Age | Commit message (Collapse) | Author |
|
Reported-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Link: https://gitlab.freedesktop.org/mesa/waffle/-/issues/78
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
|
|
Detected by clang -Wmissing-noreturn. I haven't added that warning in
meson.build here, because it is also triggered by third-party code in
third_party/threads/threads_posix.c.
Signed-off-by: Simon McVittie <smcv@collabora.com>
|
|
Wayland 1.20 introduces two new symbols wl_proxy_marshal_flags and
wl_proxy_marshal_array_flags, which need to be wrapped as well.
Closes: https://gitlab.freedesktop.org/mesa/waffle/-/issues/76
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
|
|
|
|
Add all author entries, otherwise meson will not flag a rebuild when
they are updated.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Chad has been MIA for a few years now and I (Emil) have been doing the
actual maintenance. Update the references to better indicate reality.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
For a while now, we've had https://waffle.freedesktop.org/ and somewhat
recently the original waffle-gl.org domain expired.
Update (nearly) all the existing references. Be that waffle-gl.org ones
or people.freedesktop.org/~chadversary/waffle. The latter redirect to
the former.
Nearly since some archive.waffle-gl.org artefacts are nowhere to be
found ... at a glance. We can tweak or drop those with future commits.
While we're here - httpS all the links :-)
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Our files roughly fall into the following groups:
- source code - c/c++/cocoa - 4 spaces indent
- meson - 2 spaces
- cmake - 2 spaces
- android build - mix of tabs and 4 spaces -- on their way out
- handwritten Docbook XML files (ouch) - 2 spaces
- various one-off files
Based on the above - let's add editorconfig file for meson and
"everything else". It's a decent start going forward.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
This allows automatic overrides of the waffle-1 dependency and wflifo
program, when used as a subproject/dependency fallback without having to
specify them in the consumer, as is current meson best practice.
For projects using meson > 0.54, they can then just write:
```meson
dependency('waffle-1')
```
and if they have the waffle-1 wrap installed then meson will
automatically configure the project, and use the built libwaffle-1,
without having to specify fallback parameters in the calling
meson.build. This is quite similar to the find_program_override, which
works for executables instead of dependencies.
|
|
Nothing needs to write to this.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
String constants aren't suitable for writing, so assign them to a
const char *, not a plain char *.
Detected by gcc -Wwrite-strings.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
From context, it seems this was meant to check gl_context_flags, not the
constant array context_flags.
Detected by clang -Wtautological-pointer-compare.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Previously, this would print whatever uninitialized value happened to
be on the stack.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Otherwise, we segfault trying to look up the platform from a window that
was never properly initialized.
Fixes: 6695ab42ae80 ("gbm: factor out gbm_window_{init,teardown}")
|
|
Seemingly having the comment inline breaks with the current templates.
Fixes: ca13f88 ("gitlab-ci: add clang-format stage")
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
Add a small clang-format.sh script, to manage the merge-base commit
against which we compare.
The magic in there, is copied from the ci-fairy - huge thanks to the
team behind it.
Notes:
- The clang-format-7 dependency tracking it busted across Debian
releases - we need to add git explicitly
- Don't block the build stage, on clang-format
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
Copy the clang-format config from virgl, tweaking it slightly - column
sizes and include files order.
It does highlight few mostly trivial changes through the codebase.
Albeit none of them is a deal breaker so, no point in a wholesale
"fixup". The follow-up CI stage will ensure, newly merged code is good.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
Through Mesa we use libdrm's drmDevice2 API to gather DRM device
details. Additionally the API is simpler and the code shorter.
Execution time is nearly identical and with a pending libdrm MR - it
will go down in half.
While here, remove the link-time dependency - using dlopen/dlsym ftw.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
This reverts commit 9a4f7c0e4b50f1053318cc76478da530b7b66299.
Remove this ugly workaround. Fixed versions of Mesa have been deployed
5+ years ago.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Conflicts:
src/waffle/gbm/wgbm_display.c
|
|
Flip the meson werror=true switch. It will help us catch extra issues
before they creep in.
Closes: https://gitlab.freedesktop.org/mesa/waffle/-/issues/73
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Enable warning level 1 (3 is too pedantic), alongside a few other
warnings like init-self, missing-declarations and vla.
Technically we're level=2, although cmocka throws a -Wclobbered which is
a PITA to disable when subproject is used.
Explicitly silence unused-parameter - there are too many instances
throughout the project. No point in adding dozens of (void *) casts.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Seems unhappy to compare int32_t vs enum, oops
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
The skip() function from cmocka is missing the noreturn notation. Thus
the compiler assumes that it can return and results in
implicit-fallthrough warnings.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
thrd_create() takes a `int (*)(void *)` callback, while currently we're
giving it a `bool (*)(struct thread_arg *)`
Flip the bool to an int.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Copy Wpointer-arith safe implementation from mesa and toggle the
warning in the meson build.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
None of them uses the fancy new features, but hey at least we're
consistent across the project.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Consistently set the define across the project. With that done, we can
enable -Werror=vla
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
GCC insists that assert_true(0) can return, thus assuming that we can
end using a uninitialized variable.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Keep the build clean by disabling the warnings locally. Using GCC-ism is
fine here, since the code is build only with GCC or Clang both of which
supporting it.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Drop the gratuitous wait and use the -noreset Xvfb option. This should
resolve the intermittent fallouts that we're seeing.
Huge thanks goes to Pekka and Daniel for pointing it out.
Closes: https://gitlab.freedesktop.org/mesa/waffle/-/issues/70
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
Currently wflinfo supports handling short opts, which are not space
separated. Additionally it also handles long-opt separated via =.
Neither of which is documented any more - latter was previously.
Just keep things simple and remove the extra quirky handling to those.
The program still supports them, so any users are unaffected.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
Current implementation supports space separated short and long opts.
Additionally it handles space-less short opts and = long opts.
The latter two are less common and adding support for them in the shell
completion is cumbersome. So drop them from the --help and man pages.
The program still supports them so existing users will still work.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
This produces artifacts which can be consumed by downstream piglit
builds.
Further changes can be done to restrict this to specific branches, tags,
or even releases, as publishing artifacts on every build is likely
overkill.
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
|
|
[Emil: split from a larger patch]
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|
|
Cmake version 3.19.3 throws a deprecation warning. Considering 2.8.12
has been around since 2013, it's perfectly reasonable to depend on it.
In the future we might consider bumping the requirement to something
more modern, but for now this gets silences the warning.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
|