xserver/hw/xwayland/xwayland-glamor.c

1179 lines
38 KiB
C
Raw Normal View History

/*
* Copyright © 2011-2014 Intel Corporation
*
* Permission to use, copy, modify, distribute, and sell this software
* and its documentation for any purpose is hereby granted without
* fee, provided that the above copyright notice appear in all copies
* and that both that copyright notice and this permission notice
* appear in supporting documentation, and that the name of the
* copyright holders not be used in advertising or publicity
* pertaining to distribution of the software without specific,
* written prior permission. The copyright holders make no
* representations about the suitability of this software for any
* purpose. It is provided "as is" without express or implied
* warranty.
*
* THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
* SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
* FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
* SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
* AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
* OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
* SOFTWARE.
*/
#include <xwayland-config.h>
#include <compositeext.h>
#define MESA_EGL_NO_X11_HEADERS
#define EGL_NO_X11
#include <glamor_egl.h>
#include <glamor.h>
#include <glamor_context.h>
#ifdef GLXEXT
#include "glx_extinit.h"
#endif
#include "linux-dmabuf-unstable-v1-client-protocol.h"
#include "drm-client-protocol.h"
#include <drm_fourcc.h>
#include "xwayland-glamor.h"
#include "xwayland-glx.h"
#include "xwayland-screen.h"
#include "xwayland-window.h"
#include "xwayland-window-buffers.h"
#include <sys/mman.h>
static void
xwayland: Add glamor egl_backend for EGLStreams 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>
2018-04-20 14:38:05 -04:00
glamor_egl_make_current(struct glamor_context *glamor_ctx)
{
eglMakeCurrent(glamor_ctx->display, EGL_NO_SURFACE,
EGL_NO_SURFACE, EGL_NO_CONTEXT);
if (!eglMakeCurrent(glamor_ctx->display,
EGL_NO_SURFACE, EGL_NO_SURFACE,
glamor_ctx->ctx))
FatalError("Failed to make EGL context current\n");
}
xwayland: Add glamor egl_backend for EGLStreams 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>
2018-04-20 14:38:05 -04:00
void
xwl_glamor_egl_make_current(struct xwl_screen *xwl_screen)
{
EGLContext ctx = xwl_screen->glamor_ctx->ctx;
if (lastGLContext == ctx)
xwayland: Add glamor egl_backend for EGLStreams 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>
2018-04-20 14:38:05 -04:00
return;
lastGLContext = ctx;
xwayland: Add glamor egl_backend for EGLStreams 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>
2018-04-20 14:38:05 -04:00
xwl_screen->glamor_ctx->make_current(xwl_screen->glamor_ctx);
}
void
glamor_egl_screen_init(ScreenPtr screen, struct glamor_context *glamor_ctx)
{
struct xwl_screen *xwl_screen = xwl_screen_get(screen);
glamor_set_glvnd_vendor(screen, xwl_screen->glvnd_vendor);
glamor_enable_dri3(screen);
glamor_ctx->ctx = xwl_screen->egl_context;
glamor_ctx->display = xwl_screen->egl_display;
xwayland: Add glamor egl_backend for EGLStreams 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>
2018-04-20 14:38:05 -04:00
glamor_ctx->make_current = glamor_egl_make_current;
xwl_screen->glamor_ctx = glamor_ctx;
}
Bool
xwl_glamor_check_flip(WindowPtr present_window, PixmapPtr pixmap)
{
ScreenPtr screen = pixmap->drawable.pScreen;
struct xwl_screen *xwl_screen = xwl_screen_get(screen);
PixmapPtr backing_pixmap = screen->GetWindowPixmap(present_window);
if (pixmap->drawable.depth != backing_pixmap->drawable.depth)
return FALSE;
if (!xwl_glamor_pixmap_get_wl_buffer(pixmap))
return FALSE;
if (xwl_screen->egl_backend->check_flip)
return xwl_screen->egl_backend->check_flip(pixmap);
return TRUE;
}
static Bool
xwl_glamor_is_modifier_supported_in_formats(struct xwl_format *formats, int num_formats,
uint32_t format, uint64_t modifier)
{
struct xwl_format *xwl_format = NULL;
int i;
for (i = 0; i < num_formats; i++) {
if (formats[i].format == format) {
xwl_format = &formats[i];
break;
}
}
if (xwl_format) {
for (i = 0; i < xwl_format->num_modifiers; i++) {
if (xwl_format->modifiers[i] == modifier) {
return TRUE;
}
}
}
return FALSE;
}
static Bool
xwl_feedback_is_modifier_supported(struct xwl_dmabuf_feedback *xwl_feedback,
uint32_t format, uint64_t modifier,
int supports_scanout)
{
for (int i = 0; i < xwl_feedback->dev_formats_len; i++) {
