[PATCH 00/20] Various vhost-user-gpu & UI fixes

[PATCH 00/20] Various vhost-user-gpu & UI fixes

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Hi,

Here is a collection of fixes and improvements mostly around vhost-user-gpu. In
combination with spice, they work best with the previously sent fixes ("[PATCH
0/2] Fix (non-vga) QXL async screendump" & "[PATCH v2] spice: delay starting
until display are initialized"). I also use virgl from git, which seems to work
better. You also need mesa with the recent fix !8663 from last week if you test
this.

The last patch is a simple RFC avocado test to have the basics under some
control and avoid simple regressions.

I have further refactoring work pending to separate the DisplayChangeListener
from the GL context. This should allow to run a GL-less qemu with Spice client &
vhost-user-gpu/virgl for example.

Marc-André Lureau (20):
  vhost-user-gpu: check backend for EDID support
  vhost-user-gpu: handle vhost-user-gpu features in a callback
  vhost-user-gpu: use an extandable state enum for commands
  vhost-user-gpu: handle display-info in a callback
  ui: remove extra #ifdef CONFIG_OPENGL
  ui: remove gl_ctx_get_current
  ui: add gd_gl_area_scanout_disable
  ui: annotate DCLOps callback requirements
  ui: remove console_has_gl_dmabuf()
  vhost-user-gpu: add a configuration flag for dmabuf usage
  ui: add an optional get_flags callback to GraphicHwOps
  ui: add a DCLOps callback to check dmabuf support
  ui: check hw requirements during DCL registration
  ui: add qemu_egl_has_dmabuf helper
  ui: check gtk-egl dmabuf support
  ui: add egl dmabuf import to gtkglarea
  virtio-gpu: avoid re-entering cmdq processing
  display/ui: add a callback to indicate GL state is flushed
  chardev: check if the chardev is registered for yanking
  RFC: tests: add some virtio-gpu & vhost-user-gpu acceptance test

 contrib/vhost-user-gpu/vugpu.h          |  10 +-
 include/hw/virtio/virtio-gpu.h          |   6 +-
 include/ui/console.h                    |  35 +++++-
 include/ui/egl-context.h                |   1 -
 include/ui/egl-helpers.h                |   1 +
 include/ui/gtk.h                        |   4 +
 include/ui/sdl2.h                       |   1 -
 chardev/char-socket.c                   |  53 +++++----
 contrib/vhost-user-gpu/vhost-user-gpu.c |  77 ++++++++----
 contrib/vhost-user-gpu/virgl.c          |   2 +-
 hw/display/vhost-user-gpu.c             |  17 +--
 hw/display/virtio-gpu-base.c            |  31 ++++-
 hw/display/virtio-gpu.c                 |   9 +-
 hw/display/virtio-vga.c                 |  20 ++++
 hw/vfio/display.c                       |   6 +
 ui/console.c                            |  61 +++++++---
 ui/egl-context.c                        |   5 -
 ui/egl-headless.c                       |   1 -
 ui/egl-helpers.c                        |   9 ++
 ui/gtk-egl.c                            |   3 +
 ui/gtk-gl-area.c                        |  28 +++++
 ui/gtk.c                                |  43 +++++--
 ui/sdl2-gl.c                            |  10 +-
 ui/sdl2.c                               |   1 -
 ui/spice-display.c                      |   2 +-
 tests/acceptance/virtio-gpu.py          | 150 ++++++++++++++++++++++++
 26 files changed, 484 insertions(+), 102 deletions(-)
 create mode 100644 tests/acceptance/virtio-gpu.py

-- 
2.29.0

[PATCH 01/20] vhost-user-gpu: check backend for EDID support

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

EDID has been enabled by default, but the backend may not implement
it (such as the contrib backend). This results in extra warnings and
potentially other issues in the guest.

The option shouldn't probably have been added to VIRTIO_GPU_BASE, but
it's a bit too late now, report an error and disable EDID when it's
not available.

Fixes: 0a7196625 ("edid: flip the default to enabled")

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 hw/display/vhost-user-gpu.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/hw/display/vhost-user-gpu.c b/hw/display/vhost-user-gpu.c
index 51f1747c4a..55b0ed15f2 100644
--- a/hw/display/vhost-user-gpu.c
+++ b/hw/display/vhost-user-gpu.c
@@ -555,6 +555,12 @@ vhost_user_gpu_device_realize(DeviceState *qdev, Error **errp)
     if (virtio_has_feature(g->vhost->dev.features, VIRTIO_GPU_F_VIRGL)) {
         g->parent_obj.conf.flags |= 1 << VIRTIO_GPU_FLAG_VIRGL_ENABLED;
     }
+    if (virtio_has_feature(g->vhost->dev.features, VIRTIO_GPU_F_EDID)) {
+        g->parent_obj.conf.flags |= 1 << VIRTIO_GPU_FLAG_EDID_ENABLED;
+    } else {
+        error_report("EDID requested but the backend doesn't support it.");
+        g->parent_obj.conf.flags &= ~(1 << VIRTIO_GPU_FLAG_EDID_ENABLED);
+    }
 
     if (!virtio_gpu_base_device_realize(qdev, NULL, NULL, errp)) {
         return;
-- 
2.29.0

[PATCH 02/20] vhost-user-gpu: handle vhost-user-gpu features in a callback

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Fixes a deadlock where the backend calls QEMU, while QEMU also calls the
backend simultaneously, both ends waiting for each other.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 contrib/vhost-user-gpu/vugpu.h          |  2 +-
 contrib/vhost-user-gpu/vhost-user-gpu.c | 37 ++++++++++++++++++-------
 2 files changed, 28 insertions(+), 11 deletions(-)

diff --git a/contrib/vhost-user-gpu/vugpu.h b/contrib/vhost-user-gpu/vugpu.h
index ad664c4df8..86f3ac86aa 100644
--- a/contrib/vhost-user-gpu/vugpu.h
+++ b/contrib/vhost-user-gpu/vugpu.h
@@ -118,7 +118,7 @@ typedef struct VuGpu {
     int sock_fd;
     int drm_rnode_fd;
     GSource *renderer_source;
-    guint wait_ok;
+    guint wait_in;
 
     bool virgl;
     bool virgl_inited;
diff --git a/contrib/vhost-user-gpu/vhost-user-gpu.c b/contrib/vhost-user-gpu/vhost-user-gpu.c
index f445ef28ec..85c16404fb 100644
--- a/contrib/vhost-user-gpu/vhost-user-gpu.c
+++ b/contrib/vhost-user-gpu/vhost-user-gpu.c
@@ -124,7 +124,7 @@ source_wait_cb(gint fd, GIOCondition condition, gpointer user_data)
     }
 
     /* resume */
-    g->wait_ok = 0;
+    g->wait_in = 0;
     vg_handle_ctrl(&g->dev.parent, 0);
 
     return G_SOURCE_REMOVE;
@@ -133,8 +133,8 @@ source_wait_cb(gint fd, GIOCondition condition, gpointer user_data)
 void
 vg_wait_ok(VuGpu *g)
 {
-    assert(g->wait_ok == 0);
-    g->wait_ok = g_unix_fd_add(g->sock_fd, G_IO_IN | G_IO_HUP,
+    assert(g->wait_in == 0);
+    g->wait_in = g_unix_fd_add(g->sock_fd, G_IO_IN | G_IO_HUP,
                                source_wait_cb, g);
 }
 
@@ -270,7 +270,7 @@ vg_get_display_info(VuGpu *vg, struct virtio_gpu_ctrl_command *cmd)
         .size = 0,
     };
 
-    assert(vg->wait_ok == 0);
+    assert(vg->wait_in == 0);
 
     vg_send_msg(vg, &msg, -1);
     if (!vg_recv_msg(vg, msg.request, sizeof(dpy_info), &dpy_info)) {
@@ -815,7 +815,7 @@ vg_handle_ctrl(VuDev *dev, int qidx)
     size_t len;
 
     for (;;) {
-        if (vg->wait_ok != 0) {
+        if (vg->wait_in != 0) {
             return;
         }
 
@@ -969,18 +969,17 @@ vg_queue_set_started(VuDev *dev, int qidx, bool started)
     }
 }
 
-static void
-set_gpu_protocol_features(VuGpu *g)
+static gboolean
+protocol_features_cb(gint fd, GIOCondition condition, gpointer user_data)
 {
+    VuGpu *g = user_data;
     uint64_t u64;
     VhostUserGpuMsg msg = {
         .request = VHOST_USER_GPU_GET_PROTOCOL_FEATURES
     };
 
-    assert(g->wait_ok == 0);
-    vg_send_msg(g, &msg, -1);
     if (!vg_recv_msg(g, msg.request, sizeof(u64), &u64)) {
-        return;
+        return G_SOURCE_CONTINUE;
     }
 
     msg = (VhostUserGpuMsg) {
@@ -989,6 +988,24 @@ set_gpu_protocol_features(VuGpu *g)
         .payload.u64 = 0
     };
     vg_send_msg(g, &msg, -1);
+
+    g->wait_in = 0;
+    vg_handle_ctrl(&g->dev.parent, 0);
+
+    return G_SOURCE_REMOVE;
+}
+
+static void
+set_gpu_protocol_features(VuGpu *g)
+{
+    VhostUserGpuMsg msg = {
+        .request = VHOST_USER_GPU_GET_PROTOCOL_FEATURES
+    };
+
+    vg_send_msg(g, &msg, -1);
+    assert(g->wait_in == 0);
+    g->wait_in = g_unix_fd_add(g->sock_fd, G_IO_IN | G_IO_HUP,
+                               protocol_features_cb, g);
 }
 
 static int
-- 
2.29.0

[PATCH 03/20] vhost-user-gpu: use an extandable state enum for commands

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Introduce a pending state for commands which aren't finished yet, but
are being handled. See following patch.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 contrib/vhost-user-gpu/vugpu.h          | 8 +++++++-
 contrib/vhost-user-gpu/vhost-user-gpu.c | 8 ++++----
 contrib/vhost-user-gpu/virgl.c          | 2 +-
 3 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/contrib/vhost-user-gpu/vugpu.h b/contrib/vhost-user-gpu/vugpu.h
index 86f3ac86aa..04d5615812 100644
--- a/contrib/vhost-user-gpu/vugpu.h
+++ b/contrib/vhost-user-gpu/vugpu.h
@@ -129,12 +129,18 @@ typedef struct VuGpu {
     QTAILQ_HEAD(, virtio_gpu_ctrl_command) fenceq;
 } VuGpu;
 
+enum {
+    VG_CMD_STATE_NEW,
+    VG_CMD_STATE_PENDING,
+    VG_CMD_STATE_FINISHED,
+};
+
 struct virtio_gpu_ctrl_command {
     VuVirtqElement elem;
     VuVirtq *vq;
     struct virtio_gpu_ctrl_hdr cmd_hdr;
     uint32_t error;
-    bool finished;
+    int state;
     QTAILQ_ENTRY(virtio_gpu_ctrl_command) next;
 };
 
diff --git a/contrib/vhost-user-gpu/vhost-user-gpu.c b/contrib/vhost-user-gpu/vhost-user-gpu.c
index 85c16404fb..7dcc02966c 100644
--- a/contrib/vhost-user-gpu/vhost-user-gpu.c
+++ b/contrib/vhost-user-gpu/vhost-user-gpu.c
@@ -246,7 +246,7 @@ vg_ctrl_response(VuGpu *g,
     }
     vu_queue_push(&g->dev.parent, cmd->vq, &cmd->elem, s);
     vu_queue_notify(&g->dev.parent, cmd->vq);
-    cmd->finished = true;
+    cmd->state = VG_CMD_STATE_FINISHED;
 }
 
 void
@@ -800,7 +800,7 @@ vg_process_cmd(VuGpu *vg, struct virtio_gpu_ctrl_command *cmd)
         cmd->error = VIRTIO_GPU_RESP_ERR_UNSPEC;
         break;
     }
-    if (!cmd->finished) {
+    if (cmd->state == VG_CMD_STATE_NEW) {
         vg_ctrl_response_nodata(vg, cmd, cmd->error ? cmd->error :
                                 VIRTIO_GPU_RESP_OK_NODATA);
     }
@@ -825,7 +825,7 @@ vg_handle_ctrl(VuDev *dev, int qidx)
         }
         cmd->vq = vq;
         cmd->error = 0;
-        cmd->finished = false;
+        cmd->state = VG_CMD_STATE_NEW;
 
         len = iov_to_buf(cmd->elem.out_sg, cmd->elem.out_num,
                          0, &cmd->cmd_hdr, sizeof(cmd->cmd_hdr));
@@ -844,7 +844,7 @@ vg_handle_ctrl(VuDev *dev, int qidx)
             vg_process_cmd(vg, cmd);
         }
 
