The buffer release queue has two kinds of entries:
* Pending async flips.
* Completed flips waiting for their buffer to be released by the Wayland
compositor.
xwl_present_timer_callback neither completes async flips nor releases
buffers, so the timer isn't needed for the buffer release queue.
(cherry picked from commit e6cd1c9bde)
Fixes issue #12. Presumably the problem was that Present operations on
unmapped windows were executed immediately instead of only when reaching
the target MSC.
(cherry picked from commit f541615342)
When a window is unrealized, a pending frame callback may never be
called, which could result in repeatedly freezing until the frame timer
fires after a second.
Fixes these symptoms when switching from fullscreen to windowed mode in
sauerbraten.
(cherry picked from commit 8c9538573c)
There's no need to keep track of the window which last performed a
Present flip. This fixes crashes due to the assertion in
xwl_present_flips_stop failing. Fixes issue #10.
The damage generated by a flip only needs to be ignored once, then
xwl_window::present_flipped can be cleared. This may fix freezing in
the (hypothetical) scenario where Present flips are performed on a
window, followed by other drawing requests using the window as the
destination, but nothing triggering xwl_present_flips_stop. The damage
from the latter drawing requests would continue being ignored.
(cherry picked from commit 6b016d58d2)
Completing them from xwl_present_sync_callback had at least two issues:
* It was before the MSC was incremented in xwl_present_frame_callback,
so the MSC value in the completion event could be lower than the
target specified by the client. This could cause hangs with the Mesa
Vulkan drivers.
* It allowed clients to run at a frame-rate higher than the Wayland
compositor's frame-rate, wasting energy on generating frames which
were never displayed. This isn't expected to happen unless the client
specified PresentOptionAsync (in which case flips are still completed
from xwl_present_sync_callback, allowing higher frame-rates).
v2:
* Make xwl_present_has_events return true when there's a pending
"synchronous" flip, so those complete after at most ~1 second even if
the Wayland server doesn't send a frame event.
Bugzilla: https://bugs.freedesktop.org/106713
(cherry picked from commit ace551d8a2)
Apart from simplifying the code, this should also prevent a condition
(which might only be possible with the following fix) reported in
https://gitlab.freedesktop.org/wayland/weston/issues/115#note_52467:
1. xwl_present_timer_callback indirectly calls xwl_present_reset_timer
-> xwl_present_free_timer
2. xwl_present_timer_callback then returns a non-0 value, so DoTimer
calls TimerSet with the old xwl_present_window->frame_timer pointer
which was freed in step 1 => use after free
Calling xwl_present_reset_timer explicitly passes NULL to TimerSet if
step 1 freed xwl_present_window->frame_timer, and it will allocate a new
one.
(cherry picked from commit 5e8b9a3a56)
The function `xwl_glamor_gbm_create_pixmap()` first creates a buffer
objects and then creates the xwl_pixmap from it.
However, `xwl_glamor_gbm_create_pixmap_for_bo()` is not called if the
buffer object creation fails, and `xwl_glamor_gbm_create_pixmap()`
simply returns `glamor_create_pixmap()`.
The problem with this is that if `xwl_glamor_gbm_create_pixmap_for_bo()`
is not called then neither is `xwl_pixmap_set_private()` and further
calls to `xwl_pixmap_get()` will return NULL and cause a NULL pointer
dereference if the return value is not checked:
#0 xwl_glamor_gbm_get_wl_buffer_for_pixmap ()
at hw/xwayland/xwayland-glamor-gbm.c:248
#1 xwl_window_post_damage () at hw/xwayland/xwayland.c:697
#2 xwl_display_post_damage () at hw/xwayland/xwayland.c:759
#3 block_handler () at hw/xwayland/xwayland.c:890
#4 BlockHandler () at dix/dixutils.c:388
#5 WaitForSomething () at os/WaitFor.c:201
#6 Dispatch () at dix/dispatch.c:421
#7 dix_main () at dix/main.c:276
#8 __libc_start_main () at ../csu/libc-start.c:308
#9 _start ()
(gdb) print xwl_pixmap
$1 = (struct xwl_pixmap *) 0x0
Make sure we check for `xwl_pixmap_get()` returned value where relevant
and fail gracefully if this is the case.