struct xwl_device_formats *dev_formats = &xwl_feedback->dev_formats[i];
if (supports_scanout && !dev_formats->supports_scanout)
continue;
if (xwl_glamor_is_modifier_supported_in_formats(dev_formats->formats,
dev_formats->num_formats,
format, modifier))
return TRUE;
}
return FALSE;
}
Bool
xwl_glamor_is_modifier_supported(struct xwl_screen *xwl_screen,
uint32_t format, uint64_t modifier)
{
struct xwl_window *xwl_window;
/*
* If we are using dmabuf v4, then we need to check in the main
* device and per-window format lists. For older protocol
* versions we can just check the list returned by the dmabuf.modifier
* events in xwl_screen
*/
if (xwl_screen->dmabuf_protocol_version < 4) {
return xwl_glamor_is_modifier_supported_in_formats(xwl_screen->formats,
xwl_screen->num_formats,
format, modifier);
}
if (xwl_feedback_is_modifier_supported(&xwl_screen->default_feedback, format, modifier, FALSE))
return TRUE;
xorg_list_for_each_entry(xwl_window, &xwl_screen->window_list, link_window) {
if (xwl_feedback_is_modifier_supported(&xwl_window->feedback, format, modifier, FALSE))
return TRUE;
}
return FALSE;
}
uint32_t
wl_drm_format_for_depth(int depth)
{
switch (depth) {
case 15:
return WL_DRM_FORMAT_XRGB1555;
case 16:
return WL_DRM_FORMAT_RGB565;
case 24:
return WL_DRM_FORMAT_XRGB8888;
case 30:
return WL_DRM_FORMAT_ARGB2101010;
default:
ErrorF("unexpected depth: %d\n", depth);
case 32:
return WL_DRM_FORMAT_ARGB8888;
}
}
static drmDevice *
xwl_screen_get_main_dev(struct xwl_screen *xwl_screen)
{
/*
* If we have gbm then get our main device from it. Otherwise use what
* the compositor told us.
*/
if (xwl_screen->gbm_backend.is_available)
return xwl_screen->gbm_backend.get_main_device(xwl_screen);
else
return xwl_screen->default_feedback.main_dev;
}
static Bool
xwl_get_formats(struct xwl_format *format_array, int format_array_len,
uint32_t *num_formats, uint32_t **formats)
{
*num_formats = 0;
*formats = NULL;
if (format_array_len == 0)
return TRUE;
*formats = calloc(format_array_len, sizeof(CARD32));
if (*formats == NULL)
return FALSE;
for (int i = 0; i < format_array_len; i++)
(*formats)[i] = format_array[i].format;
*num_formats = format_array_len;
return TRUE;
}
static Bool
xwl_get_formats_for_device(struct xwl_dmabuf_feedback *xwl_feedback, drmDevice *device,
uint32_t *num_formats, uint32_t **formats)
{
uint32_t *ret = NULL;
uint32_t count = 0;
/* go through all matching sets of tranches for the window's device */
for (int i = 0; i < xwl_feedback->dev_formats_len; i++) {
if (drmDevicesEqual(xwl_feedback->dev_formats[i].drm_dev, device)) {
struct xwl_device_formats *dev_formats = &xwl_feedback->dev_formats[i];
/* Append the formats from this tranche to the list */
ret = xnfreallocarray(ret, count + dev_formats->num_formats, sizeof(CARD32));
for (int j = 0; j < dev_formats->num_formats; j++) {
bool found = false;
/* Check if this format is already present in the list */
for (int k = 0; k < count; k++) {
if (ret[k] == dev_formats->formats[j].format) {
found = true;
break;
}
}
/* If this format has not yet been added, do so now */
if (!found)
ret[count++] = dev_formats->formats[j].format;
}
}
}
*num_formats = count;
*formats = ret;
return TRUE;
}
Bool
xwl_glamor_get_formats(ScreenPtr screen,
CARD32 *num_formats, CARD32 **formats)
{
struct xwl_screen *xwl_screen = xwl_screen_get(screen);
/* Explicitly zero the count as the caller may ignore the return value */
*num_formats = 0;
if (!xwl_screen->dmabuf)
return FALSE;
if (xwl_screen->dmabuf_protocol_version >= 4) {
drmDevice *main_dev = xwl_screen_get_main_dev(xwl_screen);
return xwl_get_formats_for_device(&xwl_screen->default_feedback, main_dev,
num_formats, formats);
}
return xwl_get_formats(xwl_screen->formats, xwl_screen->num_formats,
num_formats, formats);
}
static Bool
xwl_get_modifiers_for_format(struct xwl_format *format_array, int num_formats,
uint32_t format, uint32_t *num_modifiers, uint64_t **modifiers)
{
struct xwl_format *xwl_format = NULL;
int i;
*num_modifiers = 0;
*modifiers = NULL;
if (num_formats == 0)
return TRUE;
for (i = 0; i < num_formats; i++) {
if (format_array[i].format == format) {
xwl_format = &format_array[i];
break;
}
}
if (!xwl_format ||
(xwl_format->num_modifiers == 1 &&
xwl_format->modifiers[0] == DRM_FORMAT_MOD_INVALID))
return FALSE;
*modifiers = calloc(xwl_format->num_modifiers, sizeof(uint64_t));