-        if (!cmd->finished) {
+        if (cmd->state != VG_CMD_STATE_FINISHED) {
             QTAILQ_INSERT_TAIL(&vg->fenceq, cmd, next);
             vg->inflight++;
         } else {
diff --git a/contrib/vhost-user-gpu/virgl.c b/contrib/vhost-user-gpu/virgl.c
index e647278052..8bb3c563d9 100644
--- a/contrib/vhost-user-gpu/virgl.c
+++ b/contrib/vhost-user-gpu/virgl.c
@@ -482,7 +482,7 @@ void vg_virgl_process_cmd(VuGpu *g, struct virtio_gpu_ctrl_command *cmd)
         break;
     }
 
-    if (cmd->finished) {
+    if (cmd->state != VG_CMD_STATE_NEW) {
         return;
     }
 
-- 
2.29.0

[PATCH 04/20] vhost-user-gpu: handle display-info in a callback

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Fixes a deadlock where the backend calls QEMU, while QEMU also calls the
backend simultaneously, both ends waiting for each other.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 contrib/vhost-user-gpu/vhost-user-gpu.c | 32 ++++++++++++++++++++-----
 1 file changed, 26 insertions(+), 6 deletions(-)

diff --git a/contrib/vhost-user-gpu/vhost-user-gpu.c b/contrib/vhost-user-gpu/vhost-user-gpu.c
index 7dcc02966c..810e227382 100644
--- a/contrib/vhost-user-gpu/vhost-user-gpu.c
+++ b/contrib/vhost-user-gpu/vhost-user-gpu.c
@@ -261,10 +261,32 @@ vg_ctrl_response_nodata(VuGpu *g,
     vg_ctrl_response(g, cmd, &resp, sizeof(resp));
 }
 
+
+static gboolean
+get_display_info_cb(gint fd, GIOCondition condition, gpointer user_data)
+{
+    struct virtio_gpu_resp_display_info dpy_info = { {} };
+    VuGpu *vg = user_data;
+    struct virtio_gpu_ctrl_command *cmd = QTAILQ_LAST(&vg->fenceq);
+
+    g_debug("disp info cb");
+    assert(cmd->cmd_hdr.type == VIRTIO_GPU_CMD_GET_DISPLAY_INFO);
+    if (!vg_recv_msg(vg, VHOST_USER_GPU_GET_DISPLAY_INFO, sizeof(dpy_info), &dpy_info)) {
+        return G_SOURCE_CONTINUE;
+    }
+
+    QTAILQ_REMOVE(&vg->fenceq, cmd, next);
+    vg_ctrl_response(vg, cmd, &dpy_info.hdr, sizeof(dpy_info));
+
+    vg->wait_in = 0;
+    vg_handle_ctrl(&vg->dev.parent, 0);
+
+    return G_SOURCE_REMOVE;
+}
+
 void
 vg_get_display_info(VuGpu *vg, struct virtio_gpu_ctrl_command *cmd)
 {
-    struct virtio_gpu_resp_display_info dpy_info = { {} };
     VhostUserGpuMsg msg = {
         .request = VHOST_USER_GPU_GET_DISPLAY_INFO,
         .size = 0,
@@ -273,11 +295,9 @@ vg_get_display_info(VuGpu *vg, struct virtio_gpu_ctrl_command *cmd)
     assert(vg->wait_in == 0);
 
     vg_send_msg(vg, &msg, -1);
-    if (!vg_recv_msg(vg, msg.request, sizeof(dpy_info), &dpy_info)) {
-        return;
-    }
-
-    vg_ctrl_response(vg, cmd, &dpy_info.hdr, sizeof(dpy_info));
+    vg->wait_in = g_unix_fd_add(vg->sock_fd, G_IO_IN | G_IO_HUP,
+                               get_display_info_cb, vg);
+    cmd->state = VG_CMD_STATE_PENDING;
 }
 
 static void
-- 
2.29.0

[PATCH 05/20] ui: remove extra #ifdef CONFIG_OPENGL

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Since commit 5cb69566daa8081abb82a13403dcc0fffed02007 ("gtk: remove
CONFIG_GTK_GL"), some #ifdef are redundants.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 ui/gtk.c | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)

diff --git a/ui/gtk.c b/ui/gtk.c
index 26665cd2e6..e1ee0840b3 100644
--- a/ui/gtk.c
+++ b/ui/gtk.c
@@ -625,8 +625,6 @@ static const DisplayChangeListenerOps dcl_ops = {
 
 /** DisplayState Callbacks (opengl version) **/
 
-#if defined(CONFIG_OPENGL)
-
 static const DisplayChangeListenerOps dcl_gl_area_ops = {
     .dpy_name             = "gtk-egl",
     .dpy_gfx_update       = gd_gl_area_update,
@@ -644,8 +642,6 @@ static const DisplayChangeListenerOps dcl_gl_area_ops = {
     .dpy_gl_update           = gd_gl_area_scanout_flush,
 };
 
-#endif /* CONFIG_OPENGL */
-
 static const DisplayChangeListenerOps dcl_egl_ops = {
     .dpy_name             = "gtk-egl",
     .dpy_gfx_update       = gd_egl_update,
@@ -1993,13 +1989,10 @@ static GSList *gd_vc_gfx_init(GtkDisplayState *s, VirtualConsole *vc,
 
 #if defined(CONFIG_OPENGL)
     if (display_opengl) {
-#if defined(CONFIG_OPENGL)
         if (gtk_use_gl_area) {
             vc->gfx.drawing_area = gtk_gl_area_new();
             vc->gfx.dcl.ops = &dcl_gl_area_ops;
-        } else
-#endif /* CONFIG_OPENGL */
-        {
+        } else {
             vc->gfx.drawing_area = gtk_drawing_area_new();
             /*
              * gtk_widget_set_double_buffered() was deprecated in 3.14.
-- 
2.29.0

[PATCH 06/20] ui: remove gl_ctx_get_current

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

There are no users left.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/console.h     | 2 --
 include/ui/egl-context.h | 1 -
 include/ui/sdl2.h        | 1 -
 ui/console.c             | 6 ------
 ui/egl-context.c         | 5 -----
 ui/egl-headless.c        | 1 -
 ui/gtk.c                 | 2 --
 ui/sdl2-gl.c             | 8 --------
 ui/sdl2.c                | 1 -
 ui/spice-display.c       | 1 -
 10 files changed, 28 deletions(-)

diff --git a/include/ui/console.h b/include/ui/console.h
index 7a3fc11abf..ce6c72e37c 100644
--- a/include/ui/console.h
+++ b/include/ui/console.h
@@ -201,7 +201,6 @@ typedef struct DisplayChangeListenerOps {
                                QEMUGLContext ctx);
     int (*dpy_gl_ctx_make_current)(DisplayChangeListener *dcl,
                                    QEMUGLContext ctx);
-    QEMUGLContext (*dpy_gl_ctx_get_current)(DisplayChangeListener *dcl);
 
     void (*dpy_gl_scanout_disable)(DisplayChangeListener *dcl);
     void (*dpy_gl_scanout_texture)(DisplayChangeListener *dcl,
@@ -303,7 +302,6 @@ QEMUGLContext dpy_gl_ctx_create(QemuConsole *con,
                                 QEMUGLParams *params);
 void dpy_gl_ctx_destroy(QemuConsole *con, QEMUGLContext ctx);
 int dpy_gl_ctx_make_current(QemuConsole *con, QEMUGLContext ctx);
-QEMUGLContext dpy_gl_ctx_get_current(QemuConsole *con);
 
 bool console_has_gl(QemuConsole *con);
 bool console_has_gl_dmabuf(QemuConsole *con);
diff --git a/include/ui/egl-context.h b/include/ui/egl-context.h
index f004ce11a7..9374fe41e3 100644
--- a/include/ui/egl-context.h
+++ b/include/ui/egl-context.h
@@ -9,6 +9,5 @@ QEMUGLContext qemu_egl_create_context(DisplayChangeListener *dcl,
 void qemu_egl_destroy_context(DisplayChangeListener *dcl, QEMUGLContext ctx);
 int qemu_egl_make_context_current(DisplayChangeListener *dcl,
                                   QEMUGLContext ctx);
-QEMUGLContext qemu_egl_get_current_context(DisplayChangeListener *dcl);
 
 #endif /* EGL_CONTEXT_H */
diff --git a/include/ui/sdl2.h b/include/ui/sdl2.h
index 0875b8d56b..f85c117a78 100644
--- a/include/ui/sdl2.h
+++ b/include/ui/sdl2.h
@@ -70,7 +70,6 @@ QEMUGLContext sdl2_gl_create_context(DisplayChangeListener *dcl,
 void sdl2_gl_destroy_context(DisplayChangeListener *dcl, QEMUGLContext ctx);
 int sdl2_gl_make_context_current(DisplayChangeListener *dcl,
                                  QEMUGLContext ctx);
-QEMUGLContext sdl2_gl_get_current_context(DisplayChangeListener *dcl);
 
 void sdl2_gl_scanout_disable(DisplayChangeListener *dcl);
 void sdl2_gl_scanout_texture(DisplayChangeListener *dcl,
diff --git a/ui/console.c b/ui/console.c
index d80ce7037c..c0b1a3689c 100644
--- a/ui/console.c
+++ b/ui/console.c
@@ -1803,12 +1803,6 @@ int dpy_gl_ctx_make_current(QemuConsole *con, QEMUGLContext ctx)
     return con->gl->ops->dpy_gl_ctx_make_current(con->gl, ctx);
 }
 
-QEMUGLContext dpy_gl_ctx_get_current(QemuConsole *con)
-{
-    assert(con->gl);
-    return con->gl->ops->dpy_gl_ctx_get_current(con->gl);
-}
-
 void dpy_gl_scanout_disable(QemuConsole *con)
 {
     assert(con->gl);
diff --git a/ui/egl-context.c b/ui/egl-context.c
index 4aa1cbb50c..368ffa49d8 100644
--- a/ui/egl-context.c
+++ b/ui/egl-context.c
@@ -35,8 +35,3 @@ int qemu_egl_make_context_current(DisplayChangeListener *dcl,
    return eglMakeCurrent(qemu_egl_display,
                          EGL_NO_SURFACE, EGL_NO_SURFACE, ctx);
 }
-
-QEMUGLContext qemu_egl_get_current_context(DisplayChangeListener *dcl)
-{
-    return eglGetCurrentContext();
-}
diff --git a/ui/egl-headless.c b/ui/egl-headless.c
index fe2a0d1eab..da377a74af 100644
--- a/ui/egl-headless.c
+++ b/ui/egl-headless.c
@@ -160,7 +160,6 @@ static const DisplayChangeListenerOps egl_ops = {
     .dpy_gl_ctx_create       = egl_create_context,
     .dpy_gl_ctx_destroy      = qemu_egl_destroy_context,
     .dpy_gl_ctx_make_current = qemu_egl_make_context_current,
-    .dpy_gl_ctx_get_current  = qemu_egl_get_current_context,
 
