dri-devel.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: bugzilla-daemon@freedesktop.org
To: dri-devel@lists.freedesktop.org
Subject: [Bug 100303] Adding a single, meaningless if-else to a shader source leads to different image
Date: Tue, 21 Mar 2017 15:07:49 +0000	[thread overview]
Message-ID: <bug-100303-502@http.bugs.freedesktop.org/> (raw)


[-- Attachment #1.1: Type: text/plain, Size: 2748 bytes --]

https://bugs.freedesktop.org/show_bug.cgi?id=100303

            Bug ID: 100303
           Summary: Adding a single, meaningless if-else to a shader
                    source leads to different image
           Product: Mesa
           Version: git
          Hardware: x86-64 (AMD64)
                OS: Linux (All)
            Status: NEW
          Severity: normal
          Priority: medium
         Component: Drivers/Gallium/radeonsi
          Assignee: dri-devel@lists.freedesktop.org
          Reporter: h.evrard@imperial.ac.uk
        QA Contact: dri-devel@lists.freedesktop.org

Created attachment 130352
  --> https://bugs.freedesktop.org/attachment.cgi?id=130352&action=edit
Zip archive with files to reproduce

See: https://github.com/mc-imperial/shader-compiler-bugs/issues/62

Steps to Reproduce:
-------------------------------

0. This bug seems to affect AMD GPU (tested on Tonga, see below for
   more config details)

1. Obtain and build the latest release of get-image, a simple tool that
   creates a .png image from a fragment shader
   https://github.com/mc-imperial/get-image

2. From a terminal, execute:

   /path/to/get_image/linux_out/install/bin/get_image original.frag

   This will create output.png, which should look like original.png

3. From a terminal, execute:

   /path/to/get_image/linux_out/install/bin/get_image variant.frag

   This will create output.png, which should look like variant.png

Expected Results:
-------------------------------

Both shaders should render an image that looks like original.png. This
is because the only difference between the shader source files is the
addition of an if-else whose if block is empty and else block contains
the original code. We control the condition value to make sure it is
always false at runtime.

Diff:
```
27a28,29
> uniform vec2 injectionSwitch;
>
117a120,124
>     if(injectionSwitch.x > injectionSwitch.y)
>         {
>         }
>     else
>         {
129a137
>         }
```

The injectionSwitch uniform is always set to { 0.0, 1.0 }. The else
block contains the original code, untouched.

Actual Results:
-------------------------------
The shaders render different images, which they should not: the shadow
in the lower part of the ball disappears in the variant.

Additional Information:
-------------------------------
Some info on the system where the bug was found:

- CPU: AMD A10-7850K Radeon R7, 12 Compute Cores 4C+8G
- GPU: Advanced Micro Devices, Inc. [AMD/ATI] Tonga PRO [Radeon R9 285/380]
- Mesa: 17.1.0-devel (git-08df015)
- OS: Ubuntu 16.04, Linux 4.4.0-67-generic

-- 
You are receiving this mail because:
You are the assignee for the bug.

[-- Attachment #1.2: Type: text/html, Size: 4536 bytes --]

[-- Attachment #2: Type: text/plain, Size: 160 bytes --]

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

             reply	other threads:[~2017-03-21 15:07 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-03-21 15:07 bugzilla-daemon [this message]
2017-03-22  1:04 ` [Bug 100303] Adding a single, meaningless if-else to a shader source leads to different image bugzilla-daemon

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=bug-100303-502@http.bugs.freedesktop.org/ \
    --to=bugzilla-daemon@freedesktop.org \
    --cc=dri-devel@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).