Go to file
Georges Basile Stavracas Neto ada5e67f7e cogl/matrix: Calculate inverse using graphene matrices
Turns out inverting a matrix was the largest chunk of the CoglMatrix
code. By switching to Graphene, a lot of it can go away. The inverse
is still cached in the CoglMatrix struct itself, to preserve the
optimization.

However, switching to graphene_matrix_t to calculate the inverse has
a challenge: float precision. We had to work around it here, and it
needs an explanation.

The way to detect whether a matrix is invertible or not (i.e.
whether it's not a "singular" matrix, or not) is by checking
if the determinant equals 0. So far, so good.

Both graphene_matrix_t and CoglMatrix use single-precision
floats to store their 4x4 matrices. Graphene uses vectorized
operations to optimize determinant calculation, while Cogl
tries to keep track of the matrix type and has special-purpose
determinant functions for different matrix types (the most
common one being a 3D matrix).

Cogl, however, has a fundamentally flawed check for whether
the matrix is invertible or not. Have a look:

```
float det;

…

if (det*det < 1e-25)
   return FALSE;
```

Notice that 1e-25 is *way* smaller than FLT_EPSILON. This
check is fundamentally flawed.

"In practice, what does it break?", the reader might ask.
Well, in this case, the answer is opposite of that: Cogl
inverts matrices that should not be invertible. Let's see
an example: the model-view-projection of a 4K monitor. It
looks like this:

```
| +0,002693 +0,000000 +0,000000 +0,000000 |
| +0,000000 -0,002693 +0,000000 +0,000000 |
| +0,000000 +0,000000 +0,002693 +0,000000 |
| -5,169809 +2,908017 -5,036834 +1,000000 |
```

The determinant of this matrix is -0.000000019530306557.
It evidently is smaller than FLT_EPSILON. In this situation,
Cogl would happily calculate the inverse matrix, whereas
Graphene (correctly) bails out and thinks it's a singular
matrix.

This commit works around that by exploiting the maths around
it. The basis of it is:

  inverse(scalar * M) = (1/scalar) * M'

which can be extrapolated to:

  inverse(M) = scalar * inverse(scalar * M) = M'

In other words, scaling the to-be-inversed matrix, then
scaling the inverse matrix by the same factor, gives us
the desired inverse. In this commit, the scale is calculated
as 1 / (smallest value in the diagonal).

I'm sorry for everyone that has to read through this :(

https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1439
2020-10-06 15:34:47 +00:00
.gitlab/issue_templates gitlab: Add missing < in markdown comment tag 2020-02-14 03:10:28 +00:00
.gitlab-ci ci: Add gnome-shell runtime dependencies 2020-08-07 16:49:59 +02:00
clutter Don't access CoglMatrix struct fields 2020-10-06 15:34:46 +00:00
cogl cogl/matrix: Calculate inverse using graphene matrices 2020-10-06 15:34:47 +00:00
data backends: force enable modifiers on tagged devices 2020-09-28 14:00:44 +02:00
doc Add MetaGravity and replace X11 equivalent with it 2020-02-29 21:01:50 +00:00
meson build: Add postinstall script 2019-08-27 09:57:54 +00:00
po Update Catalan translation 2020-10-03 23:14:34 +02:00
src Don't access CoglMatrix struct fields 2020-10-06 15:34:46 +00:00
subprojects build: bump ABI to sysprof-capture-4 2020-07-28 11:13:30 -07:00
tools tools: Remove obsolete ppa-magic.py 2018-11-30 11:12:12 +08:00
.gitignore build: bump ABI to sysprof-capture-4 2020-07-28 11:13:30 -07:00
.gitlab-ci.yml ci: Explicitly specify job dependencies 2020-08-04 18:00:04 +02:00
config.h.meson *: Fix spelling mistakes found by codespell 2020-08-29 09:10:31 +00:00
COPYING Updated obsolete FSF postal address in COPYING 2014-01-12 08:44:30 +07:00
meson_options.txt *: Fix spelling mistakes found by codespell 2020-08-29 09:10:31 +00:00
meson.build build: Bump graphene requirement to 1.10.2 2020-10-06 15:34:45 +00:00
mutter.doap doap: Update list of maintainers 2019-10-16 12:33:47 +02:00
NEWS Bump version to 3.38.1 2020-10-05 19:55:19 +02:00
README.md README: Add contribution section 2019-02-14 15:38:46 +01:00

Mutter

Mutter is a Wayland display server and X11 window manager and compositor library.

When used as a Wayland display server, it runs on top of KMS and libinput. It implements the compositor side of the Wayland core protocol as well as various protocol extensions. It also has functionality related to running X11 applications using Xwayland.

When used on top of Xorg it acts as a X11 window manager and compositing manager.

It contains functionality related to, among other things, window management, window compositing, focus tracking, workspace management, keybindings and monitor configuration.

Internally it uses a fork of Cogl, a hardware acceleration abstraction library used to simplify usage of OpenGL pipelines, as well as a fork af Clutter, a scene graph and user interface toolkit.

Mutter is used by, for example, GNOME Shell, the GNOME core user interface, and by Gala, elementary OS's window manager. It can also be run standalone, using the command "mutter", but just running plain mutter is only intended for debugging purposes.

Contributing

To contribute, open merge requests at https://gitlab.gnome.org/GNOME/mutter.

The coding style used is primarily the GNU flavor of the GNOME coding style with some minor additions such as preferring stdint.h types over GLib fundamental types, and a soft 80 character line limit. However, in general, look at the file you're editing for inspiration.

Commit messages should follow the GNOME commit message guidelines. We require an URL to either an issue or a merge request in each commit.

License

Mutter is distributed under the terms of the GNU General Public License, version 2 or later. See the COPYING file for detalis.