     .dpy_gl_scanout_disable  = egl_scanout_disable,
     .dpy_gl_scanout_texture  = egl_scanout_texture,
diff --git a/ui/gtk.c b/ui/gtk.c
index e1ee0840b3..a0e6b60ac4 100644
--- a/ui/gtk.c
+++ b/ui/gtk.c
@@ -637,7 +637,6 @@ static const DisplayChangeListenerOps dcl_gl_area_ops = {
     .dpy_gl_ctx_create       = gd_gl_area_create_context,
     .dpy_gl_ctx_destroy      = gd_gl_area_destroy_context,
     .dpy_gl_ctx_make_current = gd_gl_area_make_current,
-    .dpy_gl_ctx_get_current  = gd_gl_area_get_current_context,
     .dpy_gl_scanout_texture  = gd_gl_area_scanout_texture,
     .dpy_gl_update           = gd_gl_area_scanout_flush,
 };
@@ -654,7 +653,6 @@ static const DisplayChangeListenerOps dcl_egl_ops = {
     .dpy_gl_ctx_create       = gd_egl_create_context,
     .dpy_gl_ctx_destroy      = qemu_egl_destroy_context,
     .dpy_gl_ctx_make_current = gd_egl_make_current,
-    .dpy_gl_ctx_get_current  = qemu_egl_get_current_context,
     .dpy_gl_scanout_disable  = gd_egl_scanout_disable,
     .dpy_gl_scanout_texture  = gd_egl_scanout_texture,
     .dpy_gl_scanout_dmabuf   = gd_egl_scanout_dmabuf,
diff --git a/ui/sdl2-gl.c b/ui/sdl2-gl.c
index c73d273bf8..82139ba49e 100644
--- a/ui/sdl2-gl.c
+++ b/ui/sdl2-gl.c
@@ -185,14 +185,6 @@ int sdl2_gl_make_context_current(DisplayChangeListener *dcl,
     return SDL_GL_MakeCurrent(scon->real_window, sdlctx);
 }
 
-QEMUGLContext sdl2_gl_get_current_context(DisplayChangeListener *dcl)
-{
-    SDL_GLContext sdlctx;
-
-    sdlctx = SDL_GL_GetCurrentContext();
-    return (QEMUGLContext)sdlctx;
-}
-
 void sdl2_gl_scanout_disable(DisplayChangeListener *dcl)
 {
     struct sdl2_console *scon = container_of(dcl, struct sdl2_console, dcl);
diff --git a/ui/sdl2.c b/ui/sdl2.c
index 5e49709283..a203cb0239 100644
--- a/ui/sdl2.c
+++ b/ui/sdl2.c
@@ -781,7 +781,6 @@ static const DisplayChangeListenerOps dcl_gl_ops = {
     .dpy_gl_ctx_create       = sdl2_gl_create_context,
     .dpy_gl_ctx_destroy      = sdl2_gl_destroy_context,
     .dpy_gl_ctx_make_current = sdl2_gl_make_context_current,
-    .dpy_gl_ctx_get_current  = sdl2_gl_get_current_context,
     .dpy_gl_scanout_disable  = sdl2_gl_scanout_disable,
     .dpy_gl_scanout_texture  = sdl2_gl_scanout_texture,
     .dpy_gl_update           = sdl2_gl_scanout_flush,
diff --git a/ui/spice-display.c b/ui/spice-display.c
index 0178d5766d..6a5e189a67 100644
--- a/ui/spice-display.c
+++ b/ui/spice-display.c
@@ -1102,7 +1102,6 @@ static const DisplayChangeListenerOps display_listener_gl_ops = {
     .dpy_gl_ctx_create       = qemu_spice_gl_create_context,
     .dpy_gl_ctx_destroy      = qemu_egl_destroy_context,
     .dpy_gl_ctx_make_current = qemu_egl_make_context_current,
-    .dpy_gl_ctx_get_current  = qemu_egl_get_current_context,
 
     .dpy_gl_scanout_disable  = qemu_spice_gl_scanout_disable,
     .dpy_gl_scanout_texture  = qemu_spice_gl_scanout_texture,
-- 
2.29.0

[PATCH 07/20] ui: add gd_gl_area_scanout_disable

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Require the callback, drop the fallback path.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/gtk.h | 1 +
 ui/console.c     | 7 +------
 ui/gtk-gl-area.c | 7 +++++++
 ui/gtk.c         | 1 +
 4 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/include/ui/gtk.h b/include/ui/gtk.h
index 3f395d7f94..7569d090fa 100644
--- a/include/ui/gtk.h
+++ b/include/ui/gtk.h
@@ -140,6 +140,7 @@ void gd_gl_area_scanout_texture(DisplayChangeListener *dcl,
                                 uint32_t backing_height,
                                 uint32_t x, uint32_t y,
                                 uint32_t w, uint32_t h);
+void gd_gl_area_scanout_disable(DisplayChangeListener *dcl);
 void gd_gl_area_scanout_flush(DisplayChangeListener *dcl,
                               uint32_t x, uint32_t y, uint32_t w, uint32_t h);
 void gtk_gl_area_init(void);
diff --git a/ui/console.c b/ui/console.c
index c0b1a3689c..ab9224429e 100644
--- a/ui/console.c
+++ b/ui/console.c
@@ -1806,12 +1806,7 @@ int dpy_gl_ctx_make_current(QemuConsole *con, QEMUGLContext ctx)
 void dpy_gl_scanout_disable(QemuConsole *con)
 {
     assert(con->gl);
-    if (con->gl->ops->dpy_gl_scanout_disable) {
-        con->gl->ops->dpy_gl_scanout_disable(con->gl);
-    } else {
-        con->gl->ops->dpy_gl_scanout_texture(con->gl, 0, false, 0, 0,
-                                             0, 0, 0, 0);
-    }
+    con->gl->ops->dpy_gl_scanout_disable(con->gl);
 }
 
 void dpy_gl_scanout_texture(QemuConsole *con,
diff --git a/ui/gtk-gl-area.c b/ui/gtk-gl-area.c
index 98c22d23f5..96fbe75387 100644
--- a/ui/gtk-gl-area.c
+++ b/ui/gtk-gl-area.c
@@ -198,6 +198,13 @@ void gd_gl_area_scanout_texture(DisplayChangeListener *dcl,
                          backing_id, false);
 }
 
+void gd_gl_area_scanout_disable(DisplayChangeListener *dcl)
+{
+    VirtualConsole *vc = container_of(dcl, VirtualConsole, gfx.dcl);
+
+    gtk_gl_area_set_scanout_mode(vc, false);
+}
+
 void gd_gl_area_scanout_flush(DisplayChangeListener *dcl,
                           uint32_t x, uint32_t y, uint32_t w, uint32_t h)
 {
diff --git a/ui/gtk.c b/ui/gtk.c
index a0e6b60ac4..00045881b1 100644
--- a/ui/gtk.c
+++ b/ui/gtk.c
@@ -638,6 +638,7 @@ static const DisplayChangeListenerOps dcl_gl_area_ops = {
     .dpy_gl_ctx_destroy      = gd_gl_area_destroy_context,
     .dpy_gl_ctx_make_current = gd_gl_area_make_current,
     .dpy_gl_scanout_texture  = gd_gl_area_scanout_texture,
+    .dpy_gl_scanout_disable  = gd_gl_area_scanout_disable,
     .dpy_gl_update           = gd_gl_area_scanout_flush,
 };
 
-- 
2.29.0

[PATCH 08/20] ui: annotate DCLOps callback requirements

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/console.h | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/include/ui/console.h b/include/ui/console.h
index ce6c72e37c..bea2b6329a 100644
--- a/include/ui/console.h
+++ b/include/ui/console.h
@@ -174,35 +174,49 @@ typedef struct DisplayState DisplayState;
 typedef struct DisplayChangeListenerOps {
     const char *dpy_name;
 
+    /* optional */
     void (*dpy_refresh)(DisplayChangeListener *dcl);
 
+    /* optional */
     void (*dpy_gfx_update)(DisplayChangeListener *dcl,
                            int x, int y, int w, int h);
+    /* optional */
     void (*dpy_gfx_switch)(DisplayChangeListener *dcl,
                            struct DisplaySurface *new_surface);
+    /* optional */
     bool (*dpy_gfx_check_format)(DisplayChangeListener *dcl,
                                  pixman_format_code_t format);
 
+    /* optional */
     void (*dpy_text_cursor)(DisplayChangeListener *dcl,
                             int x, int y);
+    /* optional */
     void (*dpy_text_resize)(DisplayChangeListener *dcl,
                             int w, int h);
+    /* optional */
     void (*dpy_text_update)(DisplayChangeListener *dcl,
                             int x, int y, int w, int h);
 
+    /* optional */
     void (*dpy_mouse_set)(DisplayChangeListener *dcl,
                           int x, int y, int on);
+    /* optional */
     void (*dpy_cursor_define)(DisplayChangeListener *dcl,
                               QEMUCursor *cursor);
 
+    /* required if GL */
     QEMUGLContext (*dpy_gl_ctx_create)(DisplayChangeListener *dcl,
                                        QEMUGLParams *params);
+    /* required if GL */
     void (*dpy_gl_ctx_destroy)(DisplayChangeListener *dcl,
                                QEMUGLContext ctx);
+    /* required if GL */
     int (*dpy_gl_ctx_make_current)(DisplayChangeListener *dcl,
                                    QEMUGLContext ctx);
 
+    /* required if GL */
     void (*dpy_gl_scanout_disable)(DisplayChangeListener *dcl);
+    /* required if GL */
     void (*dpy_gl_scanout_texture)(DisplayChangeListener *dcl,
                                    uint32_t backing_id,
                                    bool backing_y_0_top,
@@ -210,15 +224,20 @@ typedef struct DisplayChangeListenerOps {
                                    uint32_t backing_height,
                                    uint32_t x, uint32_t y,
                                    uint32_t w, uint32_t h);
+    /* optional */
     void (*dpy_gl_scanout_dmabuf)(DisplayChangeListener *dcl,
                                   QemuDmaBuf *dmabuf);
+    /* optional */
     void (*dpy_gl_cursor_dmabuf)(DisplayChangeListener *dcl,
                                  QemuDmaBuf *dmabuf, bool have_hot,
                                  uint32_t hot_x, uint32_t hot_y);
+    /* optional */
     void (*dpy_gl_cursor_position)(DisplayChangeListener *dcl,
                                    uint32_t pos_x, uint32_t pos_y);
+    /* optional */
     void (*dpy_gl_release_dmabuf)(DisplayChangeListener *dcl,
                                   QemuDmaBuf *dmabuf);
+    /* required if GL */
     void (*dpy_gl_update)(DisplayChangeListener *dcl,
                           uint32_t x, uint32_t y, uint32_t w, uint32_t h);
 
-- 
2.29.0

[PATCH 09/20] ui: remove console_has_gl_dmabuf()

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

This check is currently limited. It only is used by vhost-user-gpu (not
by vfio-display), and will print an error repeatedly during run-time.

We are going to dissociate the GL context from the
DisplayChangeListener, and listeners may come and go. The following
patches will address this differently.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/console.h        | 1 -
 hw/display/vhost-user-gpu.c | 5 -----
 ui/console.c                | 5 -----
 3 files changed, 11 deletions(-)

diff --git a/include/ui/console.h b/include/ui/console.h
index bea2b6329a..ac989fdf70 100644
--- a/include/ui/console.h
+++ b/include/ui/console.h
@@ -323,7 +323,6 @@ void dpy_gl_ctx_destroy(QemuConsole *con, QEMUGLContext ctx);
 int dpy_gl_ctx_make_current(QemuConsole *con, QEMUGLContext ctx);
 
 bool console_has_gl(QemuConsole *con);
-bool console_has_gl_dmabuf(QemuConsole *con);
 
 static inline int surface_stride(DisplaySurface *s)
 {
diff --git a/hw/display/vhost-user-gpu.c b/hw/display/vhost-user-gpu.c
index 55b0ed15f2..dd587436ff 100644
--- a/hw/display/vhost-user-gpu.c
+++ b/hw/display/vhost-user-gpu.c
@@ -224,11 +224,6 @@ vhost_user_gpu_handle_display(VhostUserGPU *g, VhostUserGpuMsg *msg)
             close(dmabuf->fd);
             dmabuf->fd = -1;
         }
-        if (!console_has_gl_dmabuf(con)) {
-            /* it would be nice to report that error earlier */
-            error_report("console doesn't support dmabuf!");
-            break;
-        }
         dpy_gl_release_dmabuf(con, dmabuf);
         if (fd == -1) {
             dpy_gl_scanout_disable(con);
diff --git a/ui/console.c b/ui/console.c
index ab9224429e..b5bc3f7699 100644
--- a/ui/console.c
+++ b/ui/console.c
@@ -1463,11 +1463,6 @@ bool console_has_gl(QemuConsole *con)
     return con->gl != NULL;
 }
 
