* [PULL v2 00/13] Ui patches
@ 2025-07-15 6:32 marcandre.lureau
2025-07-15 6:32 ` [PULL v2 01/13] ui/vnc: Do not copy z_stream marcandre.lureau
` (14 more replies)
0 siblings, 15 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel; +Cc: stefanha, Marc-André Lureau
From: Marc-André Lureau <marcandre.lureau@redhat.com>
The following changes since commit 9a4e273ddec3927920c5958d2226c6b38b543336:
Merge tag 'pull-tcg-20250711' of https://gitlab.com/rth7680/qemu into staging (2025-07-13 01:46:04 -0400)
are available in the Git repository at:
https://gitlab.com/marcandre.lureau/qemu.git tags/ui-pull-request
for you to fetch changes up to df892b3954e5b2782165e6c59e5ffd55c2f7ec5a:
tpm: "qemu -tpmdev help" should return success (2025-07-15 10:22:33 +0400)
----------------------------------------------------------------
UI-related for 10.1
- [PATCH v3 0/2] ui/vnc: Do not copy z_stream
- [PATCH v6 0/7] ui/spice: Enable gl=on option for non-local or remote clients
- [PATCH v6 0/1] Allow injection of virtio-gpu EDID name
- [PATCH 0/2] ui/gtk: Add keep-aspect-ratio and scale option
----------------------------------------------------------------
Akihiko Odaki (2):
ui/vnc: Do not copy z_stream
ui/vnc: Introduce the VncWorker type
Andrew Keesler (1):
hw/display: Allow injection of virtio-gpu EDID name
Marc-André Lureau (1):
tpm: "qemu -tpmdev help" should return success
Vivek Kasireddy (7):
ui/egl-helpers: Error check the fds in egl_dmabuf_export_texture()
ui/spice: Enable gl=on option for non-local or remote clients
ui/spice: Add an option for users to provide a preferred video codec
ui/spice: Add an option to submit gl_draw requests at fixed rate
ui/console-gl: Add a helper to create a texture with linear memory
layout
ui/spice: Create a new texture with linear layout when gl=on is
specified
ui/spice: Blit the scanout texture if its memory layout is not linear
Weifeng Liu (2):
ui/gtk: Add keep-aspect-ratio option
ui/gtk: Add scale option
qapi/ui.json | 15 +-
qapi/virtio.json | 18 +-
include/hw/display/edid.h | 2 +
include/hw/qdev-properties-system.h | 5 +
include/hw/virtio/virtio-gpu.h | 3 +
include/ui/console.h | 3 +
include/ui/gtk.h | 2 +
include/ui/spice-display.h | 5 +
include/ui/surface.h | 1 +
ui/vnc.h | 49 +--
hw/core/qdev-properties-system.c | 44 +++
hw/display/virtio-gpu-base.c | 27 ++
system/tpm.c | 5 +-
ui/console-gl.c | 54 ++++
ui/egl-helpers.c | 6 +
ui/gtk.c | 58 ++--
ui/spice-core.c | 32 ++
ui/spice-display.c | 226 +++++++++++++-
ui/vnc-enc-tight.c | 456 +++++++++++++++-------------
ui/vnc-enc-zlib.c | 47 +--
ui/vnc-enc-zrle.c | 122 ++++----
ui/vnc-jobs.c | 13 +-
ui/vnc.c | 83 +++--
ui/vnc-enc-zrle.c.inc | 20 +-
qemu-options.hx | 13 +
25 files changed, 883 insertions(+), 426 deletions(-)
--
2.50.0
^ permalink raw reply [flat|nested] 20+ messages in thread
* [PULL v2 01/13] ui/vnc: Do not copy z_stream
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-16 15:07 ` Michael Tokarev
2025-07-15 6:32 ` [PULL v2 02/13] ui/vnc: Introduce the VncWorker type marcandre.lureau
` (13 subsequent siblings)
14 siblings, 1 reply; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel; +Cc: stefanha, Akihiko Odaki, Marc-André Lureau
From: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
vnc_worker_thread_loop() copies z_stream stored in its local VncState to
the persistent VncState, and the copied one is freed with deflateEnd()
later. However, deflateEnd() refuses to operate with a copied z_stream
and returns Z_STREAM_ERROR, leaking the allocated memory.
Avoid copying the zlib state to fix the memory leak.