See also: https://gitlab.gnome.org/GNOME/mutter/issues/340
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Marco Trevisan <mail@3v1n0.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 036794bebc)
xwl_present_window->sync_callback was leaked.
The event memory was leaked if the corresponding buffer had already been
released.
(cherry picked from commit cb0de153bf)
0a9415cf apparently can tickle bugs in the GL stack where glGetString
returns NULL, presumably because the eglMakeCurrent() didn't manage to
actually install a dispatch table and you're hitting a stub function.
That's clearly not our bug, but if it happens we should at least not
crash. Notice this case and fail gently.
Signed-off-by: Adam Jackson <ajax@redhat.com>
(cherry picked from commit af151895f3)
Mesa started supporting GL_OES_EGL_image on llvmpipe in 17.3, after this
commit:
commit bbdeddd5fd0b797e1e281f058338b3da4d98029d
Author: Gurchetan Singh <gurchetansingh@chromium.org>
Date: Tue Aug 1 14:49:33 2017 -0700
st/dri: add drisw image extension
That's pretty cool, but it means glamor now thinks it can initialize on
llvmpipe. This is almost certainly not what anyone wants, as glamor on
llvmpipe is pretty much uniformly slower than fb.
This fixes both Xorg and Xwayland to refuse glamor in such a setup.
Xephyr is left alone, both because glamor is not the default there and
because Xephyr+glamor+llvmpipe is one of the easier ways to get xts to
exercise glamor.
The (very small) downside of this change is that you lose DRI3 support.
This wouldn't have helped you very much (since an lp glamor blit is
slower than a pixman blit), but it would eliminate the PutImage overhead
for llvmpipe's glXSwapBuffers. A future change should add DRI3 support
for the fb-only case.
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Adam Jackson <ajax@redhat.com>
(cherry picked from commit 0a9415cf79)
Xwayland's `xwl_destroy_window()` invokes `xwl_present_cleanup()`
before the common `DestroyWindow()`.
But then `DestroyWindow()` calls `present_destroy_window()` which will
possibly end up in `xwl_present_abort_vblank()` which will try to access
data that was previously freed by `xwl_present_cleanup()`:
Invalid read of size 8
at 0x434184: xwl_present_abort_vblank (xwayland-present.c:378)
by 0x53785B: present_wnmd_abort_vblank (present_wnmd.c:651)
by 0x53695A: present_free_window_vblank (present_screen.c:87)
by 0x53695A: present_destroy_window (present_screen.c:152)
by 0x42A90D: xwl_destroy_window (xwayland.c:653)
by 0x584298: compDestroyWindow (compwindow.c:613)
by 0x53CEE3: damageDestroyWindow (damage.c:1570)
by 0x4F1BB8: DbeDestroyWindow (dbe.c:1326)
by 0x46F7F6: FreeWindowResources (window.c:1031)
by 0x472847: DeleteWindow (window.c:1099)
by 0x46B54C: doFreeResource (resource.c:880)
by 0x46C706: FreeClientResources (resource.c:1146)
by 0x446ADE: CloseDownClient (dispatch.c:3473)
Address 0x182abde0 is 80 bytes inside a block of size 112 free'd
at 0x4C2FDAC: free (vg_replace_malloc.c:530)
by 0x42A937: xwl_destroy_window (xwayland.c:647)
by 0x584298: compDestroyWindow (compwindow.c:613)
by 0x53CEE3: damageDestroyWindow (damage.c:1570)
by 0x4F1BB8: DbeDestroyWindow (dbe.c:1326)
by 0x46F7F6: FreeWindowResources (window.c:1031)
by 0x472847: DeleteWindow (window.c:1099)
by 0x46B54C: doFreeResource (resource.c:880)
by 0x46C706: FreeClientResources (resource.c:1146)
by 0x446ADE: CloseDownClient (dispatch.c:3473)
by 0x446DA5: ProcKillClient (dispatch.c:3279)
by 0x4476AF: Dispatch (dispatch.c:479)
Block was alloc'd at
at 0x4C30B06: calloc (vg_replace_malloc.c:711)
by 0x433F46: xwl_present_window_get_priv (xwayland-present.c:54)
by 0x434228: xwl_present_get_crtc (xwayland-present.c:302)
by 0x539728: proc_present_query_capabilities (present_request.c:227)
by 0x4476AF: Dispatch (dispatch.c:479)
by 0x44B5B5: dix_main (main.c:276)
by 0x75F611A: (below main) (libc-start.c:308)
This is because `xwl_present_cleanup()` frees the memory but does not
remove it from the window's privates, and `xwl_present_abort_vblank()`
will still find it and hence try to access that freed memory...