-bool console_has_gl_dmabuf(QemuConsole *con)
-{
-    return con->gl != NULL && con->gl->ops->dpy_gl_scanout_dmabuf != NULL;
-}
-
 void register_displaychangelistener(DisplayChangeListener *dcl)
 {
     static const char nodev[] =
-- 
2.29.0

[PATCH 10/20] vhost-user-gpu: add a configuration flag for dmabuf usage

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Let's inform VirtioGPUBase that vhost-user-gpu require DMABUF messages.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/hw/virtio/virtio-gpu.h | 3 +++
 hw/display/vhost-user-gpu.c    | 2 ++
 2 files changed, 5 insertions(+)

diff --git a/include/hw/virtio/virtio-gpu.h b/include/hw/virtio/virtio-gpu.h
index 1aed7275c8..4f3dbf79f9 100644
--- a/include/hw/virtio/virtio-gpu.h
+++ b/include/hw/virtio/virtio-gpu.h
@@ -71,6 +71,7 @@ enum virtio_gpu_base_conf_flags {
     VIRTIO_GPU_FLAG_VIRGL_ENABLED = 1,
     VIRTIO_GPU_FLAG_STATS_ENABLED,
     VIRTIO_GPU_FLAG_EDID_ENABLED,
+    VIRTIO_GPU_FLAG_DMABUF_ENABLED,
 };
 
 #define virtio_gpu_virgl_enabled(_cfg) \
@@ -79,6 +80,8 @@ enum virtio_gpu_base_conf_flags {
     (_cfg.flags & (1 << VIRTIO_GPU_FLAG_STATS_ENABLED))
 #define virtio_gpu_edid_enabled(_cfg) \
     (_cfg.flags & (1 << VIRTIO_GPU_FLAG_EDID_ENABLED))
+#define virtio_gpu_dmabuf_enabled(_cfg) \
+    (_cfg.flags & (1 << VIRTIO_GPU_FLAG_DMABUF_ENABLED))
 
 struct virtio_gpu_base_conf {
     uint32_t max_outputs;
diff --git a/hw/display/vhost-user-gpu.c b/hw/display/vhost-user-gpu.c
index dd587436ff..b7bde9feb6 100644
--- a/hw/display/vhost-user-gpu.c
+++ b/hw/display/vhost-user-gpu.c
@@ -547,6 +547,8 @@ vhost_user_gpu_device_realize(DeviceState *qdev, Error **errp)
         return;
     }
 
+    /* existing backend may send DMABUF, so let's add that requirement */
+    g->parent_obj.conf.flags |= 1 << VIRTIO_GPU_FLAG_DMABUF_ENABLED;
     if (virtio_has_feature(g->vhost->dev.features, VIRTIO_GPU_F_VIRGL)) {
         g->parent_obj.conf.flags |= 1 << VIRTIO_GPU_FLAG_VIRGL_ENABLED;
     }
-- 
2.29.0

[PATCH 11/20] ui: add an optional get_flags callback to GraphicHwOps

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Those flags can be used to express different requirements for the
display or other needs.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/console.h         |  9 +++++++++
 hw/display/virtio-gpu-base.c | 16 ++++++++++++++++
 hw/display/virtio-vga.c      |  9 +++++++++
 hw/vfio/display.c            |  6 ++++++
 4 files changed, 40 insertions(+)

diff --git a/include/ui/console.h b/include/ui/console.h
index ac989fdf70..0595aa9953 100644
--- a/include/ui/console.h
+++ b/include/ui/console.h
@@ -368,7 +368,16 @@ static inline void console_write_ch(console_ch_t *dest, uint32_t ch)
     *dest = ch;
 }
 
+enum {
+    GRAPHIC_FLAGS_NONE     = 0,
+    /* require a console/display with GL callbacks */
+    GRAPHIC_FLAGS_GL       = 1 << 0,
+    /* require a console/display with DMABUF import */
+    GRAPHIC_FLAGS_DMABUF   = 1 << 1,
+};
+
 typedef struct GraphicHwOps {
+    int (*get_flags)(void *opaque); /* optional, default 0 */
     void (*invalidate)(void *opaque);
     void (*gfx_update)(void *opaque);
     bool gfx_update_async; /* if true, calls graphic_hw_update_done() */
diff --git a/hw/display/virtio-gpu-base.c b/hw/display/virtio-gpu-base.c
index 40ccd00f94..82c37aac62 100644
--- a/hw/display/virtio-gpu-base.c
+++ b/hw/display/virtio-gpu-base.c
@@ -114,7 +114,23 @@ virtio_gpu_gl_block(void *opaque, bool block)
     }
 }
 
+static int
+virtio_gpu_get_flags(void *opaque)
+{
+    VirtIOGPUBase *g = opaque;
+    int flags = GRAPHIC_FLAGS_NONE;
+
+    if (virtio_gpu_virgl_enabled(g->conf))
+        flags |= GRAPHIC_FLAGS_GL;
+
+    if (virtio_gpu_dmabuf_enabled(g->conf))
+        flags |= GRAPHIC_FLAGS_DMABUF;
+
+    return flags;
+}
+
 static const GraphicHwOps virtio_gpu_ops = {
+    .get_flags = virtio_gpu_get_flags,
     .invalidate = virtio_gpu_invalidate_display,
     .gfx_update = virtio_gpu_update_display,
     .text_update = virtio_gpu_text_update,
diff --git a/hw/display/virtio-vga.c b/hw/display/virtio-vga.c
index 81f776ee36..b071909b68 100644
--- a/hw/display/virtio-vga.c
+++ b/hw/display/virtio-vga.c
@@ -68,7 +68,16 @@ static void virtio_vga_base_gl_block(void *opaque, bool block)
     }
 }
 
+static int virtio_vga_base_get_flags(void *opaque)
+{
+    VirtIOVGABase *vvga = opaque;
+    VirtIOGPUBase *g = vvga->vgpu;
+
+    return g->hw_ops->get_flags(g);
+}
+
 static const GraphicHwOps virtio_vga_base_ops = {
+    .get_flags = virtio_vga_base_get_flags,
     .invalidate = virtio_vga_base_invalidate_display,
     .gfx_update = virtio_vga_base_update_display,
     .text_update = virtio_vga_base_text_update,
diff --git a/hw/vfio/display.c b/hw/vfio/display.c
index 42d67e870b..f04473e3ce 100644
--- a/hw/vfio/display.c
+++ b/hw/vfio/display.c
@@ -335,7 +335,13 @@ static void vfio_display_dmabuf_update(void *opaque)
     }
 }
 
+static int vfio_display_get_flags(void *opaque)
+{
+    return GRAPHIC_FLAGS_GL | GRAPHIC_FLAGS_DMABUF;
+}
+
 static const GraphicHwOps vfio_display_dmabuf_ops = {
+    .get_flags  = vfio_display_get_flags,
     .gfx_update = vfio_display_dmabuf_update,
     .ui_info    = vfio_display_edid_ui_info,
 };
-- 
2.29.0

Re: [PATCH 11/20] ui: add an optional get_flags callback to GraphicHwOps

[Gerd Hoffmann]

> +static int
> +virtio_gpu_get_flags(void *opaque)
> +{
> +    VirtIOGPUBase *g = opaque;
> +    int flags = GRAPHIC_FLAGS_NONE;
> +
> +    if (virtio_gpu_virgl_enabled(g->conf))
> +        flags |= GRAPHIC_FLAGS_GL;
> +
> +    if (virtio_gpu_dmabuf_enabled(g->conf))
> +        flags |= GRAPHIC_FLAGS_DMABUF;

fbe6ba76ac01 ui: add an optional get_flags callback to GraphicHwOps
ERROR: braces {} are necessary for all arms of this statement
#50: FILE: hw/display/virtio-gpu-base.c:123:
+    if (virtio_gpu_virgl_enabled(g->conf))
[...]

ERROR: braces {} are necessary for all arms of this statement
#53: FILE: hw/display/virtio-gpu-base.c:126:
+    if (virtio_gpu_dmabuf_enabled(g->conf))
[...]

total: 2 errors, 0 warnings, 68 lines checked

Re: [PATCH 11/20] ui: add an optional get_flags callback to GraphicHwOps

[Marc-André Lureau]

[-- Attachment #1: Type: text/plain, Size: 957 bytes --]

On Wed, Feb 3, 2021 at 7:48 PM Gerd Hoffmann <kraxel@redhat.com> wrote:

> > +static int
> > +virtio_gpu_get_flags(void *opaque)
> > +{
> > +    VirtIOGPUBase *g = opaque;
> > +    int flags = GRAPHIC_FLAGS_NONE;
> > +
> > +    if (virtio_gpu_virgl_enabled(g->conf))
> > +        flags |= GRAPHIC_FLAGS_GL;
> > +
> > +    if (virtio_gpu_dmabuf_enabled(g->conf))
> > +        flags |= GRAPHIC_FLAGS_DMABUF;
>
> fbe6ba76ac01 ui: add an optional get_flags callback to GraphicHwOps
> ERROR: braces {} are necessary for all arms of this statement
> #50: FILE: hw/display/virtio-gpu-base.c:123:
> +    if (virtio_gpu_virgl_enabled(g->conf))
> [...]
>
> ERROR: braces {} are necessary for all arms of this statement
> #53: FILE: hw/display/virtio-gpu-base.c:126:
> +    if (virtio_gpu_dmabuf_enabled(g->conf))
> [...]
>
> total: 2 errors, 0 warnings, 68 lines checked
>

If you queued the series, do you mind squashing a style fix? Otherwise I
can resend.

Thanks

[-- Attachment #2: Type: text/html, Size: 1449 bytes --]

Re: [PATCH 11/20] ui: add an optional get_flags callback to GraphicHwOps

[Gerd Hoffmann]

On Thu, Feb 04, 2021 at 02:21:02AM +0400, Marc-André Lureau wrote:
> On Wed, Feb 3, 2021 at 7:48 PM Gerd Hoffmann <kraxel@redhat.com> wrote:
> 
> > > +static int
> > > +virtio_gpu_get_flags(void *opaque)
> > > +{
> > > +    VirtIOGPUBase *g = opaque;
> > > +    int flags = GRAPHIC_FLAGS_NONE;
> > > +
> > > +    if (virtio_gpu_virgl_enabled(g->conf))
> > > +        flags |= GRAPHIC_FLAGS_GL;
> > > +
> > > +    if (virtio_gpu_dmabuf_enabled(g->conf))
> > > +        flags |= GRAPHIC_FLAGS_DMABUF;
> >
> > fbe6ba76ac01 ui: add an optional get_flags callback to GraphicHwOps
> > ERROR: braces {} are necessary for all arms of this statement
> > #50: FILE: hw/display/virtio-gpu-base.c:123:
> > +    if (virtio_gpu_virgl_enabled(g->conf))
> > [...]
> >
> > ERROR: braces {} are necessary for all arms of this statement
> > #53: FILE: hw/display/virtio-gpu-base.c:126:
> > +    if (virtio_gpu_dmabuf_enabled(g->conf))
> > [...]
> >
> > total: 2 errors, 0 warnings, 68 lines checked
> >
> 
> If you queued the series, do you mind squashing a style fix? Otherwise I
> can resend.