Fixes: bd023f953e5e ("vnc: threaded VNC server")
Signed-off-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250603-zlib-v3-1-20b857bd8d05@rsg.ci.i.u-tokyo.ac.jp>
---
ui/vnc.h | 2 +-
ui/vnc-enc-zlib.c | 30 +++++++++++++++---------------
ui/vnc.c | 13 ++++++++++---
3 files changed, 26 insertions(+), 19 deletions(-)
diff --git a/ui/vnc.h b/ui/vnc.h
index b3e07269bb..8df0cbab25 100644
--- a/ui/vnc.h
+++ b/ui/vnc.h
@@ -342,7 +342,7 @@ struct VncState
* update vnc_async_encoding_start()
*/
VncTight *tight;
- VncZlib zlib;
+ VncZlib *zlib;
VncHextile hextile;
VncZrle *zrle;
VncZywrle zywrle;
diff --git a/ui/vnc-enc-zlib.c b/ui/vnc-enc-zlib.c
index 900ae5b30f..52e9193eab 100644
--- a/ui/vnc-enc-zlib.c
+++ b/ui/vnc-enc-zlib.c
@@ -48,21 +48,21 @@ void vnc_zlib_zfree(void *x, void *addr)
static void vnc_zlib_start(VncState *vs)
{
- buffer_reset(&vs->zlib.zlib);
+ buffer_reset(&vs->zlib->zlib);
// make the output buffer be the zlib buffer, so we can compress it later
- vs->zlib.tmp = vs->output;
- vs->output = vs->zlib.zlib;
+ vs->zlib->tmp = vs->output;
+ vs->output = vs->zlib->zlib;
}
static int vnc_zlib_stop(VncState *vs)
{
- z_streamp zstream = &vs->zlib.stream;
+ z_streamp zstream = &vs->zlib->stream;
int previous_out;
// switch back to normal output/zlib buffers
- vs->zlib.zlib = vs->output;
- vs->output = vs->zlib.tmp;
+ vs->zlib->zlib = vs->output;
+ vs->output = vs->zlib->tmp;
// compress the zlib buffer
@@ -85,24 +85,24 @@ static int vnc_zlib_stop(VncState *vs)
return -1;
}
- vs->zlib.level = vs->tight->compression;
+ vs->zlib->level = vs->tight->compression;
zstream->opaque = vs;
}
- if (vs->tight->compression != vs->zlib.level) {
+ if (vs->tight->compression != vs->zlib->level) {
if (deflateParams(zstream, vs->tight->compression,
Z_DEFAULT_STRATEGY) != Z_OK) {
return -1;
}
- vs->zlib.level = vs->tight->compression;
+ vs->zlib->level = vs->tight->compression;
}
// reserve memory in output buffer
- buffer_reserve(&vs->output, vs->zlib.zlib.offset + 64);
+ buffer_reserve(&vs->output, vs->zlib->zlib.offset + 64);
// set pointers
- zstream->next_in = vs->zlib.zlib.buffer;
- zstream->avail_in = vs->zlib.zlib.offset;
+ zstream->next_in = vs->zlib->zlib.buffer;
+ zstream->avail_in = vs->zlib->zlib.offset;
zstream->next_out = vs->output.buffer + vs->output.offset;
zstream->avail_out = vs->output.capacity - vs->output.offset;
previous_out = zstream->avail_out;
@@ -147,8 +147,8 @@ int vnc_zlib_send_framebuffer_update(VncState *vs, int x, int y, int w, int h)
void vnc_zlib_clear(VncState *vs)
{
- if (vs->zlib.stream.opaque) {
- deflateEnd(&vs->zlib.stream);
+ if (vs->zlib->stream.opaque) {
+ deflateEnd(&vs->zlib->stream);
}
- buffer_free(&vs->zlib.zlib);
+ buffer_free(&vs->zlib->zlib);
}
diff --git a/ui/vnc.c b/ui/vnc.c
index e9c30aad62..ab74154e4c 100644
--- a/ui/vnc.c
+++ b/ui/vnc.c
@@ -56,6 +56,11 @@
#include "io/dns-resolver.h"
#include "monitor/monitor.h"
+typedef struct VncConnection {
+ VncState vs;
+ VncZlib zlib;
+} VncConnection;
+
#define VNC_REFRESH_INTERVAL_BASE GUI_REFRESH_INTERVAL_DEFAULT
#define VNC_REFRESH_INTERVAL_INC 50
#define VNC_REFRESH_INTERVAL_MAX GUI_REFRESH_INTERVAL_IDLE
@@ -1362,7 +1367,7 @@ void vnc_disconnect_finish(VncState *vs)
vs->magic = 0;
g_free(vs->zrle);
g_free(vs->tight);
- g_free(vs);
+ g_free(container_of(vs, VncConnection, vs));
}
size_t vnc_client_io_error(VncState *vs, ssize_t ret, Error *err)
@@ -3241,11 +3246,13 @@ static void vnc_refresh(DisplayChangeListener *dcl)
static void vnc_connect(VncDisplay *vd, QIOChannelSocket *sioc,
bool skipauth, bool websocket)
{
- VncState *vs = g_new0(VncState, 1);
+ VncConnection *vc = g_new0(VncConnection, 1);
+ VncState *vs = &vc->vs;
bool first_client = QTAILQ_EMPTY(&vd->clients);
int i;
trace_vnc_client_connect(vs, sioc);
+ vs->zlib = &vc->zlib;
vs->zrle = g_new0(VncZrle, 1);
vs->tight = g_new0(VncTight, 1);
vs->magic = VNC_MAGIC;
@@ -3268,7 +3275,7 @@ static void vnc_connect(VncDisplay *vd, QIOChannelSocket *sioc,
#ifdef CONFIG_PNG
buffer_init(&vs->tight->png, "vnc-tight-png/%p", sioc);
#endif
- buffer_init(&vs->zlib.zlib, "vnc-zlib/%p", sioc);
+ buffer_init(&vc->zlib.zlib, "vnc-zlib/%p", sioc);
buffer_init(&vs->zrle->zrle, "vnc-zrle/%p", sioc);
buffer_init(&vs->zrle->fb, "vnc-zrle-fb/%p", sioc);
buffer_init(&vs->zrle->zlib, "vnc-zrle-zlib/%p", sioc);
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 02/13] ui/vnc: Introduce the VncWorker type
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
2025-07-15 6:32 ` [PULL v2 01/13] ui/vnc: Do not copy z_stream marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 03/13] ui/egl-helpers: Error check the fds in egl_dmabuf_export_texture() marcandre.lureau
` (12 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel; +Cc: stefanha, Akihiko Odaki, Marc-André Lureau
From: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
The worker thread copies data in VncState to avoid race, but some
data are too big to copy. Such data are held with pointers to avoid
the overhead to copy, but it requires tedious memory management and
makes them vulnerable to race.