Remove `xwl_present_window` from window's privates on cleanup so that no
other function can find and reuse that data once it's freed.
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1616269
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit 3f31f56929)
xwl_output->randr_crtc is used in the update_screen_size() function :
==5331== Invalid read of size 4
==5331== at 0x15263D: update_screen_size (xwayland-output.c:190)
==5331== by 0x152C48: xwl_output_remove (xwayland-output.c:413)
==5331== by 0x6570FCD: ffi_call_unix64 (in /usr/lib/x86_64-linux-gnu/libffi.so.6.0.4)
==5331== by 0x657093E: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.6.0.4)
==5331== by 0x4DDB183: ??? (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x4DD79D8: ??? (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x4DD8EA3: wl_display_dispatch_queue_pending (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x14BCCA: xwl_read_events (xwayland.c:814)
==5331== by 0x2AC0D0: ospoll_wait (ospoll.c:651)
==5331== by 0x2A5322: WaitForSomething (WaitFor.c:208)
==5331== by 0x27574B: Dispatch (dispatch.c:421)
==5331== by 0x279945: dix_main (main.c:276)
==5331== Address 0x1aacb5f4 is 36 bytes inside a block of size 154 free'd
==5331== at 0x48369EB: free (vg_replace_malloc.c:530)
==5331== by 0x1F8AE8: RROutputDestroyResource (rroutput.c:421)
==5331== by 0x29A2AC: doFreeResource (resource.c:880)
==5331== by 0x29AE5B: FreeResource (resource.c:910)
==5331== by 0x152BE0: xwl_output_remove (xwayland-output.c:408)
==5331== by 0x6570FCD: ffi_call_unix64 (in /usr/lib/x86_64-linux-gnu/libffi.so.6.0.4)
==5331== by 0x657093E: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.6.0.4)
==5331== by 0x4DDB183: ??? (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x4DD79D8: ??? (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x4DD8EA3: wl_display_dispatch_queue_pending (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x14BCCA: xwl_read_events (xwayland.c:814)
==5331== by 0x2AC0D0: ospoll_wait (ospoll.c:651)
==5331== Block was alloc'd at
==5331== at 0x48357BF: malloc (vg_replace_malloc.c:299)
==5331== by 0x1F93E0: RROutputCreate (rroutput.c:83)
==5331== by 0x152A75: xwl_output_create (xwayland-output.c:361)
==5331== by 0x14BE59: registry_global (xwayland.c:764)
==5331== by 0x6570FCD: ffi_call_unix64 (in /usr/lib/x86_64-linux-gnu/libffi.so.6.0.4)
==5331== by 0x657093E: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.6.0.4)
==5331== by 0x4DDB183: ??? (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x4DD79D8: ??? (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x4DD8EA3: wl_display_dispatch_queue_pending (in /usr/lib/x86_64-linux-gnu/libwayland-client.so.0.3.0)
==5331== by 0x14BCCA: xwl_read_events (xwayland.c:814)
==5331== by 0x2AC0D0: ospoll_wait (ospoll.c:651)
==5331== by 0x2A5322: WaitForSomething (WaitFor.c:208)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
(cherry picked from commit 53ce2ba0a1)
This prevents multiple scroll events happening for wayland compositors
which send axis values other than 10. For example, libinput will
typically return 15 for each scroll wheel step, and if a wayland
compositor sends those to xwayland without normalising them, 2 scroll
wheel steps will end up as 3 xorg scroll events. By listening for the
discrete_axis event, this will now correctly send only 2 xorg scroll
events.