if (*modifiers == NULL)
return FALSE;
for (i = 0; i < xwl_format->num_modifiers; i++)
(*modifiers)[i] = xwl_format->modifiers[i];
*num_modifiers = xwl_format->num_modifiers;
return TRUE;
}
static Bool
xwl_get_modifiers_for_device(struct xwl_dmabuf_feedback *feedback, drmDevice *device,
uint32_t format, uint32_t *num_modifiers,
uint64_t **modifiers,
Bool *supports_scanout)
{
/* Now try to find a matching set of tranches for the window's device */
for (int i = 0; i < feedback->dev_formats_len; i++) {
struct xwl_device_formats *dev_formats = &feedback->dev_formats[i];
if (drmDevicesEqual(dev_formats->drm_dev, device) &&
xwl_get_modifiers_for_format(dev_formats->formats, dev_formats->num_formats,
format, num_modifiers, modifiers)) {
if (supports_scanout)
*supports_scanout = !!dev_formats->supports_scanout;
return TRUE;
}
}
return FALSE;
}
Bool
xwl_glamor_get_modifiers(ScreenPtr screen, uint32_t format,
uint32_t *num_modifiers, uint64_t **modifiers)
{
struct xwl_screen *xwl_screen = xwl_screen_get(screen);
drmDevice *main_dev;
/* Explicitly zero the count as the caller may ignore the return value */
*num_modifiers = 0;
*modifiers = NULL;
if (!xwl_screen->dmabuf)
return FALSE;
if (xwl_screen->dmabuf_protocol_version >= 4) {
main_dev = xwl_screen_get_main_dev(xwl_screen);
return xwl_get_modifiers_for_device(&xwl_screen->default_feedback, main_dev,
format, num_modifiers, modifiers,
NULL);
} else {
return xwl_get_modifiers_for_format(xwl_screen->formats, xwl_screen->num_formats,
format, num_modifiers, modifiers);
}
}
Bool
xwl_glamor_get_drawable_modifiers_and_scanout(DrawablePtr drawable,
uint32_t format,
uint32_t *num_modifiers,
uint64_t **modifiers,
Bool *supports_scanout)
{
struct xwl_screen *xwl_screen = xwl_screen_get(drawable->pScreen);
struct xwl_window *xwl_window;
drmDevice *main_dev;
*num_modifiers = 0;
*modifiers = NULL;
if (supports_scanout)
*supports_scanout = FALSE;
/* We can only return per-drawable modifiers if the compositor supports feedback */
if (xwl_screen->dmabuf_protocol_version < 4)
return TRUE;
if (drawable->type != DRAWABLE_WINDOW || !xwl_screen->dmabuf)
return FALSE;
xwl_window = xwl_window_from_window((WindowPtr)drawable);
/* couldn't find drawable for window */
if (!xwl_window)
return FALSE;
main_dev = xwl_screen_get_main_dev(xwl_screen);
return xwl_get_modifiers_for_device(&xwl_window->feedback, main_dev,
format, num_modifiers, modifiers,
supports_scanout);
}
Bool
xwl_glamor_get_drawable_modifiers(DrawablePtr drawable, uint32_t format,
uint32_t *num_modifiers, uint64_t **modifiers)
{
return xwl_glamor_get_drawable_modifiers_and_scanout(drawable,
format, num_modifiers,
modifiers, NULL);
}
static void
xwl_dmabuf_handle_format(void *data, struct zwp_linux_dmabuf_v1 *dmabuf,
uint32_t format)
{
}
static void
xwl_add_format_and_mod_to_list(struct xwl_format **formats,
uint32_t *num_formats,
uint32_t format,
uint64_t modifier)
{
struct xwl_format *xwl_format = NULL;
int i;
for (i = 0; i < *num_formats; i++) {
if ((*formats)[i].format == format) {
xwl_format = &(*formats)[i];
break;
}
}
if (xwl_format == NULL) {
(*num_formats)++;
*formats = xnfrealloc(*formats, *num_formats * sizeof(*xwl_format));
xwl_format = &(*formats)[*num_formats - 1];
xwl_format->format = format;
xwl_format->num_modifiers = 0;
xwl_format->modifiers = NULL;
}
for (i = 0; i < xwl_format->num_modifiers; i++) {
/* don't add it if the modifier already exists */
if (xwl_format->modifiers[i] == modifier)
return;
}
xwl_format->num_modifiers++;
xwl_format->modifiers = xnfrealloc(xwl_format->modifiers,
xwl_format->num_modifiers * sizeof(uint64_t));
xwl_format->modifiers[xwl_format->num_modifiers - 1] = modifier;
}
static void
xwl_dmabuf_handle_modifier(void *data, struct zwp_linux_dmabuf_v1 *dmabuf,
uint32_t format, uint32_t modifier_hi,
uint32_t modifier_lo)
{
struct xwl_screen *xwl_screen = data;
xwl_add_format_and_mod_to_list(&xwl_screen->formats, &xwl_screen->num_formats,
format,
((uint64_t)modifier_hi << 32 | (uint64_t)modifier_lo));
}
static const struct zwp_linux_dmabuf_v1_listener xwl_dmabuf_listener = {
.format = xwl_dmabuf_handle_format,
.modifier = xwl_dmabuf_handle_modifier
};
/*
* We need to check if the compositor is resending all of the tranche
* information. Each tranche event will call this method to see
* if the existing format info should be cleared before refilling.