Introduce the VncWorker type to contain all data shared without copying.
It allows allocating and freeing all shared data at once and shows that
the race with the worker thread needs to be taken care of when
accessing them.
Signed-off-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20250603-zlib-v3-2-20b857bd8d05@rsg.ci.i.u-tokyo.ac.jp>
---
ui/vnc.h | 49 +++--
ui/vnc-enc-tight.c | 456 ++++++++++++++++++++++--------------------
ui/vnc-enc-zlib.c | 47 ++---
ui/vnc-enc-zrle.c | 122 +++++------
ui/vnc-jobs.c | 13 +-
ui/vnc.c | 86 ++++----
ui/vnc-enc-zrle.c.inc | 20 +-
7 files changed, 405 insertions(+), 388 deletions(-)
diff --git a/ui/vnc.h b/ui/vnc.h
index 8df0cbab25..f2dab2f4d9 100644
--- a/ui/vnc.h
+++ b/ui/vnc.h
@@ -272,8 +272,6 @@ struct VncState
gboolean disconnecting;
DECLARE_BITMAP(dirty[VNC_MAX_HEIGHT], VNC_DIRTY_BITS);
- uint8_t **lossy_rect; /* Not an Array to avoid costly memcpy in
- * vnc-jobs-async.c */
VncDisplay *vd;
VncStateUpdate update; /* Most recent pending request from client */
@@ -341,10 +339,7 @@ struct VncState
/* Encoding specific, if you add something here, don't forget to
* update vnc_async_encoding_start()
*/
- VncTight *tight;
- VncZlib *zlib;
VncHextile hextile;
- VncZrle *zrle;
VncZywrle zywrle;
Notifier mouse_mode_notifier;
@@ -356,6 +351,19 @@ struct VncState
QTAILQ_ENTRY(VncState) next;
};
+typedef struct VncWorker {
+ uint8_t lossy_rect[VNC_STAT_ROWS][VNC_STAT_COLS];
+
+ VncTight tight;
+ VncZlib zlib;
+ VncZrle zrle;
+} VncWorker;
+
+typedef struct VncConnection {
+ VncState vs;
+ VncWorker worker;
+} VncConnection;
+
/*****************************************************************************
*
@@ -602,10 +610,11 @@ int vnc_server_fb_stride(VncDisplay *vd);
void vnc_convert_pixel(VncState *vs, uint8_t *buf, uint32_t v);
double vnc_update_freq(VncState *vs, int x, int y, int w, int h);
-void vnc_sent_lossy_rect(VncState *vs, int x, int y, int w, int h);
+void vnc_sent_lossy_rect(VncWorker *worker, int x, int y, int w, int h);
/* Encodings */
-int vnc_send_framebuffer_update(VncState *vs, int x, int y, int w, int h);
+int vnc_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h);
int vnc_raw_send_framebuffer_update(VncState *vs, int x, int y, int w, int h);
@@ -615,17 +624,21 @@ void vnc_hextile_set_pixel_conversion(VncState *vs, int generic);
void *vnc_zlib_zalloc(void *x, unsigned items, unsigned size);
void vnc_zlib_zfree(void *x, void *addr);
-int vnc_zlib_send_framebuffer_update(VncState *vs, int x, int y, int w, int h);
-void vnc_zlib_clear(VncState *vs);
-
-int vnc_tight_send_framebuffer_update(VncState *vs, int x, int y, int w, int h);
-int vnc_tight_png_send_framebuffer_update(VncState *vs, int x, int y,
- int w, int h);
-void vnc_tight_clear(VncState *vs);
-
-int vnc_zrle_send_framebuffer_update(VncState *vs, int x, int y, int w, int h);
-int vnc_zywrle_send_framebuffer_update(VncState *vs, int x, int y, int w, int h);
-void vnc_zrle_clear(VncState *vs);
+int vnc_zlib_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h);
+void vnc_zlib_clear(VncWorker *worker);
+
+int vnc_tight_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h);
+int vnc_tight_png_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h);
+void vnc_tight_clear(VncWorker *worker);
+
+int vnc_zrle_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h);
+int vnc_zywrle_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h);
+void vnc_zrle_clear(VncWorker *worker);
/* vnc-clipboard.c */
void vnc_server_cut_text_caps(VncState *vs);
diff --git a/ui/vnc-enc-tight.c b/ui/vnc-enc-tight.c