The wayland protocol gurantees that there will always be an axis event
following an axis_discrete event. However, it does not gurantee that
other events (including other axis_discrete+axis pairs) will not happen
in between them. So we must keep a list of outstanding axis_discrete
events.
Signed-off-by: Scott Anderson <scott@anderso.nz>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit cd285922cd)
The logical size is the size of the output in the global compositor
space. The mode width/height should be scaled as in the logical
size, but shouldn't be transformed. Thus we need to rotate back
the logical size to be able to use it as the mode width/height.
This fixes issues with pointer input on transformed outputs.
Signed-Off-By: Simon Ser <contact@emersion.fr>
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit ce2dde9ed0)
glamor_fds_from_pixmap() will bail out early if DRI3 is not enabled,
unfortunately Xwayland's glamor code would not set it as enabled which
would lead to blank pixmaps when using texture from pixmap.
Make sure to mark DRI3 as enabled from glamor_egl_screen_init() in
Xwayland.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107287
Fixes: c8c276c956 ("glamor: Implement PixmapFromBuffers and BuffersFromPixmap")
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
(cherry picked from commit cdec2b3c19)
When support for allocating GBM BOs with modifiers was added,
glamor_fd_from_pixmap() was changed so that it would return an error if
it got a bo with modifiers set from glamor_fds_from_pixmap(). The
problem is that on systems that support BOs with modifiers,
glamor_fds_from_pixmap() will always return BOs with modifiers.
This means that glamor_fd_from_pixmap() was broken entirely, which broke
a number of other things including glamor_shareable_fd_from_pixmap(),
which meant that modesetting using multiple GPUs with the modesetting
DDX was also broken. Easy reproducer:
- Find a laptop with DRI prime that has outputs connected to the
dedicated GPU and integrated GPU
- Try to enable one display on each using the modesetting DDX
- Fail
Since there isn't a way to ask for no modifiers from
glamor_fds_from_pixmap, we create a shared _glamor_fds_from_pixmap()
function used by both glamor_fds_from_pixmap() and
glamor_fd_from_pixmap() that calls down to the appropriate
glamor_egl_fd*_from_pixmap() function.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Cc: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Fixes: c8c276c956 ("glamor: Implement PixmapFromBuffers and BuffersFromPixmap")
(cherry picked from commit 186a21c4ba)
The API init_wl_registry() and has_wl_interfaces() are marked as being
optional, but both GBM And EGLStream backends implement them so there is
point in keeping those optional.
Suggested-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit 92daeb31fa)
When retrieving the Wayland buffer from a pixmap, if the buffer already
exists, the GBM backend will return that existing buffer.
However, as seen with the Present issues, if the call had previously
passed a wrong size, that buffer will remain at the wrong size for as
long as the buffer exists, which is error prone.
Considering that the width/height passed to get_wl_buffer() is always the
actual pixmap drawable size, and considering that the EGLStream backend
makes no use of the size either, there is really no point in passing the
width/height around.
Simplify the xwl_glamor_pixmap_get_wl_buffer() and EGL backends API by
removing the pixmap size, and use the drawable size instead.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit 792359057b)
xwl_glamor_eglstream_init_egl() uses "EGL_IMG_context_priority"
extension, make sure it's actually available before using it.
Suggested-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit bdadaa25f5)
Now that we have separate backends for EGLStream and GBM, we can
explicitly check for the EGLStream backend to disable present support
in that case.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit 5d843f6947)
To be able to check for availability of the Wayland interfaces required
to run a given EGL backend (either GBM or EGLStream for now), we need
to have each backend structures and vfuncs in place before we enter the
Wayland registry dance.
That basically means that we should init all backends at first, connect
to the Wayland compositor and query the available interfaces and then
decide which backend is available and should be used (or none if either
the Wayland interfaces or the EGL extensions are not available).
For this purpose, hold an egl_backend struct for each backend we are to
consider prior to connect to the Wayland display so that, when we get to
query the Wayland interfaces, everything is in place for each backend to
handle the various Wayland interfaces.