*/
static void
xwl_check_reset_tranche_info(struct xwl_dmabuf_feedback *xwl_feedback)
{
if (!xwl_feedback->feedback_done)
return;
xwl_feedback->feedback_done = false;
xwl_dmabuf_feedback_clear_dev_formats(xwl_feedback);
}
static void
xwl_dmabuf_feedback_main_device(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
struct wl_array *dev)
{
struct xwl_dmabuf_feedback *xwl_feedback = data;
dev_t devid;
xwl_check_reset_tranche_info(xwl_feedback);
assert(dev->size == sizeof(dev_t));
memcpy(&devid, dev->data, sizeof(dev_t));
if (drmGetDeviceFromDevId(devid, 0, &xwl_feedback->main_dev) != 0)
ErrorF("linux_dmabuf_feedback.main_device: Failed to fetch DRM device\n");
}
static void
xwl_dmabuf_feedback_tranche_target_device(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
struct wl_array *dev)
{
struct xwl_dmabuf_feedback *xwl_feedback = data;
dev_t devid;
xwl_check_reset_tranche_info(xwl_feedback);
assert(dev->size == sizeof(dev_t));
memcpy(&devid, dev->data, sizeof(dev_t));
if (drmGetDeviceFromDevId(devid, 0, &xwl_feedback->tmp_tranche.drm_dev) != 0)
ErrorF("linux_dmabuf_feedback.tranche_target_device: Failed to fetch DRM device\n");
}
static void
xwl_dmabuf_feedback_tranche_flags(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
uint32_t flags)
{
struct xwl_dmabuf_feedback *xwl_feedback = data;
xwl_check_reset_tranche_info(xwl_feedback);
if (flags & ZWP_LINUX_DMABUF_FEEDBACK_V1_TRANCHE_FLAGS_SCANOUT)
xwl_feedback->tmp_tranche.supports_scanout = true;
}
static void
xwl_dmabuf_feedback_tranche_formats(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
struct wl_array *indices)
{
struct xwl_dmabuf_feedback *xwl_feedback = data;
struct xwl_device_formats *tranche = &xwl_feedback->tmp_tranche;
uint16_t *index;
xwl_check_reset_tranche_info(xwl_feedback);
wl_array_for_each(index, indices) {
if (*index >= xwl_feedback->format_table.len) {
ErrorF("linux_dmabuf_feedback.tranche_formats: Index given to us by the compositor"
" is too large to fit in the format table\n");
continue;
}
/* Look up this format/mod in the format table */
struct xwl_format_table_entry *entry = &xwl_feedback->format_table.entry[*index];
/* Add it to the in-progress tranche */
xwl_add_format_and_mod_to_list(&tranche->formats, &tranche->num_formats,
entry->format,
entry->modifier);
}
}
static void
xwl_append_to_tranche(struct xwl_device_formats *dst, struct xwl_device_formats *src)
{
struct xwl_format *format;
for (int i = 0; i < src->num_formats; i++) {
format = &src->formats[i];
for (int j = 0; j < format->num_modifiers; j++)
xwl_add_format_and_mod_to_list(&dst->formats, &dst->num_formats,
format->format,
format->modifiers[j]);
}
}
static void
xwl_dmabuf_feedback_tranche_done(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback)
{
struct xwl_dmabuf_feedback *xwl_feedback = data;
struct xwl_device_formats *tranche;
int appended = false;
/*
* No need to call xwl_check_reset_tranche_info, the other events should have been
* triggered first
*/
if (xwl_feedback->tmp_tranche.drm_dev == NULL) {
xwl_device_formats_destroy(&xwl_feedback->tmp_tranche);
goto out;
}
/*
* First check if there is an existing tranche for this device+flags combo. We
* will combine it with this tranche, since we can only send one modifier list
* in DRI3 but the compositor may report multiple tranches per device (KDE
* does this)
*/
for (int i = 0; i < xwl_feedback->dev_formats_len; i++) {
tranche = &xwl_feedback->dev_formats[i];
if (tranche->drm_dev == xwl_feedback->tmp_tranche.drm_dev &&
tranche->supports_scanout == xwl_feedback->tmp_tranche.supports_scanout) {
appended = true;
/* Add all format/mods to this tranche */
xwl_append_to_tranche(tranche, &xwl_feedback->tmp_tranche);
/* Now free our temp tranche's allocations */
xwl_device_formats_destroy(&xwl_feedback->tmp_tranche);
break;
}
}
if (!appended) {
xwl_feedback->dev_formats_len++;
xwl_feedback->dev_formats = xnfrealloc(xwl_feedback->dev_formats,
sizeof(struct xwl_device_formats) *
xwl_feedback->dev_formats_len);
/* copy the temporary tranche into the official array */
memcpy(&xwl_feedback->dev_formats[xwl_feedback->dev_formats_len - 1],
&xwl_feedback->tmp_tranche,
sizeof(struct xwl_device_formats));
}
out:
/* reset the tranche */
memset(&xwl_feedback->tmp_tranche, 0, sizeof(struct xwl_device_formats));
}
static void
xwl_dmabuf_feedback_done(void *data, struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback)