index 25c7b2c788..9dfe6ae5a2 100644
--- a/ui/vnc-enc-tight.c
+++ b/ui/vnc-enc-tight.c
@@ -72,8 +72,8 @@ static const struct {
};
-static int tight_send_framebuffer_update(VncState *vs, int x, int y,
- int w, int h);
+static int tight_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h);
#ifdef CONFIG_VNC_JPEG
static const struct {
@@ -111,12 +111,12 @@ static const struct {
{ 9, PNG_ALL_FILTERS },
};
-static int send_png_rect(VncState *vs, int x, int y, int w, int h,
- VncPalette *palette);
+static int send_png_rect(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h, VncPalette *palette);
-static bool tight_can_send_png_rect(VncState *vs, int w, int h)
+static bool tight_can_send_png_rect(VncState *vs, VncTight *tight, int w, int h)
{
- if (vs->tight->type != VNC_ENCODING_TIGHT_PNG) {
+ if (tight->type != VNC_ENCODING_TIGHT_PNG) {
return false;
}
@@ -135,7 +135,7 @@ static bool tight_can_send_png_rect(VncState *vs, int w, int h)
*/
static unsigned int
-tight_detect_smooth_image24(VncState *vs, int w, int h)
+tight_detect_smooth_image24(VncState *vs, VncTight *tight, int w, int h)
{
int off;
int x, y, d, dx;
@@ -144,7 +144,7 @@ tight_detect_smooth_image24(VncState *vs, int w, int h)
int pixels = 0;
int pix, left[3];
unsigned int errors;
- unsigned char *buf = vs->tight->tight.buffer;
+ unsigned char *buf = tight->tight.buffer;
/*
* If client is big-endian, color samples begin from the second
@@ -205,7 +205,8 @@ tight_detect_smooth_image24(VncState *vs, int w, int h)
#define DEFINE_DETECT_FUNCTION(bpp) \
\
static unsigned int \
- tight_detect_smooth_image##bpp(VncState *vs, int w, int h) { \
+ tight_detect_smooth_image##bpp(VncState *vs, VncTight *tight, \
+ int w, int h) { \
bool endian; \
uint##bpp##_t pix; \
int max[3], shift[3]; \
@@ -215,7 +216,7 @@ tight_detect_smooth_image24(VncState *vs, int w, int h)
int pixels = 0; \
int sample, sum, left[3]; \
unsigned int errors; \
- unsigned char *buf = vs->tight->tight.buffer; \
+ unsigned char *buf = tight->tight.buffer; \
\
endian = 0; /* FIXME */ \
\
@@ -293,11 +294,11 @@ DEFINE_DETECT_FUNCTION(16)
DEFINE_DETECT_FUNCTION(32)
static int
-tight_detect_smooth_image(VncState *vs, int w, int h)
+tight_detect_smooth_image(VncState *vs, VncTight *tight, int w, int h)
{
unsigned int errors;
- int compression = vs->tight->compression;
- int quality = vs->tight->quality;
+ int compression = tight->compression;
+ int quality = tight->quality;
if (!vs->vd->lossy) {
return 0;
@@ -309,7 +310,7 @@ tight_detect_smooth_image(VncState *vs, int w, int h)
return 0;
}
- if (vs->tight->quality != (uint8_t)-1) {
+ if (tight->quality != (uint8_t)-1) {
if (w * h < VNC_TIGHT_JPEG_MIN_RECT_SIZE) {
return 0;
}
@@ -320,17 +321,17 @@ tight_detect_smooth_image(VncState *vs, int w, int h)
}
if (vs->client_pf.bytes_per_pixel == 4) {
- if (vs->tight->pixel24) {
- errors = tight_detect_smooth_image24(vs, w, h);
- if (vs->tight->quality != (uint8_t)-1) {
+ if (tight->pixel24) {
+ errors = tight_detect_smooth_image24(vs, tight, w, h);
+ if (tight->quality != (uint8_t)-1) {
return (errors < tight_conf[quality].jpeg_threshold24);
}
return (errors < tight_conf[compression].gradient_threshold24);
} else {
- errors = tight_detect_smooth_image32(vs, w, h);
+ errors = tight_detect_smooth_image32(vs, tight, w, h);
}
} else {
- errors = tight_detect_smooth_image16(vs, w, h);
+ errors = tight_detect_smooth_image16(vs, tight, w, h);
}
if (quality != (uint8_t)-1) {
return (errors < tight_conf[quality].jpeg_threshold);
@@ -344,15 +345,15 @@ tight_detect_smooth_image(VncState *vs, int w, int h)