Whatever is easier for you (note there are more codestyle warnings in
following patches, see patchew report).

take care,
  Gerd

[PATCH 12/20] ui: add a DCLOps callback to check dmabuf support

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/console.h |  2 ++
 ui/console.c         | 13 +++++++++++++
 2 files changed, 15 insertions(+)

diff --git a/include/ui/console.h b/include/ui/console.h
index 0595aa9953..875885d9c7 100644
--- a/include/ui/console.h
+++ b/include/ui/console.h
@@ -224,6 +224,8 @@ typedef struct DisplayChangeListenerOps {
                                    uint32_t backing_height,
                                    uint32_t x, uint32_t y,
                                    uint32_t w, uint32_t h);
+    /* optional (default to true if has dpy_gl_scanout_dmabuf) */
+    bool (*dpy_has_dmabuf)(DisplayChangeListener *dcl);
     /* optional */
     void (*dpy_gl_scanout_dmabuf)(DisplayChangeListener *dcl,
                                   QemuDmaBuf *dmabuf);
diff --git a/ui/console.c b/ui/console.c
index b5bc3f7699..a645418ada 100644
--- a/ui/console.c
+++ b/ui/console.c
@@ -1463,6 +1463,19 @@ bool console_has_gl(QemuConsole *con)
     return con->gl != NULL;
 }
 
+static bool displaychangelistener_has_dmabuf(DisplayChangeListener *dcl)
+{
+    if (dcl->ops->dpy_has_dmabuf) {
+        return dcl->ops->dpy_has_dmabuf(dcl);
+    }
+
+    if (dcl->ops->dpy_gl_scanout_dmabuf) {
+        return true;
+    }
+
+    return false;
+}
+
 void register_displaychangelistener(DisplayChangeListener *dcl)
 {
     static const char nodev[] =
-- 
2.29.0

[PATCH 13/20] ui: check hw requirements during DCL registration

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 ui/console.c | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)

diff --git a/ui/console.c b/ui/console.c
index a645418ada..47124e8cad 100644
--- a/ui/console.c
+++ b/ui/console.c
@@ -1476,12 +1476,34 @@ static bool displaychangelistener_has_dmabuf(DisplayChangeListener *dcl)
     return false;
 }
 
+static bool dpy_compatible_with(QemuConsole *con, DisplayChangeListener *dcl, Error **errp)
+{
+    ERRP_GUARD();
+    int flags;
+
+    flags = con->hw_ops->get_flags ? con->hw_ops->get_flags(con->hw) : 0;
+
+    if (flags & GRAPHIC_FLAGS_GL && !console_has_gl(con)) {
+        error_setg(errp, "The console requires a GL context.");
+        return false;
+
+    }
+
+    if (flags & GRAPHIC_FLAGS_DMABUF && !displaychangelistener_has_dmabuf(dcl)) {
+        error_setg(errp, "The console requires display DMABUF support.");
+        return false;
+    }
+
+    return true;
+}
+
 void register_displaychangelistener(DisplayChangeListener *dcl)
 {
     static const char nodev[] =
         "This VM has no graphic display device.";
     static DisplaySurface *dummy;
     QemuConsole *con;
+    Error *err = NULL;
 
     assert(!dcl->ds);
 
@@ -1496,6 +1518,11 @@ void register_displaychangelistener(DisplayChangeListener *dcl)
         dcl->con->gl = dcl;
     }
 
+    if (dcl->con && !dpy_compatible_with(dcl->con, dcl, &err)) {
+        error_report_err(err);
+        exit(1);
+    }
+
     trace_displaychangelistener_register(dcl, dcl->ops->dpy_name);
     dcl->ds = get_alloc_displaystate();
     QLIST_INSERT_HEAD(&dcl->ds->listeners, dcl, next);
-- 
2.29.0

[PATCH 14/20] ui: add qemu_egl_has_dmabuf helper

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/egl-helpers.h | 1 +
 ui/egl-helpers.c         | 9 +++++++++
 2 files changed, 10 insertions(+)

diff --git a/include/ui/egl-helpers.h b/include/ui/egl-helpers.h
index 94a4b3e6f3..5b1f7fafe0 100644
--- a/include/ui/egl-helpers.h
+++ b/include/ui/egl-helpers.h
@@ -51,5 +51,6 @@ EGLSurface qemu_egl_init_surface_x11(EGLContext ectx, EGLNativeWindowType win);
 int qemu_egl_init_dpy_x11(EGLNativeDisplayType dpy, DisplayGLMode mode);
 int qemu_egl_init_dpy_mesa(EGLNativeDisplayType dpy, DisplayGLMode mode);
 EGLContext qemu_egl_init_ctx(void);
+bool qemu_egl_has_dmabuf(void);
 
 #endif /* EGL_HELPERS_H */
diff --git a/ui/egl-helpers.c b/ui/egl-helpers.c
index 7c530c2825..0fe62fb885 100644
--- a/ui/egl-helpers.c
+++ b/ui/egl-helpers.c
@@ -441,6 +441,15 @@ int qemu_egl_init_dpy_mesa(EGLNativeDisplayType dpy, DisplayGLMode mode)
 #endif
 }
 
+bool qemu_egl_has_dmabuf(void)
+{
+    if (qemu_egl_display == EGL_NO_DISPLAY ||
+        !epoxy_has_egl_extension(qemu_egl_display, "EGL_EXT_image_dma_buf_import")) {
+        return false;
+    }
+    return true;
+}
+
 EGLContext qemu_egl_init_ctx(void)
 {
     static const EGLint ctx_att_core[] = {
-- 
2.29.0

[PATCH 15/20] ui: check gtk-egl dmabuf support

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/gtk.h | 1 +
 ui/gtk.c         | 9 +++++++++
 2 files changed, 10 insertions(+)

diff --git a/include/ui/gtk.h b/include/ui/gtk.h
index 7569d090fa..aaef884b95 100644
--- a/include/ui/gtk.h
+++ b/include/ui/gtk.h
@@ -48,6 +48,7 @@ typedef struct VirtualGfxConsole {
     int cursor_y;
     bool y0_top;
     bool scanout_mode;
+    bool has_dmabuf;
 #endif
 } VirtualGfxConsole;
 
diff --git a/ui/gtk.c b/ui/gtk.c
index 00045881b1..f41c396cb9 100644
--- a/ui/gtk.c
+++ b/ui/gtk.c
@@ -623,6 +623,13 @@ static const DisplayChangeListenerOps dcl_ops = {
 
 #if defined(CONFIG_OPENGL)
 
+static bool gd_has_dmabuf(DisplayChangeListener *dcl)
+{
+    VirtualConsole *vc = container_of(dcl, VirtualConsole, gfx.dcl);
+
+    return vc->gfx.has_dmabuf;
+}
+
 /** DisplayState Callbacks (opengl version) **/
 
 static const DisplayChangeListenerOps dcl_gl_area_ops = {
@@ -661,6 +668,7 @@ static const DisplayChangeListenerOps dcl_egl_ops = {
     .dpy_gl_cursor_position  = gd_egl_cursor_position,
     .dpy_gl_release_dmabuf   = gd_egl_release_dmabuf,
     .dpy_gl_update           = gd_egl_scanout_flush,
+    .dpy_has_dmabuf          = gd_has_dmabuf,
 };
 
 #endif /* CONFIG_OPENGL */
@@ -2004,6 +2012,7 @@ static GSList *gd_vc_gfx_init(GtkDisplayState *s, VirtualConsole *vc,
             gtk_widget_set_double_buffered(vc->gfx.drawing_area, FALSE);
 #pragma GCC diagnostic pop
             vc->gfx.dcl.ops = &dcl_egl_ops;
+            vc->gfx.has_dmabuf = qemu_egl_has_dmabuf();
         }
     } else
 #endif
-- 
2.29.0

[PATCH 16/20] ui: add egl dmabuf import to gtkglarea

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

GtkGLArea is used on wayland, where EGL is usually available.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/ui/gtk.h |  2 ++
 ui/gtk-gl-area.c | 18 ++++++++++++++++++
 ui/gtk.c         | 22 ++++++++++++++++++++++
 3 files changed, 42 insertions(+)

diff --git a/include/ui/gtk.h b/include/ui/gtk.h
index aaef884b95..3c1cd98db8 100644
--- a/include/ui/gtk.h
+++ b/include/ui/gtk.h
@@ -134,6 +134,8 @@ QEMUGLContext gd_gl_area_create_context(DisplayChangeListener *dcl,
                                         QEMUGLParams *params);
 void gd_gl_area_destroy_context(DisplayChangeListener *dcl,
                                 QEMUGLContext ctx);
+void gd_gl_area_scanout_dmabuf(DisplayChangeListener *dcl,
+                               QemuDmaBuf *dmabuf);
 void gd_gl_area_scanout_texture(DisplayChangeListener *dcl,
                                 uint32_t backing_id,
                                 bool backing_y_0_top,
diff --git a/ui/gtk-gl-area.c b/ui/gtk-gl-area.c
index 96fbe75387..72bcd94918 100644
--- a/ui/gtk-gl-area.c
+++ b/ui/gtk-gl-area.c
@@ -213,6 +213,24 @@ void gd_gl_area_scanout_flush(DisplayChangeListener *dcl,
     gtk_gl_area_queue_render(GTK_GL_AREA(vc->gfx.drawing_area));
 }
 
+void gd_gl_area_scanout_dmabuf(DisplayChangeListener *dcl,
+                               QemuDmaBuf *dmabuf)
+{
+#ifdef CONFIG_OPENGL_DMABUF
+    VirtualConsole *vc = container_of(dcl, VirtualConsole, gfx.dcl);
+
+    gtk_gl_area_make_current(GTK_GL_AREA(vc->gfx.drawing_area));
+    egl_dmabuf_import_texture(dmabuf);
+    if (!dmabuf->texture) {
+        return;
+    }
+
+    gd_gl_area_scanout_texture(dcl, dmabuf->texture,
+                               false, dmabuf->width, dmabuf->height,
+                               0, 0, dmabuf->width, dmabuf->height);
+#endif
+}
+
 void gtk_gl_area_init(void)
 {
     display_opengl = 1;
diff --git a/ui/gtk.c b/ui/gtk.c
index f41c396cb9..79dc240120 100644
--- a/ui/gtk.c
+++ b/ui/gtk.c
@@ -627,6 +627,12 @@ static bool gd_has_dmabuf(DisplayChangeListener *dcl)
 {
     VirtualConsole *vc = container_of(dcl, VirtualConsole, gfx.dcl);
 
+    if (gtk_use_gl_area && !gtk_widget_get_realized(vc->gfx.drawing_area)) {
+        /* FIXME: Assume it will work, actual check done after realize */
+        /* fixing this would require delaying listener registration */
+        return true;
+    }
+
     return vc->gfx.has_dmabuf;
 }
 
@@ -647,6 +653,8 @@ static const DisplayChangeListenerOps dcl_gl_area_ops = {
     .dpy_gl_scanout_texture  = gd_gl_area_scanout_texture,
     .dpy_gl_scanout_disable  = gd_gl_area_scanout_disable,
     .dpy_gl_update           = gd_gl_area_scanout_flush,
+    .dpy_gl_scanout_dmabuf   = gd_gl_area_scanout_dmabuf,
+    .dpy_has_dmabuf          = gd_has_dmabuf,
 };
 
 static const DisplayChangeListenerOps dcl_egl_ops = {
@@ -1983,6 +1991,18 @@ static GtkWidget *gd_create_menu_machine(GtkDisplayState *s)
     return machine_menu;
 }
 
+#if defined(CONFIG_OPENGL)
+static void gl_area_realize(GtkGLArea *area, VirtualConsole *vc)
+{
+    gtk_gl_area_make_current(area);
+    qemu_egl_display = eglGetCurrentDisplay();
+    vc->gfx.has_dmabuf = qemu_egl_has_dmabuf();
+    if (!vc->gfx.has_dmabuf) {
+        error_report("GtkGLArea console lacks DMABUF support.");
+    }
+}
+#endif
+
 static GSList *gd_vc_gfx_init(GtkDisplayState *s, VirtualConsole *vc,
                               QemuConsole *con, int idx,
                               GSList *group, GtkWidget *view_menu)
@@ -1998,6 +2018,8 @@ static GSList *gd_vc_gfx_init(GtkDisplayState *s, VirtualConsole *vc,
     if (display_opengl) {
         if (gtk_use_gl_area) {
             vc->gfx.drawing_area = gtk_gl_area_new();
+            g_signal_connect(vc->gfx.drawing_area, "realize",
+                             G_CALLBACK(gl_area_realize), vc);
             vc->gfx.dcl.ops = &dcl_gl_area_ops;
         } else {
             vc->gfx.drawing_area = gtk_drawing_area_new();
-- 
2.29.0

[PATCH 17/20] virtio-gpu: avoid re-entering cmdq processing

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

The next patch will notify the GL context got flush, which will resume
the queue processing. However, if this happens within the caller
context, it will end up with a stack overflow flush/update loop.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/hw/virtio/virtio-gpu.h | 1 +
 hw/display/virtio-gpu.c        | 5 +++++
 2 files changed, 6 insertions(+)

diff --git a/include/hw/virtio/virtio-gpu.h b/include/hw/virtio/virtio-gpu.h
index 4f3dbf79f9..0043268e90 100644
--- a/include/hw/virtio/virtio-gpu.h
+++ b/include/hw/virtio/virtio-gpu.h
@@ -148,6 +148,7 @@ struct VirtIOGPU {
 
     uint64_t hostmem;
 
+    bool processing_cmdq;
     bool renderer_inited;
     bool renderer_reset;
     QEMUTimer *fence_poll;
diff --git a/hw/display/virtio-gpu.c b/hw/display/virtio-gpu.c
index 0e833a462b..7eb4265a6d 100644
--- a/hw/display/virtio-gpu.c
+++ b/hw/display/virtio-gpu.c
@@ -814,6 +814,10 @@ void virtio_gpu_process_cmdq(VirtIOGPU *g)
 {
     struct virtio_gpu_ctrl_command *cmd;
 
+    if (g->processing_cmdq) {
+        return;
+    }
+    g->processing_cmdq = true;
     while (!QTAILQ_EMPTY(&g->cmdq)) {
         cmd = QTAILQ_FIRST(&g->cmdq);
 
@@ -843,6 +847,7 @@ void virtio_gpu_process_cmdq(VirtIOGPU *g)
             g_free(cmd);
         }
     }
+    g->processing_cmdq = false;
 }
 
 static void virtio_gpu_gl_unblock(VirtIOGPUBase *b)
-- 
2.29.0

[PATCH 18/20] display/ui: add a callback to indicate GL state is flushed

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Displaying rendered resources requires blocking qemu GPU to avoid extra
framebuffer copies. For an external display, via Spice currently, there
is a callback to block/unblock the rendering in the same thread.