{
struct xwl_dmabuf_feedback *xwl_feedback = data;
xwl_feedback->feedback_done = true;
xwl_feedback->unprocessed_feedback_pending = true;
}
static void
xwl_dmabuf_feedback_format_table(void *data,
struct zwp_linux_dmabuf_feedback_v1 *zwp_linux_dmabuf_feedback_v1,
int32_t fd, uint32_t size)
{
struct xwl_dmabuf_feedback *xwl_feedback = data;
/* Unmap the old table */
if (xwl_feedback->format_table.entry) {
munmap(xwl_feedback->format_table.entry,
xwl_feedback->format_table.len * sizeof(struct xwl_format_table_entry));
}
assert(size % sizeof(struct xwl_format_table_entry) == 0);
xwl_feedback->format_table.len = size / sizeof(struct xwl_format_table_entry);
xwl_feedback->format_table.entry = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
close(fd);
if (xwl_feedback->format_table.entry == MAP_FAILED) {
ErrorF("linux_dmabuf_feedback.format_table: Could not map the format"
" table: Compositor bug or out of resources\n");
xwl_feedback->format_table.len = 0;
}
}
static const struct zwp_linux_dmabuf_feedback_v1_listener xwl_dmabuf_feedback_listener = {
.done = xwl_dmabuf_feedback_done,
.format_table = xwl_dmabuf_feedback_format_table,
.main_device = xwl_dmabuf_feedback_main_device,
.tranche_done = xwl_dmabuf_feedback_tranche_done,
.tranche_target_device = xwl_dmabuf_feedback_tranche_target_device,
.tranche_formats = xwl_dmabuf_feedback_tranche_formats,
.tranche_flags = xwl_dmabuf_feedback_tranche_flags,
};
Bool
xwl_screen_set_dmabuf_interface(struct xwl_screen *xwl_screen,
uint32_t id, uint32_t version)
{
/* We either support versions 3 or 4. 4 is needed for dmabuf feedback */
int supported_version = version >= 4 ? 4 : 3;
if (version < 3)
return FALSE;
xwl_screen->dmabuf =
wl_registry_bind(xwl_screen->registry, id, &zwp_linux_dmabuf_v1_interface, supported_version);
xwl_screen->dmabuf_protocol_version = supported_version;
zwp_linux_dmabuf_v1_add_listener(xwl_screen->dmabuf, &xwl_dmabuf_listener, xwl_screen);
/* If the compositor supports it, request the default feedback hints */
if (version >= 4) {
xwl_screen->default_feedback.dmabuf_feedback =
zwp_linux_dmabuf_v1_get_default_feedback(xwl_screen->dmabuf);
if (!xwl_screen->default_feedback.dmabuf_feedback)
return FALSE;
zwp_linux_dmabuf_feedback_v1_add_listener(xwl_screen->default_feedback.dmabuf_feedback,
&xwl_dmabuf_feedback_listener,
&xwl_screen->default_feedback);
}
return TRUE;
}
static void
xwl_window_dmabuf_feedback_main_device(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
struct wl_array *dev)
{
struct xwl_window *xwl_window = data;
xwl_dmabuf_feedback_main_device(&xwl_window->feedback, dmabuf_feedback, dev);
}
static void
xwl_window_dmabuf_feedback_tranche_target_device(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
struct wl_array *dev)
{
struct xwl_window *xwl_window = data;
xwl_dmabuf_feedback_tranche_target_device(&xwl_window->feedback, dmabuf_feedback, dev);
}
static void
xwl_window_dmabuf_feedback_tranche_flags(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
uint32_t flags)
{
struct xwl_window *xwl_window = data;
xwl_dmabuf_feedback_tranche_flags(&xwl_window->feedback, dmabuf_feedback, flags);
}
static void
xwl_window_dmabuf_feedback_tranche_formats(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
struct wl_array *indices)
{
struct xwl_window *xwl_window = data;
xwl_dmabuf_feedback_tranche_formats(&xwl_window->feedback, dmabuf_feedback, indices);
}
static void
xwl_window_dmabuf_feedback_tranche_done(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback)
{
struct xwl_window *xwl_window = data;
xwl_dmabuf_feedback_tranche_done(&xwl_window->feedback, dmabuf_feedback);
}
static void
xwl_window_dmabuf_feedback_done(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback)
{
struct xwl_window *xwl_window = data;
uint32_t format = wl_drm_format_for_depth(xwl_window->window->drawable.depth);
xwl_dmabuf_feedback_done(&xwl_window->feedback, dmabuf_feedback);
xwl_window->has_implicit_scanout_support =
xwl_feedback_is_modifier_supported(&xwl_window->feedback, format,
DRM_FORMAT_MOD_INVALID, TRUE);
DebugF("XWAYLAND: Window 0x%x can%s get implicit scanout support\n",
xwl_window->window->drawable.id,
xwl_window->has_implicit_scanout_support ? "" : "not");
/* If the linux-dmabuf v4 per-surface feedback changed, recycle the
* window buffers so that they get re-created with appropriate parameters.