#define DEFINE_FILL_PALETTE_FUNCTION(bpp) \
\
static int \
- tight_fill_palette##bpp(VncState *vs, int x, int y, \
- int max, size_t count, \
+ tight_fill_palette##bpp(VncState *vs, VncTight *tight, \
+ int x, int y, int max, size_t count, \
uint32_t *bg, uint32_t *fg, \
VncPalette *palette) { \
uint##bpp##_t *data; \
uint##bpp##_t c0, c1, ci; \
int i, n0, n1; \
\
- data = (uint##bpp##_t *)vs->tight->tight.buffer; \
+ data = (uint##bpp##_t *)tight->tight.buffer; \
\
c0 = data[0]; \
i = 1; \
@@ -417,15 +418,15 @@ DEFINE_FILL_PALETTE_FUNCTION(8)
DEFINE_FILL_PALETTE_FUNCTION(16)
DEFINE_FILL_PALETTE_FUNCTION(32)
-static int tight_fill_palette(VncState *vs, int x, int y,
+static int tight_fill_palette(VncState *vs, VncTight *tight, int x, int y,
size_t count, uint32_t *bg, uint32_t *fg,
VncPalette *palette)
{
int max;
- max = count / tight_conf[vs->tight->compression].idx_max_colors_divisor;
+ max = count / tight_conf[tight->compression].idx_max_colors_divisor;
if (max < 2 &&
- count >= tight_conf[vs->tight->compression].mono_min_rect_size) {
+ count >= tight_conf[tight->compression].mono_min_rect_size) {
max = 2;
}
if (max >= 256) {
@@ -434,12 +435,15 @@ static int tight_fill_palette(VncState *vs, int x, int y,
switch (vs->client_pf.bytes_per_pixel) {
case 4:
- return tight_fill_palette32(vs, x, y, max, count, bg, fg, palette);
+ return tight_fill_palette32(vs, tight, x, y, max, count, bg, fg,
+ palette);
case 2:
- return tight_fill_palette16(vs, x, y, max, count, bg, fg, palette);
+ return tight_fill_palette16(vs, tight, x, y, max, count, bg, fg,
+ palette);
default:
max = 2;
- return tight_fill_palette8(vs, x, y, max, count, bg, fg, palette);
+ return tight_fill_palette8(vs, tight, x, y, max, count, bg, fg,
+ palette);
}
return 0;
}
@@ -547,7 +551,8 @@ DEFINE_MONO_ENCODE_FUNCTION(32)
*/
static void
-tight_filter_gradient24(VncState *vs, uint8_t *buf, int w, int h)
+tight_filter_gradient24(VncState *vs, VncTight *tight, uint8_t *buf,
+ int w, int h)
{
uint32_t *buf32;
uint32_t pix32;
@@ -558,7 +563,7 @@ tight_filter_gradient24(VncState *vs, uint8_t *buf, int w, int h)
int x, y, c;
buf32 = (uint32_t *)buf;
- memset(vs->tight->gradient.buffer, 0, w * 3 * sizeof(int));
+ memset(tight->gradient.buffer, 0, w * 3 * sizeof(int));
if (1 /* FIXME */) {
shift[0] = vs->client_pf.rshift;
@@ -575,7 +580,7 @@ tight_filter_gradient24(VncState *vs, uint8_t *buf, int w, int h)
upper[c] = 0;
here[c] = 0;
}
- prev = (int *)vs->tight->gradient.buffer;
+ prev = (int *)tight->gradient.buffer;
for (x = 0; x < w; x++) {
pix32 = *buf32++;
for (c = 0; c < 3; c++) {
@@ -605,8 +610,8 @@ tight_filter_gradient24(VncState *vs, uint8_t *buf, int w, int h)
#define DEFINE_GRADIENT_FILTER_FUNCTION(bpp) \
\
static void \
- tight_filter_gradient##bpp(VncState *vs, uint##bpp##_t *buf, \
- int w, int h) { \
+ tight_filter_gradient##bpp(VncState *vs, VncTight *tight, \
+ uint##bpp##_t *buf, int w, int h) { \
uint##bpp##_t pix, diff; \
bool endian; \
int *prev; \
@@ -615,7 +620,7 @@ tight_filter_gradient24(VncState *vs, uint8_t *buf, int w, int h)
int prediction; \
int x, y, c; \
\
- memset(vs->tight->gradient.buffer, 0, w * 3 * sizeof(int)); \
+ memset(tight->gradient.buffer, 0, w * 3 * sizeof(int)); \
\
endian = 0; /* FIXME */ \
\
@@ -631,7 +636,7 @@ tight_filter_gradient24(VncState *vs, uint8_t *buf, int w, int h)
upper[c] = 0; \
here[c] = 0; \
} \
- prev = (int *)vs->tight->gradient.buffer; \
+ prev = (int *)tight->gradient.buffer; \
for (x = 0; x < w; x++) { \
pix = *buf; \
if (endian) { \
@@ -782,10 +787,10 @@ static void extend_solid_area(VncState *vs, int x, int y, int w, int h,