Eventually, when we need to chose which EGL backend to use for glamor,
the available Wayland interfaces and EGL extensions available are all
known to Xwayland.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit d7185a84b6)
Move EGL backends initialization to its own function in
xwayland-glamor.c
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit 48f037a27c)
Introduces a new egl_backend function to let the EGL backend check for
the presence of the required Wayland interfaces.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit f2fcb4877e)
EGL backend availability requires both EGL extensions and Wayland
interfaces to be present, so we will need to consider multiple backends
during initialization.
As a preliminary work, move the egl_backend to its own struct so that we
can have more than one backend at any given time.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit b74b0f18b8)
If using a render node, we can skip DRM authentication.
Suggested-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit de004eefc6)
Surely, we should fail to init GBM backend if "GL_OES_EGL_image" is
missing.
This seems to have been lost with commit 1545e2dba ("xwayland: Decouple
GBM from glamor").
Suggested-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit b823b43dca)
Both xwl_glamor_init_wl_registry() and the Wayland global registry
handler use the interface id/name in that order, using name/id in the
egl_backend vfunc makes things confusing and error prone.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit 78ce4aa979)
Functions such as:
xwl_glamor_egl_supports_device_probing()
xwl_glamor_egl_get_devices()
xwl_glamor_egl_device_has_egl_extensions()
Are of no use outside of EGLStream support, move them to the relevant
source file.
Similarly, the other glamor functions such as:
xwl_glamor_init()
xwl_screen_set_drm_interface()
xwl_screen_set_dmabuf_interface()
xwl_glamor_pixmap_get_wl_buffer()
xwl_glamor_init_wl_registry()
xwl_glamor_post_damage()
xwl_glamor_allow_commits()
xwl_glamor_egl_make_current()
Are useless without glamor support enabled, move those within a
a "#ifdef XWL_HAS_GLAMOR" in xwayland.h
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit f6b2109c1b)
Make xwl_output_get_xdg_output() private, it doesn't need to be
available elsewhere.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit d31a7be15e)
EGLStream requires glamor, but the opposite is not true. So if someone
passes "-eglstream" with a GPU which does not support EGLStream, we
could maybe still try GBM and be lucky.
That allows Wayland compositors to pass "-eglstream" regardless of the
actual hardware, if they want to enable EGLStream on GPU which support
it.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit e16a6da79d)
eglQueryDevicesEXT() would abort if the required extensions are not
available, meaning that enabling “-eglstream” on a non-EGLStream
capable hardware would lead to an abort().
Check that "EGL_EXT_device_base" extension is available and bail out
early if not, so we don't abort() later in eglQueryDevicesEXT().
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit dbde3fec32)
The command line option "-eglstream" used to enable EGLStream support
for NVidia GPU was made available only when Xwayland was built with
EGLStream support enabled.
Wayland compositors who spawn Xwayland have no easy way to tell whether
or not Xwayland was built with EGLStream support enabled, and adding
"-eglstream" command line option to Xwayland when it wasn't built with
EGLStream support would prevent Xwayland from starting (“Unrecognized
option” error).
Make sure we support the command line option "-eglstream" regardless of
EGLStream support in Xwayland. Obviously, if Xwayland was built without
EGLStream support, this has no effect.
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
(cherry picked from commit 06c31e782e)
If the pixmap size does not match the present box size, flickering
occurs.
This can happen when the client changes its size (e.g. switching to
fullscreen), and since the buffer is kept as long as the pixmap is
valid, once the buffer is created, it remains at the wrong (old) size
and causes continuous flickering.
Use the actual pixmap's drawable size instead of the present box to
create the buffer so that it's sized appropriately.
Bugzilla: https://bugs.freedesktop.org/106841
Fixes: 0fb2cca193 "xwayland: Preliminary support for Present's new
window flip mode"
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Roman Gilg <subdiff@gmail.com>
(cherry picked from commit 1993f147d0)
Pointed out on irc by q66.
Reviewed-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
(cherry picked from commit 6300049a9a)
We probably don't want a fake crtc to be visible to clients, and we
definitely don't want to generate events every time we create such a
fake (which would happen as a side effect from RRCrtcCreate hitting
RRTellChanged). As it happens we're not actually using that crtc for
anything because xwayland doesn't store any state on the crtc object,
so it suffices to use the real crtc for the screen.