*/
xwl_window_buffers_recycle(xwl_window);
}
static void
xwl_window_dmabuf_feedback_format_table(void *data,
struct zwp_linux_dmabuf_feedback_v1 *dmabuf_feedback,
int32_t fd, uint32_t size)
{
struct xwl_window *xwl_window = data;
xwl_dmabuf_feedback_format_table(&xwl_window->feedback, dmabuf_feedback, fd, size);
}
static const struct zwp_linux_dmabuf_feedback_v1_listener xwl_window_dmabuf_feedback_listener = {
.done = xwl_window_dmabuf_feedback_done,
.format_table = xwl_window_dmabuf_feedback_format_table,
.main_device = xwl_window_dmabuf_feedback_main_device,
.tranche_done = xwl_window_dmabuf_feedback_tranche_done,
.tranche_target_device = xwl_window_dmabuf_feedback_tranche_target_device,
.tranche_formats = xwl_window_dmabuf_feedback_tranche_formats,
.tranche_flags = xwl_window_dmabuf_feedback_tranche_flags,
};
Bool
xwl_dmabuf_setup_feedback_for_window(struct xwl_window *xwl_window)
{
struct xwl_screen *xwl_screen = xwl_window->xwl_screen;
xwl_window->feedback.dmabuf_feedback =
zwp_linux_dmabuf_v1_get_surface_feedback(xwl_screen->dmabuf, xwl_window->surface);
if (!xwl_window->feedback.dmabuf_feedback)
return FALSE;
zwp_linux_dmabuf_feedback_v1_add_listener(xwl_window->feedback.dmabuf_feedback,
&xwl_window_dmabuf_feedback_listener,
xwl_window);
return TRUE;
}
void
xwl_glamor_init_wl_registry(struct xwl_screen *xwl_screen,
struct wl_registry *registry,
uint32_t id, const char *interface,
uint32_t version)
{
if (xwl_screen->gbm_backend.is_available &&
xwl_screen->gbm_backend.init_wl_registry(xwl_screen,
registry,
id,
interface,
version)) {
/* no-op */
} else if (xwl_screen->eglstream_backend.is_available &&
xwl_screen->eglstream_backend.init_wl_registry(xwl_screen,
registry,
id,
interface,
version)) {
/* no-op */
}
}
Bool
xwl_glamor_has_wl_interfaces(struct xwl_screen *xwl_screen,
struct xwl_egl_backend *xwl_egl_backend)
{
return xwl_egl_backend->has_wl_interfaces(xwl_screen);
}
struct wl_buffer *
xwl_glamor_pixmap_get_wl_buffer(PixmapPtr pixmap)
{
struct xwl_screen *xwl_screen = xwl_screen_get(pixmap->drawable.pScreen);
if (xwl_screen->egl_backend->get_wl_buffer_for_pixmap)
return xwl_screen->egl_backend->get_wl_buffer_for_pixmap(pixmap);
return NULL;
}
Bool
xwayland: Add glamor egl_backend for EGLStreams 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>
2018-04-20 14:38:05 -04:00
xwl_glamor_post_damage(struct xwl_window *xwl_window,
PixmapPtr pixmap, RegionPtr region)
{
struct xwl_screen *xwl_screen = xwl_window->xwl_screen;
if (xwl_screen->egl_backend->post_damage)
return xwl_screen->egl_backend->post_damage(xwl_window, pixmap, region);
return TRUE;
xwayland: Add glamor egl_backend for EGLStreams 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>
2018-04-20 14:38:05 -04:00
}
Bool
xwl_glamor_allow_commits(struct xwl_window *xwl_window)
{
struct xwl_screen *xwl_screen = xwl_window->xwl_screen;
if (xwl_screen->egl_backend->allow_commits)
return xwl_screen->egl_backend->allow_commits(xwl_window);
xwayland: Add glamor egl_backend for EGLStreams 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>
2018-04-20 14:38:05 -04:00
else
return TRUE;
}
static void
xwl_avoid_implicit_redirect(WindowPtr window)
{
ScreenPtr screen = window->drawable.pScreen;
WindowOptPtr parent_optional;
VisualPtr parent_visual = NULL;
VisualPtr window_visual = NULL;
DepthPtr depth32 = NULL;
int i;
if (!window->optional)
return;