*w_ptr += cx - (*x_ptr + *w_ptr);
}
-static int tight_init_stream(VncState *vs, int stream_id,
+static int tight_init_stream(VncState *vs, VncTight *tight, int stream_id,
int level, int strategy)
{
- z_streamp zstream = &vs->tight->stream[stream_id];
+ z_streamp zstream = &tight->stream[stream_id];
if (zstream->opaque == NULL) {
int err;
@@ -803,15 +808,15 @@ static int tight_init_stream(VncState *vs, int stream_id,
return -1;
}
- vs->tight->levels[stream_id] = level;
+ tight->levels[stream_id] = level;
zstream->opaque = vs;
}
- if (vs->tight->levels[stream_id] != level) {
+ if (tight->levels[stream_id] != level) {
if (deflateParams(zstream, level, strategy) != Z_OK) {
return -1;
}
- vs->tight->levels[stream_id] = level;
+ tight->levels[stream_id] = level;
}
return 0;
}
@@ -836,29 +841,29 @@ static void tight_send_compact_size(VncState *vs, size_t len)
}
}
-static int tight_compress_data(VncState *vs, int stream_id, size_t bytes,
- int level, int strategy)
+static int tight_compress_data(VncState *vs, VncTight *tight, int stream_id,
+ size_t bytes, int level, int strategy)
{
- z_streamp zstream = &vs->tight->stream[stream_id];
+ z_streamp zstream = &tight->stream[stream_id];
int previous_out;
if (bytes < VNC_TIGHT_MIN_TO_COMPRESS) {
- vnc_write(vs, vs->tight->tight.buffer, vs->tight->tight.offset);
+ vnc_write(vs, tight->tight.buffer, tight->tight.offset);
return bytes;
}
- if (tight_init_stream(vs, stream_id, level, strategy)) {
+ if (tight_init_stream(vs, tight, stream_id, level, strategy)) {
return -1;
}
/* reserve memory in output buffer */
- buffer_reserve(&vs->tight->zlib, bytes + 64);
+ buffer_reserve(&tight->zlib, bytes + 64);
/* set pointers */
- zstream->next_in = vs->tight->tight.buffer;
- zstream->avail_in = vs->tight->tight.offset;
- zstream->next_out = vs->tight->zlib.buffer + vs->tight->zlib.offset;
- zstream->avail_out = vs->tight->zlib.capacity - vs->tight->zlib.offset;
+ zstream->next_in = tight->tight.buffer;
+ zstream->avail_in = tight->tight.offset;
+ zstream->next_out = tight->zlib.buffer + tight->zlib.offset;
+ zstream->avail_out = tight->zlib.capacity - tight->zlib.offset;
previous_out = zstream->avail_out;
zstream->data_type = Z_BINARY;
@@ -868,14 +873,14 @@ static int tight_compress_data(VncState *vs, int stream_id, size_t bytes,
return -1;
}
- vs->tight->zlib.offset = vs->tight->zlib.capacity - zstream->avail_out;
+ tight->zlib.offset = tight->zlib.capacity - zstream->avail_out;
/* ...how much data has actually been produced by deflate() */
bytes = previous_out - zstream->avail_out;
tight_send_compact_size(vs, bytes);
- vnc_write(vs, vs->tight->zlib.buffer, bytes);
+ vnc_write(vs, tight->zlib.buffer, bytes);
- buffer_reset(&vs->tight->zlib);
+ buffer_reset(&tight->zlib);
return bytes;
}
@@ -914,67 +919,69 @@ static void tight_pack24(VncState *vs, uint8_t *buf, size_t count, size_t *ret)
}
}
-static int send_full_color_rect(VncState *vs, int x, int y, int w, int h)
+static int send_full_color_rect(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
+ VncTight *tight = &worker->tight;
+ int level = tight_conf[tight->compression].raw_zlib_level;
int stream = 0;
ssize_t bytes;
#ifdef CONFIG_PNG
- if (tight_can_send_png_rect(vs, w, h)) {
- return send_png_rect(vs, x, y, w, h, NULL);
+ if (tight_can_send_png_rect(vs, tight, w, h)) {
+ return send_png_rect(vs, worker, x, y, w, h, NULL);
}
#endif
vnc_write_u8(vs, stream << 4); /* no flushing, no filter */
- if (vs->tight->pixel24) {
- tight_pack24(vs, vs->tight->tight.buffer, w * h,
- &vs->tight->tight.offset);
+ if (tight->pixel24) {