Signed-off-by: Adam Jackson <ajax@redhat.com>
Tested-by: Roman Gilg <subdiff@gmail.com>
Reviewed-by: Roman Gilg <subdiff@gmail.com>
We were mixing stdint and CARD* types, causing compiler warnings on
32-bit. Just switch over to stdint, which is what we'd like the server
to be using long term, anyway.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Instead of reusing xwl_window introduce a persistent window struct for every
window, that asks for Present flips.
This struct saves all relevant data and is only freed on window destroy.
Signed-off-by: Roman Gilg <subdiff@gmail.com>
Tested-by: Olivier Fourdan <ofourdan@redhat.com>
Clean up only if the request points to the presenting window or its top
parent window.
Since in this case all events are removed unconditionally, always stop
the timer.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Roman Gilg <subdiff@gmail.com>
There's no real point - if we don't have EGL then the extension check is
also going to fail - and the entrypoint is new in 1.5.0, which we don't
need to require yet.
Signed-off-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
This adds initial support for displaying Xwayland applications through
the use of EGLStreams and nvidia's custom wayland protocol by adding
another egl_backend driver. This also adds some additional egl_backend
hooks that are required to make things work properly.
EGLStreams work a lot differently then the traditional way of handling
buffers with wayland. Unfortunately, there are also a LOT of various
pitfalls baked into it's design that need to be explained.
This has a very large and unfortunate implication: direct rendering is,
for the time being at least, impossible to do through EGLStreams. The
main reason being that the EGLStream spec mandates that we lose the
entire color buffer contents with each eglSwapBuffers(), which goes
against X's requirement of not losing data with pixmaps. no way to use
an allocated EGLSurface as the storage for glamor rendering like we do
with GBM, we have to rely on blitting each pixmap to it's respective
EGLSurface producer each frame. In order to pull this off, we add two
different additional egl_backend hooks that GBM opts out of
implementing:
- egl_backend.allow_commits for holding off displaying any EGLStream
backed pixmaps until the point where it's stream is completely
initialized and ready for use
- egl_backend.post_damage for blitting the content of the EGLStream
surface producer before Xwayland actually damages and commits the
wl_surface to the screen.
The other big pitfall here is that using nvidia's wayland-eglstreams
helper library is also not possible for the most part. All of it's API
for creating and destroying streams rely on being able to perform a
roundtrip in order to bring each stream to completion since the wayland
compositor must perform it's job of connecting a consumer to each
EGLstream. Because Xwayland has to potentially handle both responding to
the wayland compositor and it's own X clients, the situation of the
wayland compositor being one of our X clients must be considered. If we
perform a roundtrip with the Wayland compositor, it's possible that the
wayland compositor might currently be connected to us as an X client and
thus hang while both Xwayland and the wayland compositor await responses
from eachother. To avoid this, we work directly with the wayland
protocol and use wl_display_sync() events along with release() events to
set up and destroy EGLStreams asynchronously alongside handling X
clients.
Additionally, since setting up EGLStreams is not an atomic operation we
have to take into consideration the fact that an EGLStream can
potentially be created in response to a window resize, then immediately
deleted due to another pending window resize in the same X client's
pending reqests before Xwayland hits the part of it's event loop where
we read from the wayland compositor. To make this even more painful, we
also have to take into consideration that since EGLStreams are not
atomic that it's possible we could delete wayland resources for an
EGLStream before the compositor even finishes using them and thus run
into errors. So, we use quite a bit of tracking logic to keep EGLStream
objects alive until we know the compositor isn't using them (even if
this means the stream outlives the pixmap it backed).
While the default backend for glamor remains GBM, this patch exists for
users who have had to deal with the reprecussion of their GPU
manufacturers ignoring the advice of upstream and the standardization of
GBM across most major GPU manufacturers. It is not intended to be a
final solution to the GBM debate, but merely a baindaid so our users
don't have to suffer from the consequences of companies avoiding working
upstream. New drivers are strongly encouraged not to use this as a
backend, and use GBM like everyone else. We even spit this out as an
error from Xwayland when using the eglstream backend.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Just a small autogenerated header that will soon contain more then just
one macro.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
This takes all of the gbm related code in wayland-glamor.c and moves it
into it's own EGL backend for Xwayland, xwayland-glamor-gbm.c.