parent_optional = FindWindowWithOptional(window)->optional;
if (window->optional == parent_optional ||
window->optional->visual == parent_optional->visual ||
CompositeIsImplicitRedirectException(screen, parent_optional->visual,
window->optional->visual))
return;
for (i = 0; i < screen->numDepths; i++) {
if (screen->allowedDepths[i].depth == 32) {
depth32 = &screen->allowedDepths[i];
break;
}
}
if (!depth32)
return;
for (i = 0; i < depth32->numVids; i++) {
XID argb_vid = depth32->vids[i];
if (argb_vid != parent_optional->visual)
continue;
if (!compIsAlternateVisual(screen, argb_vid))
break;
for (i = 0; i < screen->numVisuals; i++) {
if (screen->visuals[i].vid == argb_vid) {
parent_visual = &screen->visuals[i];
break;
}
}
}
if (!parent_visual)
return;
for (i = 0; i < screen->numVisuals; i++) {
if (screen->visuals[i].vid == window->optional->visual) {
window_visual = &screen->visuals[i];
break;
}
}
if ((window_visual->class != TrueColor &&
window_visual->class != DirectColor) ||
window_visual->redMask != parent_visual->redMask ||
window_visual->greenMask != parent_visual->greenMask ||
window_visual->blueMask != parent_visual->blueMask ||
window_visual->offsetRed != parent_visual->offsetRed ||
window_visual->offsetGreen != parent_visual->offsetGreen ||
window_visual->offsetBlue != parent_visual->offsetBlue)
return;
CompositeRegisterImplicitRedirectionException(screen, parent_visual->vid, window_visual->vid);
}
static Bool
xwl_glamor_create_window(WindowPtr window)
{
ScreenPtr screen = window->drawable.pScreen;
struct xwl_screen *xwl_screen = xwl_screen_get(screen);
Bool ret;
if (window->parent)
xwl_avoid_implicit_redirect(window);
screen->CreateWindow = xwl_screen->CreateWindow;
ret = (*screen->CreateWindow) (window);
xwl_screen->CreateWindow = screen->CreateWindow;
screen->CreateWindow = xwl_glamor_create_window;
return ret;
}
static void
xwl_glamor_reparent_window(WindowPtr window, WindowPtr old_parent)
{
ScreenPtr screen = window->drawable.pScreen;
struct xwl_screen *xwl_screen = xwl_screen_get(screen);
xwl_avoid_implicit_redirect(window);
screen->ReparentWindow = xwl_screen->ReparentWindow;
(*screen->ReparentWindow) (window, old_parent);
xwl_screen->ReparentWindow = screen->ReparentWindow;
screen->ReparentWindow = xwl_glamor_reparent_window;
}
static Bool
xwl_glamor_create_screen_resources(ScreenPtr screen)
{
struct xwl_screen *xwl_screen = xwl_screen_get(screen);
int ret;
screen->CreateScreenResources = xwl_screen->CreateScreenResources;
ret = (*screen->CreateScreenResources) (screen);
xwl_screen->CreateScreenResources = screen->CreateScreenResources;
screen->CreateScreenResources = xwl_glamor_create_screen_resources;
if (!ret)
return ret;
xwl_screen->CreateWindow = screen->CreateWindow;
screen->CreateWindow = xwl_glamor_create_window;
xwl_screen->ReparentWindow = screen->ReparentWindow;
screen->ReparentWindow = xwl_glamor_reparent_window;
if (xwl_screen->rootless) {
screen->devPrivate =
fbCreatePixmap(screen, 0, 0, screen->rootDepth, 0);
}
else {
screen->devPrivate = screen->CreatePixmap(
screen, screen->width, screen->height, screen->rootDepth,
CREATE_PIXMAP_USAGE_BACKING_PIXMAP);
}
SetRootClip(screen, xwl_screen->root_clip_mode);
return screen->devPrivate != NULL;
}
int
glamor_egl_fd_name_from_pixmap(ScreenPtr screen,
PixmapPtr pixmap,
CARD16 *stride, CARD32 *size)
{
return 0;
}
Bool
xwl_glamor_needs_buffer_flush(struct xwl_screen *xwl_screen)
{