+ tight_pack24(vs, tight->tight.buffer, w * h, &tight->tight.offset);
bytes = 3;
} else {
bytes = vs->client_pf.bytes_per_pixel;
}
- bytes = tight_compress_data(vs, stream, w * h * bytes,
- tight_conf[vs->tight->compression].raw_zlib_level,
+ bytes = tight_compress_data(vs, tight, stream, w * h * bytes, level,
Z_DEFAULT_STRATEGY);
return (bytes >= 0);
}
-static int send_solid_rect(VncState *vs)
+static int send_solid_rect(VncState *vs, VncWorker *worker)
{
+ VncTight *tight = &worker->tight;
size_t bytes;
vnc_write_u8(vs, VNC_TIGHT_FILL << 4); /* no flushing, no filter */
- if (vs->tight->pixel24) {
- tight_pack24(vs, vs->tight->tight.buffer, 1, &vs->tight->tight.offset);
+ if (tight->pixel24) {
+ tight_pack24(vs, tight->tight.buffer, 1, &tight->tight.offset);
bytes = 3;
} else {
bytes = vs->client_pf.bytes_per_pixel;
}
- vnc_write(vs, vs->tight->tight.buffer, bytes);
+ vnc_write(vs, tight->tight.buffer, bytes);
return 1;
}
-static int send_mono_rect(VncState *vs, int x, int y,
+static int send_mono_rect(VncState *vs, VncWorker *worker, int x, int y,
int w, int h, uint32_t bg, uint32_t fg)
{
ssize_t bytes;
int stream = 1;
- int level = tight_conf[vs->tight->compression].mono_zlib_level;
+ int level = tight_conf[worker->tight.compression].mono_zlib_level;
#ifdef CONFIG_PNG
- if (tight_can_send_png_rect(vs, w, h)) {
+ if (tight_can_send_png_rect(vs, &worker->tight, w, h)) {
int ret;
int bpp = vs->client_pf.bytes_per_pixel * 8;
VncPalette *palette = palette_new(2, bpp);
palette_put(palette, bg);
palette_put(palette, fg);
- ret = send_png_rect(vs, x, y, w, h, palette);
+ ret = send_png_rect(vs, worker, x, y, w, h, palette);
palette_destroy(palette);
return ret;
}
@@ -992,12 +999,12 @@ static int send_mono_rect(VncState *vs, int x, int y,
uint32_t buf[2] = {bg, fg};
size_t ret = sizeof (buf);
- if (vs->tight->pixel24) {
+ if (worker->tight.pixel24) {
tight_pack24(vs, (unsigned char*)buf, 2, &ret);
}
vnc_write(vs, buf, ret);
- tight_encode_mono_rect32(vs->tight->tight.buffer, w, h, bg, fg);
+ tight_encode_mono_rect32(worker->tight.tight.buffer, w, h, bg, fg);
break;
}
case 2:
@@ -1006,7 +1013,7 @@ static int send_mono_rect(VncState *vs, int x, int y,
uint16_t fg16 = fg;
vnc_write(vs, &bg16, 2);
vnc_write(vs, &fg16, 2);
- tight_encode_mono_rect16(vs->tight->tight.buffer, w, h, bg, fg);
+ tight_encode_mono_rect16(worker->tight.tight.buffer, w, h, bg, fg);
break;
}
default:
@@ -1015,18 +1022,20 @@ static int send_mono_rect(VncState *vs, int x, int y,
uint8_t fg8 = fg;
vnc_write_u8(vs, bg8);
vnc_write_u8(vs, fg8);
- tight_encode_mono_rect8(vs->tight->tight.buffer, w, h, bg, fg);
+ tight_encode_mono_rect8(worker->tight.tight.buffer, w, h, bg, fg);
break;
}
}
- vs->tight->tight.offset = bytes;
+ worker->tight.tight.offset = bytes;
- bytes = tight_compress_data(vs, stream, bytes, level, Z_DEFAULT_STRATEGY);
+ bytes = tight_compress_data(vs, &worker->tight, stream, bytes, level,
+ Z_DEFAULT_STRATEGY);
return (bytes >= 0);
}
struct palette_cb_priv {
VncState *vs;
+ VncTight *tight;
uint8_t *header;
#ifdef CONFIG_PNG
png_colorp png_palette;
@@ -1046,53 +1055,58 @@ static void write_palette(int idx, uint32_t color, void *opaque)
}
}
-static bool send_gradient_rect(VncState *vs, int x, int y, int w, int h)
+static bool send_gradient_rect(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
+ VncTight *tight = &worker->tight;
int stream = 3;
- int level = tight_conf[vs->tight->compression].gradient_zlib_level;
+ int level = tight_conf[tight->compression].gradient_zlib_level;
ssize_t bytes;
if (vs->client_pf.bytes_per_pixel == 1) {
