GitLab

We cover all Render ops under the "blend" group, so when we're doing
the cross products of some formats for the masking operation, skip
most of the ops (covering just zero, one, and src/dst alpha blend
factors along with a definitely non-glamor-accelerated one) .

All the tests now complete in <20s of runtime on my skylake.
Signed-off-by: Eric Anholt <eric@anholt.net>

This gets us parallelism between rendercheck tests at the cost of
spinning up more Xvfbs, and nicer logging of the tests that are run.
Signed-off-by: Eric Anholt <eric@anholt.net>

I've found that it's hard to find the original error in a dump full of
xserver debug spam for the passing cases, and when I needed to look at
this I end up using a proper editor on the file from the artifacts
view anyway.
Signed-off-by: Eric Anholt <eric@anholt.net>

Signed-off-by: Eric Anholt <eric@anholt.net>

Michel noted that I could do this instead of bumping the tag as I
developed, so leave that note for the next person.
Signed-off-by: Eric Anholt <eric@anholt.net>

For testing xephyr-glamor on top of Xvfb in CI better, I want to be
able to make one command line describing the nested server invocation,
but that means I need to get two simple-xinits to split client/server
on different "--" arguments.
Signed-off-by: Eric Anholt <eric@anholt.net>

Early fatal errors may leave us with nothing in the displayfd pipe,
and we'd block forever.
Signed-off-by: Eric Anholt <eric@anholt.net>

Benjamin Tissoires authored Mar 18, 2019 and Michel Dänzer committed Mar 27, 2019

There shouldn't be a difference for users, but this way we do manage
all of our containers from freedesktop.org
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

systemd-logind since version 234 (released 2017-07-12) supports being
restarted without losing state [1]. From the systemd NEWS file [2]:

 * systemd-logind may now be restarted without losing state. It stores
   the file descriptors for devices it manages in the system manager
   using the FDSTORE= mechanism. Please note that further changes in
   other components may be required to make use of this (for example
   Xorg has code to listen for stops of systemd-logind and terminate
   itself when logind is stopped or restarted, in order to avoid using
   stale file descriptors for graphical devices, which is now
   counterproductive and must be reverted in order for restarts of
   systemd-logind to be safe. See
   https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)

This reverts commit dc48bd65.

Closes: #531

[1] https://github.com/systemd/systemd/pull/5600
[2] https://github.com/systemd/systemd/commit/9f09a95a7e6127f526c0ef7b6fd060cc0b01cdcc

Jon Turney authored Mar 14, 2019 and Peter Hutterer committed Mar 20, 2019

Applying get_pkgconfig_variable() to a not-found dependency was always
documented as an error, but meson 0.49 now actually raises an error[1]:

  meson.build:110:4: ERROR:  'xkbcomp' is not a pkgconfig dependency

Check xkbcomp_dep is a suitable dependency type before applying
get_pkgconfig_variable() to it.

[1] but this is more by accident than design (see the discusssion at [2]
et seq.), so who knows where things will come to rest...

[2] https://github.com/mesonbuild/meson/pull/4444#issuecomment-442443301

Michel Dänzer authored Feb 27, 2019 and Adam Jackson committed Mar 15, 2019

If they don't, flipping will result in corrupted display.

Test case:

* Run Xorg at 1920x1080 with no window manager
* glxgears -geometry 2048x1080

Jon Turney authored Mar 13, 2019 and Adam Jackson committed Mar 14, 2019

Drop some code conditional on meson version, which is not needed
since 7e046b94 bumped the minimum meson version

stub os support also needs to provide xf86OSInputThreadInit, omitted in
ea1527a8

Michel Dänzer authored Mar 08, 2019 and Michel Dänzer committed Mar 13, 2019

xwl_present_cleanup frees the struct xwl_present_window memory,
so if there's a pending callback, we have to destroy it to prevent
use-after-free in xwl_present_sync_callback.

Should fix issue #645.
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>

Adam Jackson authored Sep 30, 2014 and Adam Jackson committed Mar 11, 2019

This is maybe one more comparison, but it catches FullyObscured windows
slightly earlier, so it's kind of a wash. The important thing is this
allows for paintable but unmapped windows, which will have non-empty
clipList.

Adam Jackson authored Mar 06, 2019 and Adam Jackson committed Mar 11, 2019

No functional change, just folding some conditionals together.

Olivier Fourdan authored Mar 08, 2019 and Emma Anholt committed Mar 11, 2019

If `_glamor_create_tex()` fails to allocate the FBO because of
GL_OUT_OF_MEMORY error, the `pixmap_priv->fbo` is NULL.

However, `glamor_get_pixmap_texture()` doesn't actually check whether
the `pixmap_priv->fbo` is NULL and will segfault with a NULL pointer
dereference trying to access the `pixmap_priv->fbo->tex`.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Closes: xorg/xserver#647

We currently support two modes of operation for RW PBO buffers: one
that allocates a pack buffer with GL memory and one that uses system
memory when the former is not supported.

Since allocation with system memory is less likely to fail, add a
fallback to system memory when GL memory failed instead of bailing
out.
Signed-off-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>

glamor_prepare_access can fail for a few reasons, especially when
failing to allocate a PBO buffer. Take this in account and bail in
the copy helpers that call the helper when a failure happens.
Signed-off-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>

Packed buffer allocation (which happens at glBufferData time with the
buffer bound) can fail when there is no GL memory left.

Pick up the error when it happens, print a proper error message, do
some cleanup and bail.
Signed-off-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>

When uploading a picture to a texture, glamor_upload_picture_to_texture
calls glamor_pixmap_ensure_fbo to ensure that there is backing FBO.
The FBO will be allocated if the picture's drawable pixmap does not have
one already, which can fail when there is no GL memory left.

glamor_upload_picture_to_texture checks that the call succeeded and will
enter the failure path if it did not. However, unlike many other
functions in glamor, this one has ret set to TRUE initially, so it needs
to be set to FALSE when a failure happens.