if (!xwl_screen->glamor || !xwl_screen->egl_backend)
return FALSE;
return (xwl_screen->egl_backend->backend_flags &
XWL_EGL_BACKEND_NEEDS_BUFFER_FLUSH);
}
Bool
xwl_glamor_needs_n_buffering(struct xwl_screen *xwl_screen)
{
/* wl_shm benefits from n-buffering */
if (!xwl_screen->glamor || !xwl_screen->egl_backend)
return TRUE;
return (xwl_screen->egl_backend->backend_flags &
XWL_EGL_BACKEND_NEEDS_N_BUFFERING);
}
PixmapPtr
xwl_glamor_create_pixmap_for_window(struct xwl_window *xwl_window)
{
struct xwl_screen *xwl_screen = xwl_window->xwl_screen;
if (!xwl_screen->glamor || !xwl_screen->egl_backend)
return NullPixmap;
if (xwl_screen->egl_backend->create_pixmap_for_window)
return xwl_screen->egl_backend->create_pixmap_for_window(xwl_window);
else
return NullPixmap;
}
void
xwl_glamor_init_backends(struct xwl_screen *xwl_screen, Bool use_eglstream)
{
#ifdef GLAMOR_HAS_GBM
xwl_glamor_init_gbm(xwl_screen);
if (!xwl_screen->gbm_backend.is_available && !use_eglstream)
ErrorF("Xwayland glamor: GBM backend (default) is not available\n");
#endif
#ifdef XWL_HAS_EGLSTREAM
xwl_glamor_init_eglstream(xwl_screen);
if (!xwl_screen->eglstream_backend.is_available && use_eglstream)
ErrorF("Xwayland glamor: EGLStream backend requested but not available\n");
#endif
}
static Bool
xwl_glamor_select_gbm_backend(struct xwl_screen *xwl_screen)
{
#ifdef GLAMOR_HAS_GBM
if (xwl_screen->gbm_backend.is_available &&
xwl_glamor_has_wl_interfaces(xwl_screen, &xwl_screen->gbm_backend)) {
xwl_screen->egl_backend = &xwl_screen->gbm_backend;
LogMessageVerb(X_INFO, 3, "glamor: Using GBM backend\n");
return TRUE;
}
else
LogMessageVerb(X_INFO, 3,
"Missing Wayland requirements for glamor GBM backend\n");
#endif
return FALSE;
}
static Bool
xwl_glamor_select_eglstream_backend(struct xwl_screen *xwl_screen)
{
#ifdef XWL_HAS_EGLSTREAM
if (xwl_screen->eglstream_backend.is_available &&
xwl_glamor_has_wl_interfaces(xwl_screen, &xwl_screen->eglstream_backend)) {
xwl_screen->egl_backend = &xwl_screen->eglstream_backend;
LogMessageVerb(X_INFO, 3, "glamor: Using EGLStream backend\n");
return TRUE;
}
else
LogMessageVerb(X_INFO, 3,
"Missing Wayland requirements for glamor EGLStream backend\n");
#endif
return FALSE;
}
void
xwl_glamor_select_backend(struct xwl_screen *xwl_screen, Bool use_eglstream)
{
if (!xwl_glamor_select_eglstream_backend(xwl_screen)) {
if (!use_eglstream)
xwl_glamor_select_gbm_backend(xwl_screen);
}
}
Bool
xwl_glamor_init(struct xwl_screen *xwl_screen)
{
ScreenPtr screen = xwl_screen->screen;
const char *no_glamor_env;
no_glamor_env = getenv("XWAYLAND_NO_GLAMOR");
if (no_glamor_env && *no_glamor_env != '0') {
ErrorF("Disabling glamor and dri3 support, XWAYLAND_NO_GLAMOR is set\n");
return FALSE;
}
if (!xwl_screen->egl_backend->init_egl(xwl_screen)) {
ErrorF("EGL setup failed, disabling glamor\n");
return FALSE;
}
if (!glamor_init(xwl_screen->screen, GLAMOR_USE_EGL_SCREEN)) {
ErrorF("Failed to initialize glamor\n");
return FALSE;
}
if (!xwl_screen->egl_backend->init_screen(xwl_screen)) {
ErrorF("EGL backend init_screen() failed, disabling glamor\n");
return FALSE;
}
xwl_screen->CreateScreenResources = screen->CreateScreenResources;
screen->CreateScreenResources = xwl_glamor_create_screen_resources;
#ifdef XV
if (!xwl_glamor_xv_init(screen))
ErrorF("Failed to initialize glamor Xv extension\n");
#endif
#ifdef GLXEXT
GlxPushProvider(&glamor_provider);
#endif
return TRUE;
}