Discussion:
[Mesa-dev] [Bug 102891] [radv] glitches on rpcs3 emulator (green zones)
b***@freedesktop.org
2017-09-20 07:24:48 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

Bug ID: 102891
Summary: [radv] glitches on rpcs3 emulator (green zones)
Product: Mesa
Version: 17.2
Hardware: x86-64 (AMD64)
OS: Linux (All)
Status: NEW
Severity: normal
Priority: medium
Component: Drivers/Vulkan/radeon
Assignee: mesa-***@lists.freedesktop.org
Reporter: ***@gmail.com
QA Contact: mesa-***@lists.freedesktop.org

Created attachment 134353
--> https://bugs.freedesktop.org/attachment.cgi?id=134353&action=edit
Green zones with radv

On rpcs3 emulator, the game shovel knight[npub31682] renders differently on
opengl and vulkan. When there are zones with moving water, the radv driver
displays only a green zone. OpenGl is fine.
I already reported the bug here
https://github.com/RPCS3/rpcs3/issues/2880
but it seems to be a driver bug as other vulkan/os systems render correctly.
Tested on ArchLinux, R9 Fury, mesa stable 17.2.0
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-09-20 07:25:45 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #1 from ***@gmail.com ---
Created attachment 134354
--> https://bugs.freedesktop.org/attachment.cgi?id=134354&action=edit
Correct rendering with OpenGl on moving water
--
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
b***@freedesktop.org
2017-09-20 09:06:26 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #2 from Grazvydas Ignotas <***@gmail.com> ---
Can you make a renderdoc capture?
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-09-20 10:02:40 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #3 from ***@gmail.com ---
Hope that's fine (1st use of renderdoc here).
I captured the frame with vulkan.
With renderdoc, I'm seeing warning messages on the console
spirv/spirv_to_nir.c:2763 WARNING: Unsupported SPIR-V capability:
SpvCapabilityImageMSArray
So, incomplete driver is probably the culprit.

The renderdoc file is big. I uploaded it here:
https://www.dropbox.com/s/096jt15fzvtlvif/rpcs3_2017.09.20_09.52.38_frame3641.rdc?dl=0
--
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
b***@freedesktop.org
2017-09-20 12:17:04 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #4 from Ilia Mirkin <***@alum.mit.edu> ---
(In reply to jdruel from comment #3)
Post by b***@freedesktop.org
With renderdoc, I'm seeing warning messages on the console
SpvCapabilityImageMSArray
So, incomplete driver is probably the culprit.
This capability is only supposed to be used if shaderStorageImageMultisample is
set to true, which it is not for either radv or anv. It's an application bug to
be passing it in.
--
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
b***@freedesktop.org
2017-09-21 15:48:45 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #5 from Bas Nieuwenhuizen <***@basnieuwenhuizen.nl> ---
IIRC that SpvCapabilityImageMSArray warning is always there with renderdoc
though, so probably a red herring.
--
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
b***@freedesktop.org
2017-11-09 00:10:05 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #6 from Dave Airlie <***@freedesktop.org> ---
Did someone already try RADV_DEBUG=zerovram to see if it helps?

The trace replays badly on amdgpu-pro which suggests the bad stuff is in ram
before recording.
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-11-09 08:01:52 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #7 from Samuel Pitoiset <***@gmail.com> ---
Yeah, zerovram doesn't change anything.
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-11-10 07:14:55 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

Dave Airlie <***@freedesktop.org> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |NEEDINFO

--- Comment #8 from Dave Airlie <***@freedesktop.org> ---
jdruel can you try the env var on the actual game?

The problem is the damage is capture in the trace, so Samuel's check didn't
help.
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-11-10 22:29:51 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #9 from ***@gmail.com ---
Created attachment 135388
--> https://bugs.freedesktop.org/attachment.cgi?id=135388&action=edit
new qrenderdoc capture

I did an
export RADV_DEBUG=zerovram
before running qrenderdoc, but as I'm on mesa 17.2, I don't think this option
is activated.
Let me know if this capture is enough for you, else I'll see if I can switch to
mesa-git without too much pain (=breaking my system) (I'm on Arch).
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-11-14 10:08:39 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #11 from ***@gmail.com ---
Created attachment 135446
--> https://bugs.freedesktop.org/attachment.cgi?id=135446&action=edit
with zerovram on mesa-git

I installed mesa-git+vulkan-radeon-git+xf86-video-amdgpu-git
My new test was done like this:
export RADV_DEBUG=zerovram
qrenderdoc

Please tell me if this file is correct for you.
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-11-13 09:30:49 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #10 from Samuel Pitoiset <***@gmail.com> ---
yeah, zerovram is quite recent (mesa 17.3).
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-12-20 10:17:59 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #12 from ***@gmail.com ---
Created attachment 136313
--> https://bugs.freedesktop.org/attachment.cgi?id=136313&action=edit
Shovel Knight-Rpcs3-pb01

Tested on mesa 17.3 with
RADV_DEBUG=zerovram qrenderdoc
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-12-20 10:19:07 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

--- Comment #13 from ***@gmail.com ---
Created attachment 136314
--> https://bugs.freedesktop.org/attachment.cgi?id=136314&action=edit
Shovel Knight-Rpcs3-pb02

Tested on mesa 17.3 with
RADV_DEBUG=zerovram qrenderdoc
It's in another stage: same behaviour
--
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
b***@freedesktop.org
2017-12-20 10:19:30 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

***@gmail.com changed:

What |Removed |Added
----------------------------------------------------------------------------
Version|17.2 |17.3
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
b***@freedesktop.org
2017-12-27 16:18:20 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

***@gmail.com changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |***@gmail.com
--
You are receiving this mail because:
You are the assignee for the bug.
b***@freedesktop.org
2018-01-13 08:23:34 UTC
Permalink
https://bugs.freedesktop.org/show_bug.cgi?id=102891

***@gmail.com changed:

What |Removed |Added
----------------------------------------------------------------------------
Resolution|--- |FIXED
Status|NEEDINFO |RESOLVED

--- Comment #14 from ***@gmail.com ---
solved in 17.3.2
--
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
Loading...