviv zbug-2-etnaviv

Test case (correct output with radeon):

zbug-3-radeon zbug-4-radeon

Edited Jun 09, 2020 by Lukas F. Hartmann
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information
Assignee
Assign to
Time tracking