Otherwise, the error is not propagated and the failure path return TRUE.
This leads to a fault when trying to access the FBO pointer later on.
Signed-off-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>

Andy Ritger authored Feb 13, 2019 and Adam Jackson committed Mar 06, 2019

Normally, the X server infers the initial screen size based on any
connected outputs.  However, if no outputs are connected, the X server
picks a default screen size of 1024 x 768.  This option overrides the
default screen size to use when no outputs are connected.  In contrast
to the "Virtual" Display SubSection entry, which applies unconditionally,
"NoOutputInitialSize" is only used if no outputs are detected when the
X server starts.

Parse this option in the new exported helper function
xf86AssignNoOutputInitialSize(), so that other XFree86 loadable drivers
can use it, even if they don't use xf86InitialConfiguration().
Signed-off-by: Andy Ritger <aritger@nvidia.com>
Reviewed-by: Keith Packard <keithp@keithp.com>

We were being naughty:

WARNING: Project specifies a minimum meson_version '>= 0.42.0' but uses features which were added in newer versions:
 * 0.46.0: {'compiler.has_multi_link_argument', 'compiler.has_link_argument'}

Michel Dänzer authored Feb 27, 2019 and Michel Dänzer committed Feb 27, 2019

This reverts commit 8694395f.

Some scenarios have come to light where this failed to ensure the docker
image exists:

* If the master branch of a forked repository is used for an MR which
  doesn't modify .gitlab-ci.yml, the docker-image job may not run.
* If the docker-image job of the first pipeline in a forked repository
  is cancelled or fails for any reason, and .gitlab-ci.yml isn't
  modified for the next pipeline run.

Alan Coopersmith authored Feb 21, 2019 and Adam Jackson committed Feb 25, 2019

Since the Solaris kernel tracks IOPL per thread, and doesn't inherit
raised IOPL levels when creating a new thread, we need to turn it on
in the input thread for input drivers like vmmouse that need register
access to work correctly.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>

Alan Coopersmith authored Feb 21, 2019 and Adam Jackson committed Feb 25, 2019

Allows os backends to run additional code as necessary to set up the
input thread.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>

Alan Coopersmith authored Feb 21, 2019 and Adam Jackson committed Feb 25, 2019

Allows ddx's to run additional code as necessary to set up the
input thread.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>

Alan Coopersmith authored Feb 21, 2019 and Adam Jackson committed Feb 25, 2019

Keeping track of kernel state in user space doesn't buy us anything,
and introduces bugs, as we were keeping global state but the Solaris
kernel tracks IOPL per thread.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>

Michel Dänzer authored Feb 22, 2019 and Michel Dänzer committed Feb 22, 2019

We don't set the PIGLIT_DIR and XTS_DIR environment variables for make
distcheck for now, otherwise it complains about files left behind by
piglit.

Michel Dänzer authored Feb 22, 2019 and Michel Dänzer committed Feb 22, 2019

Guard BUILT_SOURCES and CLEANFILES by XWIN_GLX_WINDOWS/XQUARTZ.

Michel Dänzer authored Feb 14, 2019 and Michel Dänzer committed Feb 20, 2019

Meson picks it up automatically.

Based on:

* mesa/mesa!240
* https://gould.cx/ted/blog/2017/06/10/ccache-for-Gitlab-CI/
* https://stackoverflow.com/questions/53659419/ccache-no-hits-in-gitlab-ci

v2 based on the corresponding Mesa change:
* Quote CCACHE_(BASE)DIR environment variables.
* Clear ccache stats in before_script.
* Move cache stanza to the build-and-test job, the cache isn't used in
  the docker-image job.

Reviewed-by: Adam Jackson <ajax@redhat.com> # v1

Michel Dänzer authored Feb 19, 2019 and Michel Dänzer committed Feb 19, 2019

We're going to make use of these in build & test jobs.

It doesn't require shared memory dir and thus allows
to avoid cases when this dir is detected incorrectly,
as in https://bugreports.qt.io/browse/QTBUG-71440

Signed-off-by: Alexander Volkov <a.volkov@rusbitech.ru>

The input thread checks the barriers for pointer positioning, swapping the
list out from underneath is considered impolite.
Reported-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

Michel Dänzer authored Feb 11, 2019 and Michel Dänzer committed Feb 13, 2019

The name of a forked repository can be changed later, in which case this
would fail to refer to the main repository.

Pointed out by Eric Engestrom in
mesa/mesa!224 .

Michel Dänzer authored Feb 07, 2019 and Michel Dänzer committed Feb 13, 2019

Otherwise there's normally no need to run it. It will also run when a
new branch is created, which ensures that the docker image always exists
(e.g. in a newly forked repository).

Inspired by mesa/mesa!143

Adam Jackson authored Jul 08, 2014 and Adam Jackson committed Jan 30, 2019

Now that there are no ramdac drivers this can go.
Signed-off-by: Adam Jackson <ajax@redhat.com>

Adam Jackson authored Feb 28, 2014 and Adam Jackson committed Jan 30, 2019

External RAMDACs are a very 1996 kind of thing, this code really doesn't
belong in the server.
Signed-off-by: Adam Jackson <ajax@redhat.com>

GetTimeInMillis is called first, which sets clockid to
CLOCK_MONOTONIC_COARSE, which is typically much lower resolution than
the callers of GetTimeInMicros want.

Prior to a779fda2

, GetTimeInMillis and
GetTimeInMicros did not share a clockid.

Restore the clockid split to fix the granularity of GetTimeInMicros.
Signed-off-by: Peter Harris <pharris@opentext.com>