But with the vhost-user-gpu backend, the qemu process doesn't handle
the rendering itself, and the blocking callback isn't effective.
Instead, the backend must be notified when the display code is done.

Fix this by adding a new GraphicHwOps callback to indicate the GL state
is flushed, and we are done manipulating the shared GL resources. Call
it from gtk and spice display.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 include/hw/virtio/virtio-gpu.h |  2 +-
 include/ui/console.h           |  2 ++
 hw/display/vhost-user-gpu.c    |  4 ++--
 hw/display/virtio-gpu-base.c   | 17 ++++++++++++-----
 hw/display/virtio-gpu.c        |  4 ++--
 hw/display/virtio-vga.c        | 11 +++++++++++
 ui/console.c                   |  9 +++++++++
 ui/gtk-egl.c                   |  3 +++
 ui/gtk-gl-area.c               |  3 +++
 ui/sdl2-gl.c                   |  2 ++
 ui/spice-display.c             |  1 +
 11 files changed, 48 insertions(+), 10 deletions(-)

diff --git a/include/hw/virtio/virtio-gpu.h b/include/hw/virtio/virtio-gpu.h
index 0043268e90..fae149235c 100644
--- a/include/hw/virtio/virtio-gpu.h
+++ b/include/hw/virtio/virtio-gpu.h
@@ -121,7 +121,7 @@ struct VirtIOGPUBase {
 struct VirtIOGPUBaseClass {
     VirtioDeviceClass parent;
 
-    void (*gl_unblock)(VirtIOGPUBase *g);
+    void (*gl_flushed)(VirtIOGPUBase *g);
 };
 
 #define VIRTIO_GPU_BASE_PROPERTIES(_state, _conf)                       \
diff --git a/include/ui/console.h b/include/ui/console.h
index 875885d9c7..d30e972d0b 100644
--- a/include/ui/console.h
+++ b/include/ui/console.h
@@ -387,6 +387,7 @@ typedef struct GraphicHwOps {
     void (*update_interval)(void *opaque, uint64_t interval);
     int (*ui_info)(void *opaque, uint32_t head, QemuUIInfo *info);
     void (*gl_block)(void *opaque, bool block);
+    void (*gl_flushed)(void *opaque);
 } GraphicHwOps;
 
 QemuConsole *graphic_console_init(DeviceState *dev, uint32_t head,
@@ -402,6 +403,7 @@ void graphic_hw_update_done(QemuConsole *con);
 void graphic_hw_invalidate(QemuConsole *con);
 void graphic_hw_text_update(QemuConsole *con, console_ch_t *chardata);
 void graphic_hw_gl_block(QemuConsole *con, bool block);
+void graphic_hw_gl_flushed(QemuConsole *con);
 
 void qemu_console_early_init(void);
 
diff --git a/hw/display/vhost-user-gpu.c b/hw/display/vhost-user-gpu.c
index b7bde9feb6..4d8cb3525b 100644
--- a/hw/display/vhost-user-gpu.c
+++ b/hw/display/vhost-user-gpu.c
@@ -360,7 +360,7 @@ vhost_user_gpu_update_blocked(VhostUserGPU *g, bool blocked)
 }
 
 static void
-vhost_user_gpu_gl_unblock(VirtIOGPUBase *b)
+vhost_user_gpu_gl_flushed(VirtIOGPUBase *b)
 {
     VhostUserGPU *g = VHOST_USER_GPU(b);
 
@@ -578,7 +578,7 @@ vhost_user_gpu_class_init(ObjectClass *klass, void *data)
     VirtioDeviceClass *vdc = VIRTIO_DEVICE_CLASS(klass);
     VirtIOGPUBaseClass *vgc = VIRTIO_GPU_BASE_CLASS(klass);
 
-    vgc->gl_unblock = vhost_user_gpu_gl_unblock;
+    vgc->gl_flushed = vhost_user_gpu_gl_flushed;
 
     vdc->realize = vhost_user_gpu_device_realize;
     vdc->reset = vhost_user_gpu_reset;
diff --git a/hw/display/virtio-gpu-base.c b/hw/display/virtio-gpu-base.c
index 82c37aac62..6d6b5eed63 100644
--- a/hw/display/virtio-gpu-base.c
+++ b/hw/display/virtio-gpu-base.c
@@ -97,21 +97,27 @@ static int virtio_gpu_ui_info(void *opaque, uint32_t idx, QemuUIInfo *info)
 }
 
 static void
-virtio_gpu_gl_block(void *opaque, bool block)
+virtio_gpu_gl_flushed(void *opaque)
 {
     VirtIOGPUBase *g = opaque;
     VirtIOGPUBaseClass *vgc = VIRTIO_GPU_BASE_GET_CLASS(g);
 
+    if (vgc->gl_flushed) {
+        vgc->gl_flushed(g);
+    }
+}
+
+static void
+virtio_gpu_gl_block(void *opaque, bool block)
+{
+    VirtIOGPUBase *g = opaque;
+
     if (block) {
         g->renderer_blocked++;
     } else {
         g->renderer_blocked--;
     }
     assert(g->renderer_blocked >= 0);
-
-    if (g->renderer_blocked == 0) {
-        vgc->gl_unblock(g);
-    }
 }
 
 static int
@@ -136,6 +142,7 @@ static const GraphicHwOps virtio_gpu_ops = {
     .text_update = virtio_gpu_text_update,
     .ui_info = virtio_gpu_ui_info,
     .gl_block = virtio_gpu_gl_block,
+    .gl_flushed = virtio_gpu_gl_flushed,
 };
 
 bool
diff --git a/hw/display/virtio-gpu.c b/hw/display/virtio-gpu.c
index 7eb4265a6d..2e4a9822b6 100644
--- a/hw/display/virtio-gpu.c
+++ b/hw/display/virtio-gpu.c
@@ -850,7 +850,7 @@ void virtio_gpu_process_cmdq(VirtIOGPU *g)
     g->processing_cmdq = false;
 }
 
-static void virtio_gpu_gl_unblock(VirtIOGPUBase *b)
+static void virtio_gpu_gl_flushed(VirtIOGPUBase *b)
 {
     VirtIOGPU *g = VIRTIO_GPU(b);
 
@@ -1257,7 +1257,7 @@ static void virtio_gpu_class_init(ObjectClass *klass, void *data)
     VirtioDeviceClass *vdc = VIRTIO_DEVICE_CLASS(klass);
     VirtIOGPUBaseClass *vgc = VIRTIO_GPU_BASE_CLASS(klass);
 
-    vgc->gl_unblock = virtio_gpu_gl_unblock;
+    vgc->gl_flushed = virtio_gpu_gl_flushed;
     vdc->realize = virtio_gpu_device_realize;
     vdc->reset = virtio_gpu_reset;
     vdc->get_config = virtio_gpu_get_config;
diff --git a/hw/display/virtio-vga.c b/hw/display/virtio-vga.c
index b071909b68..d3c6404061 100644
--- a/hw/display/virtio-vga.c
+++ b/hw/display/virtio-vga.c
@@ -68,6 +68,16 @@ static void virtio_vga_base_gl_block(void *opaque, bool block)
     }
 }
 
+static void virtio_vga_base_gl_flushed(void *opaque)
+{
+    VirtIOVGABase *vvga = opaque;
+    VirtIOGPUBase *g = vvga->vgpu;
+
+    if (g->hw_ops->gl_flushed) {
+        g->hw_ops->gl_flushed(g);
+    }
+}
+
 static int virtio_vga_base_get_flags(void *opaque)
 {
     VirtIOVGABase *vvga = opaque;
@@ -83,6 +93,7 @@ static const GraphicHwOps virtio_vga_base_ops = {
     .text_update = virtio_vga_base_text_update,
     .ui_info = virtio_vga_base_ui_info,
     .gl_block = virtio_vga_base_gl_block,
+    .gl_flushed = virtio_vga_base_gl_flushed,
 };
 
 static const VMStateDescription vmstate_virtio_vga_base = {
diff --git a/ui/console.c b/ui/console.c
index 47124e8cad..0de287713f 100644
--- a/ui/console.c
+++ b/ui/console.c
@@ -294,6 +294,15 @@ void graphic_hw_gl_block(QemuConsole *con, bool block)
     }
 }
 
+void graphic_hw_gl_flushed(QemuConsole *con)
+{
+    assert(con != NULL);
+
+    if (con->hw_ops->gl_flushed) {
+        con->hw_ops->gl_flushed(con->hw);
+    }
+}
+
 int qemu_console_get_window_id(QemuConsole *con)
 {
     return con->window_id;
diff --git a/ui/gtk-egl.c b/ui/gtk-egl.c
index 71c3d698b4..588e7b1bb1 100644
--- a/ui/gtk-egl.c
+++ b/ui/gtk-egl.c
@@ -92,6 +92,9 @@ void gd_egl_draw(VirtualConsole *vc)
         vc->gfx.scale_x = (double)ww / surface_width(vc->gfx.ds);
         vc->gfx.scale_y = (double)wh / surface_height(vc->gfx.ds);
     }
+
+    glFlush();
+    graphic_hw_gl_flushed(vc->gfx.dcl.con);
 }
 
 void gd_egl_update(DisplayChangeListener *dcl,
diff --git a/ui/gtk-gl-area.c b/ui/gtk-gl-area.c
index 72bcd94918..e7ca73c7b1 100644
--- a/ui/gtk-gl-area.c
+++ b/ui/gtk-gl-area.c
@@ -70,6 +70,9 @@ void gd_gl_area_draw(VirtualConsole *vc)
         surface_gl_setup_viewport(vc->gfx.gls, vc->gfx.ds, ww, wh);
         surface_gl_render_texture(vc->gfx.gls, vc->gfx.ds);
     }
+
+    glFlush();
+    graphic_hw_gl_flushed(vc->gfx.dcl.con);
 }
 
 void gd_gl_area_update(DisplayChangeListener *dcl,
diff --git a/ui/sdl2-gl.c b/ui/sdl2-gl.c
index 82139ba49e..fd594d7461 100644
--- a/ui/sdl2-gl.c
+++ b/ui/sdl2-gl.c
@@ -58,6 +58,7 @@ static void sdl2_gl_render_surface(struct sdl2_console *scon)
 
     surface_gl_render_texture(scon->gls, scon->surface);
     SDL_GL_SwapWindow(scon->real_window);
+    graphic_hw_gl_flushed(scon->dcl.con);
 }
 
 void sdl2_gl_update(DisplayChangeListener *dcl,
@@ -240,4 +241,5 @@ void sdl2_gl_scanout_flush(DisplayChangeListener *dcl,
     egl_fb_blit(&scon->win_fb, &scon->guest_fb, !scon->y0_top);
 