Additionally, we add the egl_backend struct into xwl_screen in order to
provide hooks for alternative EGL backends such as nvidia's EGLStreams.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Regardless of the order we un-realize windows.
Suggested-by: Roman Gilg <subdiff@gmail.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Roman Gilg <subdiff@gmail.com>
xwl_unrealize_window() would use freed xwl_window which can lead to
various memory corruption and crashes, as reported by valgrind:
Invalid read of size 8
at 0x42C802: xwl_present_cleanup (xwayland-present.c:84)
by 0x42BA67: xwl_unrealize_window (xwayland.c:601)
by 0x541EE9: compUnrealizeWindow (compwindow.c:285)
by 0x57E1FA: UnrealizeTree (window.c:2816)
by 0x581189: UnmapWindow (window.c:2874)
by 0x54EB26: ProcUnmapWindow (dispatch.c:879)
by 0x554B7D: Dispatch (dispatch.c:479)
by 0x558BE5: dix_main (main.c:276)
by 0x7C4B1BA: (below main) (libc-start.c:308)
Address 0xf520f60 is 96 bytes inside a block of size 184 free'd
at 0x4C2EDAC: free (vg_replace_malloc.c:530)
by 0x42B9FB: xwl_unrealize_window (xwayland.c:624)
by 0x541EE9: compUnrealizeWindow (compwindow.c:285)
by 0x57E1FA: UnrealizeTree (window.c:2816)
by 0x581189: UnmapWindow (window.c:2874)
by 0x54EB26: ProcUnmapWindow (dispatch.c:879)
by 0x554B7D: Dispatch (dispatch.c:479)
by 0x558BE5: dix_main (main.c:276)
by 0x7C4B1BA: (below main) (libc-start.c:308)
Block was alloc'd at
at 0x4C2FB06: calloc (vg_replace_malloc.c:711)
by 0x42B307: xwl_realize_window (xwayland.c:488)
by 0x541E59: compRealizeWindow (compwindow.c:268)
by 0x57DA40: RealizeTree (window.c:2617)
by 0x580B28: MapWindow (window.c:2694)
by 0x54EA2A: ProcMapWindow (dispatch.c:845)
by 0x554B7D: Dispatch (dispatch.c:479)
by 0x558BE5: dix_main (main.c:276)
by 0x7C4B1BA: (below main) (libc-start.c:308)
This is because UnrealizeTree() traverses the tree from top to bottom,
which invalidates the assumption that if the Window doesn't feature an
xwl_window on its own, it's the xwl_window of its first ancestor with
one.
This reverts commit 82df2ce3
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Present support in Xwayland relies on glamor, make sure Xwayland can
be built without glamor by moving references to Present code inside
the conditional GLAMOR_HAS_GBM.
Reported-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Roman Gilg <subdiff@gmail.com>
Turns out that's legal, and xts exercises it, and we crash:
Thread 1 "Xwayland" received signal SIGSEGV, Segmentation fault.
dixGetPrivate (key=0x813660 <xwl_window_private_key>, privates=0x20) at ../../include/privates.h:122
122 return (char *) (*privates) + key->offset;
(gdb) bt
#0 dixGetPrivate (key=0x813660 <xwl_window_private_key>, privates=0x20) at ../../include/privates.h:122
#1 dixLookupPrivate (key=0x813660 <xwl_window_private_key>, privates=0x20) at ../../include/privates.h:166
#2 xwl_window_of_top (window=0x0) at xwayland.c:128
#3 xwl_cursor_warped_to (device=<optimized out>, screen=0x268b6e0, client=<optimized out>, window=0x0, sprite=0x300bb30,
x=2400, y=1350) at xwayland.c:292
#4 0x00000000005622ec in ProcWarpPointer (client=0x32755d0) at events.c:3618
In this case, x/y are the screen-space coordinates where the pointer
ends up, and we need to look up the (X) window there.
Signed-off-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>