     SDL_GL_SwapWindow(scon->real_window);
+    graphic_hw_gl_flushed(dcl->con);
 }
diff --git a/ui/spice-display.c b/ui/spice-display.c
index 6a5e189a67..7247a93924 100644
--- a/ui/spice-display.c
+++ b/ui/spice-display.c
@@ -826,6 +826,7 @@ static void qemu_spice_gl_unblock_bh(void *opaque)
     SimpleSpiceDisplay *ssd = opaque;
 
     qemu_spice_gl_block(ssd, false);
+    graphic_hw_gl_flushed(ssd->dcl.con);
 }
 
 static void qemu_spice_gl_block_timer(void *opaque)
-- 
2.29.0

[PATCH 19/20] chardev: check if the chardev is registered for yanking

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

Not all chardevs are created via qmp_chardev_open_socket(), and those
should not call the yank function registration, as this will eventually
assert() not being registered.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 chardev/char-socket.c | 53 +++++++++++++++++++++++++++----------------
 1 file changed, 33 insertions(+), 20 deletions(-)

diff --git a/chardev/char-socket.c b/chardev/char-socket.c
index 8a707d766c..9061981f6d 100644
--- a/chardev/char-socket.c
+++ b/chardev/char-socket.c
@@ -417,8 +417,9 @@ static void tcp_chr_free_connection(Chardev *chr)
 
     tcp_set_msgfds(chr, NULL, 0);
     remove_fd_in_watch(chr);
-    if (s->state == TCP_CHARDEV_STATE_CONNECTING
-        || s->state == TCP_CHARDEV_STATE_CONNECTED) {
+    if (s->registered_yank &&
+        (s->state == TCP_CHARDEV_STATE_CONNECTING
+        || s->state == TCP_CHARDEV_STATE_CONNECTED)) {
         yank_unregister_function(CHARDEV_YANK_INSTANCE(chr->label),
                                  yank_generic_iochannel,
                                  QIO_CHANNEL(s->sioc));
@@ -940,9 +941,11 @@ static int tcp_chr_add_client(Chardev *chr, int fd)
     }
     tcp_chr_change_state(s, TCP_CHARDEV_STATE_CONNECTING);
     tcp_chr_set_client_ioc_name(chr, sioc);
-    yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
-                           yank_generic_iochannel,
-                           QIO_CHANNEL(sioc));
+    if (s->registered_yank) {
+        yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
+                               yank_generic_iochannel,
+                               QIO_CHANNEL(sioc));
+    }
     ret = tcp_chr_new_client(chr, sioc);
     object_unref(OBJECT(sioc));
     return ret;
@@ -957,9 +960,11 @@ static void tcp_chr_accept(QIONetListener *listener,
 
     tcp_chr_change_state(s, TCP_CHARDEV_STATE_CONNECTING);
     tcp_chr_set_client_ioc_name(chr, cioc);
-    yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
-                           yank_generic_iochannel,
-                           QIO_CHANNEL(cioc));
+    if (s->registered_yank) {
+        yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
+                               yank_generic_iochannel,
+                               QIO_CHANNEL(cioc));
+    }
     tcp_chr_new_client(chr, cioc);
 }
 
@@ -975,9 +980,11 @@ static int tcp_chr_connect_client_sync(Chardev *chr, Error **errp)
         object_unref(OBJECT(sioc));
         return -1;
     }
-    yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
-                           yank_generic_iochannel,
-                           QIO_CHANNEL(sioc));
+    if (s->registered_yank) {
+        yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
+                               yank_generic_iochannel,
+                               QIO_CHANNEL(sioc));
+    }
     tcp_chr_new_client(chr, sioc);
     object_unref(OBJECT(sioc));
     return 0;
@@ -993,9 +1000,11 @@ static void tcp_chr_accept_server_sync(Chardev *chr)
     tcp_chr_change_state(s, TCP_CHARDEV_STATE_CONNECTING);
     sioc = qio_net_listener_wait_client(s->listener);
     tcp_chr_set_client_ioc_name(chr, sioc);
-    yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
-                           yank_generic_iochannel,
-                           QIO_CHANNEL(sioc));
+    if (s->registered_yank) {
+        yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
+                               yank_generic_iochannel,
+                               QIO_CHANNEL(sioc));
+    }
     tcp_chr_new_client(chr, sioc);
     object_unref(OBJECT(sioc));
 }
@@ -1124,9 +1133,11 @@ static void qemu_chr_socket_connected(QIOTask *task, void *opaque)
 
     if (qio_task_propagate_error(task, &err)) {
         tcp_chr_change_state(s, TCP_CHARDEV_STATE_DISCONNECTED);
-        yank_unregister_function(CHARDEV_YANK_INSTANCE(chr->label),
-                                 yank_generic_iochannel,
-                                 QIO_CHANNEL(sioc));
+        if (s->registered_yank) {
+            yank_unregister_function(CHARDEV_YANK_INSTANCE(chr->label),
+                                     yank_generic_iochannel,
+                                     QIO_CHANNEL(sioc));
+        }
         check_report_connect_error(chr, err);
         goto cleanup;
     }
@@ -1160,9 +1171,11 @@ static void tcp_chr_connect_client_async(Chardev *chr)
     tcp_chr_change_state(s, TCP_CHARDEV_STATE_CONNECTING);
     sioc = qio_channel_socket_new();
     tcp_chr_set_client_ioc_name(chr, sioc);
-    yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
-                           yank_generic_iochannel,
-                           QIO_CHANNEL(sioc));
+    if (s->registered_yank) {
+        yank_register_function(CHARDEV_YANK_INSTANCE(chr->label),
+                               yank_generic_iochannel,
+                               QIO_CHANNEL(sioc));
+    }
     /*
      * Normally code would use the qio_channel_socket_connect_async
      * method which uses a QIOTask + qio_task_set_error internally
-- 
2.29.0

[PATCH 20/20] RFC: tests: add some virtio-gpu & vhost-user-gpu acceptance test

[marcandre.lureau]

From: Marc-André Lureau <marcandre.lureau@redhat.com>

This will check virtio/vhost-user-vga & virgl are correctly initialized
by the Linux kernel on an egl-headless display.

There are many other things that could be checked, but that's a start. I
also don't know yet how to nicely skip on incompatible host &
configurations.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
---
 tests/acceptance/virtio-gpu.py | 150 +++++++++++++++++++++++++++++++++
 1 file changed, 150 insertions(+)
 create mode 100644 tests/acceptance/virtio-gpu.py

diff --git a/tests/acceptance/virtio-gpu.py b/tests/acceptance/virtio-gpu.py
new file mode 100644
index 0000000000..8da20dcc4a
--- /dev/null
+++ b/tests/acceptance/virtio-gpu.py
@@ -0,0 +1,150 @@
+# virtio-gpu tests
+#
+# This work is licensed under the terms of the GNU GPL, version 2 or
+# later.  See the COPYING file in the top-level directory.
+
+
+from avocado_qemu import Test
+from avocado_qemu import BUILD_DIR
+from avocado_qemu import wait_for_console_pattern
+from avocado_qemu import exec_command_and_wait_for_pattern
+from avocado_qemu import is_readable_executable_file
+
+from qemu.accel import kvm_available
+
+import os
+import socket
+import subprocess
+
+
+ACCEL_NOT_AVAILABLE_FMT = "%s accelerator does not seem to be available"
+KVM_NOT_AVAILABLE = ACCEL_NOT_AVAILABLE_FMT % "KVM"
+
+
+def pick_default_vug_bin():
+    relative_path = "./contrib/vhost-user-gpu/vhost-user-gpu"
+    if is_readable_executable_file(relative_path):
+        return relative_path
+
+    bld_dir_path = os.path.join(BUILD_DIR, relative_path)
+    if is_readable_executable_file(bld_dir_path):
+        return bld_dir_path
+
+
+class VirtioGPUx86(Test):
+    """
+    :avocado: tags=virtio-gpu
+    """
+
+    KERNEL_COMMON_COMMAND_LINE = "printk.time=0 "
+    KERNEL_URL = (
+        "https://archives.fedoraproject.org/pub/fedora"
+        "/linux/releases/33/Everything/x86_64/os/images"
+        "/pxeboot/vmlinuz"
+    )
+    INITRD_URL = (
+        "https://archives.fedoraproject.org/pub/fedora"
+        "/linux/releases/33/Everything/x86_64/os/images"
+        "/pxeboot/initrd.img"
+    )
+
+    def wait_for_console_pattern(self, success_message, vm=None):
+        wait_for_console_pattern(
+            self, success_message, failure_message="Kernel panic - not syncing", vm=vm
+        )
+
+    def test_virtio_vga_virgl(self):
+        """
+        :avocado: tags=arch:x86_64
+        :avocado: tags=device:virtio-vga
+        """
+        kernel_command_line = (
+            self.KERNEL_COMMON_COMMAND_LINE + "console=ttyS0 rdinit=/bin/bash"
+        )
+        # FIXME: should check presence of virtio, virgl etc
+        if not kvm_available(self.arch, self.qemu_bin):
+            self.cancel(KVM_NOT_AVAILABLE)
+
+        kernel_path = self.fetch_asset(self.KERNEL_URL)
+        initrd_path = self.fetch_asset(self.INITRD_URL)
+
+        self.vm.set_console()
+        self.vm.add_args("-cpu", "host")
+        self.vm.add_args("-m", "2G")
+        self.vm.add_args("-machine", "pc,accel=kvm")
+        self.vm.add_args("-device", "virtio-vga,virgl=on")
+        self.vm.add_args("-display", "egl-headless")
+        self.vm.add_args(
+            "-kernel",
+            kernel_path,
+            "-initrd",
+            initrd_path,
+            "-append",
+            kernel_command_line,
+        )
+        self.vm.launch()
+        self.wait_for_console_pattern("as init process")
+        exec_command_and_wait_for_pattern(self, "/usr/sbin/modprobe virtio_gpu", "")
+        self.wait_for_console_pattern("features: +virgl +edid")
+
+    def test_vhost_user_vga_virgl(self):
+        """
+        :avocado: tags=arch:x86_64
+        :avocado: tags=device:vhost-user-vga
+        """
+        kernel_command_line = (
+            self.KERNEL_COMMON_COMMAND_LINE + "console=ttyS0 rdinit=/bin/bash"
+        )
+        # FIXME: should check presence of vhost-user-gpu, virgl, memfd etc
+        if not kvm_available(self.arch, self.qemu_bin):
+            self.cancel(KVM_NOT_AVAILABLE)
+
+        vug = pick_default_vug_bin()
+        if not vug:
+            self.cancel("Could not find vhost-user-gpu")
+
+        kernel_path = self.fetch_asset(self.KERNEL_URL)
+        initrd_path = self.fetch_asset(self.INITRD_URL)
+
+        # Create socketpair to connect proxy and remote processes
+        qemu_sock, vug_sock = socket.socketpair(socket.AF_UNIX, socket.SOCK_STREAM)
+        os.set_inheritable(qemu_sock.fileno(), True)
+        os.set_inheritable(vug_sock.fileno(), True)
+
+        self._vug_log_path = os.path.join(self.vm._test_dir, "vhost-user-gpu.log")
+        self._vug_log_file = open(self._vug_log_path, "wb")
+        print(self._vug_log_path)
+
+        vugp = subprocess.Popen(
+            [vug, "--virgl", "--fd=%d" % vug_sock.fileno()],
+            stdin=subprocess.DEVNULL,
+            stdout=self._vug_log_file,
+            stderr=subprocess.STDOUT,
+            shell=False,
+            close_fds=False,
+        )
+
+        self.vm.set_console()
+        self.vm.add_args("-cpu", "host")
+        self.vm.add_args("-m", "2G")
+        self.vm.add_args("-object", "memory-backend-memfd,id=mem,size=2G")
+        self.vm.add_args("-machine", "pc,memory-backend=mem,accel=kvm")
+        self.vm.add_args("-chardev", "socket,id=vug,fd=%d" % qemu_sock.fileno())
+        self.vm.add_args("-device", "vhost-user-vga,chardev=vug")
+        self.vm.add_args("-display", "egl-headless")
+        self.vm.add_args(
+            "-kernel",
+            kernel_path,
+            "-initrd",
+            initrd_path,
+            "-append",
+            kernel_command_line,
+        )
+        self.vm.launch()
+        self.wait_for_console_pattern("as init process")
+        exec_command_and_wait_for_pattern(self, "/usr/sbin/modprobe virtio_gpu", "")
+        self.wait_for_console_pattern("features: +virgl -edid")
+        self.vm.shutdown()
+        qemu_sock.close()
+        vugp.terminate()
+        vugp.wait()
-- 
2.29.0

Re: [PATCH 20/20] RFC: tests: add some virtio-gpu & vhost-user-gpu acceptance test

[Gerd Hoffmann]

On Tue, Feb 02, 2021 at 06:26:25PM +0400, marcandre.lureau@redhat.com wrote:
> From: Marc-André Lureau <marcandre.lureau@redhat.com>
> 
> This will check virtio/vhost-user-vga & virgl are correctly initialized
> by the Linux kernel on an egl-headless display.
> 
> There are many other things that could be checked, but that's a start. I
> also don't know yet how to nicely skip on incompatible host &
> configurations.

You can annotate tests like this:

       @avocado.skipUnless(os.path.exists(pick_default_vug_bin()), "no vhost-user-gpu")
> +    def test_vhost_user_vga_virgl(self):

[ queued whole series + some other pending ui/vga bits, kicked CI, lets
  see how it is going.  I suspect I'll have to drop this patch though. ]

HTH,
  Gerd

Re: [PATCH 20/20] RFC: tests: add some virtio-gpu & vhost-user-gpu acceptance test

[Marc-André Lureau]

Hi

On Wed, Feb 3, 2021 at 3:44 PM Gerd Hoffmann <kraxel@redhat.com> wrote:
>
> On Tue, Feb 02, 2021 at 06:26:25PM +0400, marcandre.lureau@redhat.com wrote:
> > From: Marc-André Lureau <marcandre.lureau@redhat.com>
> >
> > This will check virtio/vhost-user-vga & virgl are correctly initialized
> > by the Linux kernel on an egl-headless display.
> >
> > There are many other things that could be checked, but that's a start. I
> > also don't know yet how to nicely skip on incompatible host &
> > configurations.
>
> You can annotate tests like this:
>
>        @avocado.skipUnless(os.path.exists(pick_default_vug_bin()), "no vhost-user-gpu")
> > +    def test_vhost_user_vga_virgl(self):

Ah interesting, thanks!

>
> [ queued whole series + some other pending ui/vga bits, kicked CI, lets
>   see how it is going.  I suspect I'll have to drop this patch though. ]
>

Yes, Cleber was going to take a look, and perhaps send another avocado
series I could take some inspiration from.

thanks!



-- 
Marc-André Lureau

Re: [PATCH 20/20] RFC: tests: add some virtio-gpu & vhost-user-gpu acceptance test

[Gerd Hoffmann]

> > [ queued whole series + some other pending ui/vga bits, kicked CI, lets
> >   see how it is going.  I suspect I'll have to drop this patch though. ]
> >
> 
> Yes, Cleber was going to take a look, and perhaps send another avocado
> series I could take some inspiration from.

Doesn't look too bad, gitlab passed, cirrus still running.

So include it?  Or do you want send an improved non-rfc version anyway?

take care,
  Gerd

Re: [PATCH 20/20] RFC: tests: add some virtio-gpu & vhost-user-gpu acceptance test

[Marc-André Lureau]

Hi

On Wed, Feb 3, 2021 at 6:16 PM Gerd Hoffmann <kraxel@redhat.com> wrote:
>
> > > [ queued whole series + some other pending ui/vga bits, kicked CI, lets
> > >   see how it is going.  I suspect I'll have to drop this patch though. ]
> > >
> >
> > Yes, Cleber was going to take a look, and perhaps send another avocado
> > series I could take some inspiration from.
>
> Doesn't look too bad, gitlab passed, cirrus still running.
>
> So include it?  Or do you want send an improved non-rfc version anyway?
>

If it passes it all (including Peter's gate), it's good to go I would
think. Easier to improve based on something that is known to work.



-- 
Marc-André Lureau

Re: [PATCH 00/20] Various vhost-user-gpu & UI fixes

[no-reply]

Patchew URL: https://patchew.org/QEMU/20210202142625.609070-1-marcandre.lureau@redhat.com/



Hi,

This series seems to have some coding style problems. See output below for
more information:

Type: series
Message-id: 20210202142625.609070-1-marcandre.lureau@redhat.com
Subject: [PATCH 00/20] Various vhost-user-gpu & UI fixes

=== TEST SCRIPT BEGIN ===
#!/bin/bash
git rev-parse base > /dev/null || exit 0
git config --local diff.renamelimit 0
git config --local diff.renames True
git config --local diff.algorithm histogram
./scripts/checkpatch.pl --mailback base..
=== TEST SCRIPT END ===

Updating 3c8cf5a9c21ff8782164d1def7f44bd888713384
From https://github.com/patchew-project/qemu
 - [tag update]      patchew/20210128224141.638790-1-richard.henderson@linaro.org -> patchew/20210128224141.638790-1-richard.henderson@linaro.org
 - [tag update]      patchew/20210202135522.127380-1-dgilbert@redhat.com -> patchew/20210202135522.127380-1-dgilbert@redhat.com
 * [new tag]         patchew/20210202142625.609070-1-marcandre.lureau@redhat.com -> patchew/20210202142625.609070-1-marcandre.lureau@redhat.com
Switched to a new branch 'test'
22d6630 RFC: tests: add some virtio-gpu & vhost-user-gpu acceptance test
e3c9583 chardev: check if the chardev is registered for yanking
b1fb728 display/ui: add a callback to indicate GL state is flushed
8e9a9d9 virtio-gpu: avoid re-entering cmdq processing
6495eb7 ui: add egl dmabuf import to gtkglarea
9fe1210 ui: check gtk-egl dmabuf support
165cebf ui: add qemu_egl_has_dmabuf helper
6339b5b ui: check hw requirements during DCL registration
483c6af ui: add a DCLOps callback to check dmabuf support
43e7283 ui: add an optional get_flags callback to GraphicHwOps
24a96b4 vhost-user-gpu: add a configuration flag for dmabuf usage
851661e ui: remove console_has_gl_dmabuf()
6f62883 ui: annotate DCLOps callback requirements
0843e91 ui: add gd_gl_area_scanout_disable
ea06ae9 ui: remove gl_ctx_get_current
82f8fd2 ui: remove extra #ifdef CONFIG_OPENGL
3eeb59f vhost-user-gpu: handle display-info in a callback
ca98f97 vhost-user-gpu: use an extandable state enum for commands
f3f5b71 vhost-user-gpu: handle vhost-user-gpu features in a callback
679108c vhost-user-gpu: check backend for EDID support

=== OUTPUT BEGIN ===
1/20 Checking commit 679108c6db9d (vhost-user-gpu: check backend for EDID support)
2/20 Checking commit f3f5b712d643 (vhost-user-gpu: handle vhost-user-gpu features in a callback)
3/20 Checking commit ca98f9798940 (vhost-user-gpu: use an extandable state enum for commands)
4/20 Checking commit 3eeb59f5eda8 (vhost-user-gpu: handle display-info in a callback)
WARNING: line over 80 characters
#34: FILE: contrib/vhost-user-gpu/vhost-user-gpu.c:274:
+    if (!vg_recv_msg(vg, VHOST_USER_GPU_GET_DISPLAY_INFO, sizeof(dpy_info), &dpy_info)) {

total: 0 errors, 1 warnings, 47 lines checked

Patch 4/20 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
5/20 Checking commit 82f8fd22e61a (ui: remove extra #ifdef CONFIG_OPENGL)
6/20 Checking commit ea06ae968a65 (ui: remove gl_ctx_get_current)
7/20 Checking commit 0843e91e2077 (ui: add gd_gl_area_scanout_disable)
8/20 Checking commit 6f6288374e98 (ui: annotate DCLOps callback requirements)
9/20 Checking commit 851661e80efb (ui: remove console_has_gl_dmabuf())
10/20 Checking commit 24a96b4c5616 (vhost-user-gpu: add a configuration flag for dmabuf usage)
11/20 Checking commit 43e7283e2d2e (ui: add an optional get_flags callback to GraphicHwOps)
ERROR: braces {} are necessary for all arms of this statement
#33: FILE: hw/display/virtio-gpu-base.c:123:
+    if (virtio_gpu_virgl_enabled(g->conf))
[...]

ERROR: braces {} are necessary for all arms of this statement
#36: FILE: hw/display/virtio-gpu-base.c:126:
+    if (virtio_gpu_dmabuf_enabled(g->conf))
[...]

total: 2 errors, 0 warnings, 68 lines checked

Patch 11/20 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.

12/20 Checking commit 483c6affb8c5 (ui: add a DCLOps callback to check dmabuf support)
13/20 Checking commit 6339b5bcfe3b (ui: check hw requirements during DCL registration)
ERROR: line over 90 characters
#21: FILE: ui/console.c:1479:
+static bool dpy_compatible_with(QemuConsole *con, DisplayChangeListener *dcl, Error **errp)

WARNING: line over 80 characters
#34: FILE: ui/console.c:1492:
+    if (flags & GRAPHIC_FLAGS_DMABUF && !displaychangelistener_has_dmabuf(dcl)) {

total: 1 errors, 1 warnings, 45 lines checked

Patch 13/20 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.

14/20 Checking commit 165cebfa7beb (ui: add qemu_egl_has_dmabuf helper)
WARNING: line over 80 characters
#36: FILE: ui/egl-helpers.c:447:
+        !epoxy_has_egl_extension(qemu_egl_display, "EGL_EXT_image_dma_buf_import")) {

total: 0 errors, 1 warnings, 21 lines checked

Patch 14/20 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
15/20 Checking commit 9fe1210ac000 (ui: check gtk-egl dmabuf support)
16/20 Checking commit 6495eb768a5f (ui: add egl dmabuf import to gtkglarea)
17/20 Checking commit 8e9a9d99a7cc (virtio-gpu: avoid re-entering cmdq processing)
18/20 Checking commit b1fb7283b8da (display/ui: add a callback to indicate GL state is flushed)
19/20 Checking commit e3c958379468 (chardev: check if the chardev is registered for yanking)
20/20 Checking commit 22d663059dd3 (RFC: tests: add some virtio-gpu & vhost-user-gpu acceptance test)
WARNING: added, moved or deleted file(s), does MAINTAINERS need updating?
#21: 
new file mode 100644

WARNING: line over 80 characters
#78: FILE: tests/acceptance/virtio-gpu.py:53:
+            self, success_message, failure_message="Kernel panic - not syncing", vm=vm

WARNING: line over 80 characters
#112: FILE: tests/acceptance/virtio-gpu.py:87:
+        exec_command_and_wait_for_pattern(self, "/usr/sbin/modprobe virtio_gpu", "")

WARNING: line over 80 characters
#135: FILE: tests/acceptance/virtio-gpu.py:110:
+        qemu_sock, vug_sock = socket.socketpair(socket.AF_UNIX, socket.SOCK_STREAM)

WARNING: line over 80 characters
#139: FILE: tests/acceptance/virtio-gpu.py:114:
+        self._vug_log_path = os.path.join(self.vm._test_dir, "vhost-user-gpu.log")

WARNING: line over 80 characters
#170: FILE: tests/acceptance/virtio-gpu.py:145:
+        exec_command_and_wait_for_pattern(self, "/usr/sbin/modprobe virtio_gpu", "")

total: 0 errors, 6 warnings, 150 lines checked

Patch 20/20 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
=== OUTPUT END ===

Test command exited with code: 1


The full log is available at
http://patchew.org/logs/20210202142625.609070-1-marcandre.lureau@redhat.com/testing.checkpatch/?type=message.
---
Email generated automatically by Patchew [https://patchew.org/].
Please send your feedback to patchew-devel@redhat.com