- return send_full_color_rect(vs, x, y, w, h);
+ return send_full_color_rect(vs, worker, x, y, w, h);
}
vnc_write_u8(vs, (stream | VNC_TIGHT_EXPLICIT_FILTER) << 4);
vnc_write_u8(vs, VNC_TIGHT_FILTER_GRADIENT);
- buffer_reserve(&vs->tight->gradient, w * 3 * sizeof(int));
+ buffer_reserve(&tight->gradient, w * 3 * sizeof(int));
- if (vs->tight->pixel24) {
- tight_filter_gradient24(vs, vs->tight->tight.buffer, w, h);
+ if (tight->pixel24) {
+ tight_filter_gradient24(vs, tight, tight->tight.buffer, w, h);
bytes = 3;
} else if (vs->client_pf.bytes_per_pixel == 4) {
- tight_filter_gradient32(vs, (uint32_t *)vs->tight->tight.buffer, w, h);
+ tight_filter_gradient32(vs, tight, (uint32_t *)tight->tight.buffer,
+ w, h);
bytes = 4;
} else {
- tight_filter_gradient16(vs, (uint16_t *)vs->tight->tight.buffer, w, h);
+ tight_filter_gradient16(vs, tight, (uint16_t *)tight->tight.buffer,
+ w, h);
bytes = 2;
}
- buffer_reset(&vs->tight->gradient);
+ buffer_reset(&tight->gradient);
bytes = w * h * bytes;
- vs->tight->tight.offset = bytes;
+ tight->tight.offset = bytes;
- bytes = tight_compress_data(vs, stream, bytes,
+ bytes = tight_compress_data(vs, tight, stream, bytes,
level, Z_FILTERED);
return (bytes >= 0);
}
-static int send_palette_rect(VncState *vs, int x, int y,
+static int send_palette_rect(VncState *vs, VncWorker *worker, int x, int y,
int w, int h, VncPalette *palette)
{
+ VncTight *tight = &worker->tight;
int stream = 2;
- int level = tight_conf[vs->tight->compression].idx_zlib_level;
+ int level = tight_conf[tight->compression].idx_zlib_level;
int colors;
ssize_t bytes;
#ifdef CONFIG_PNG
- if (tight_can_send_png_rect(vs, w, h)) {
- return send_png_rect(vs, x, y, w, h, palette);
+ if (tight_can_send_png_rect(vs, tight, w, h)) {
+ return send_png_rect(vs, worker, x, y, w, h, palette);
}
#endif
@@ -1107,38 +1121,38 @@ static int send_palette_rect(VncState *vs, int x, int y,
{
size_t old_offset, offset, palette_sz = palette_size(palette);
g_autofree uint32_t *header = g_new(uint32_t, palette_sz);
- struct palette_cb_priv priv = { vs, (uint8_t *)header };
+ struct palette_cb_priv priv = { vs, tight, (uint8_t *)header };
old_offset = vs->output.offset;
palette_iter(palette, write_palette, &priv);
vnc_write(vs, header, palette_sz * sizeof(uint32_t));
- if (vs->tight->pixel24) {
+ if (tight->pixel24) {
tight_pack24(vs, vs->output.buffer + old_offset, colors, &offset);
vs->output.offset = old_offset + offset;
}
- tight_encode_indexed_rect32(vs->tight->tight.buffer, w * h, palette);
+ tight_encode_indexed_rect32(tight->tight.buffer, w * h, palette);
break;
}
case 2:
{
size_t palette_sz = palette_size(palette);
g_autofree uint16_t *header = g_new(uint16_t, palette_sz);
- struct palette_cb_priv priv = { vs, (uint8_t *)header };
+ struct palette_cb_priv priv = { vs, tight, (uint8_t *)header };
palette_iter(palette, write_palette, &priv);
vnc_write(vs, header, palette_sz * sizeof(uint16_t));
- tight_encode_indexed_rect16(vs->tight->tight.buffer, w * h, palette);
+ tight_encode_indexed_rect16(tight->tight.buffer, w * h, palette);
break;
}
default:
return -1; /* No palette for 8bits colors */
}
bytes = w * h;
- vs->tight->tight.offset = bytes;
+ tight->tight.offset = bytes;
- bytes = tight_compress_data(vs, stream, bytes,
+ bytes = tight_compress_data(vs, tight, stream, bytes,
level, Z_DEFAULT_STRATEGY);
return (bytes >= 0);
}
@@ -1154,8 +1168,8 @@ static int send_palette_rect(VncState *vs, int x, int y,
/* This is called once per encoding */
static void jpeg_init_destination(j_compress_ptr cinfo)
{
- VncState *vs = cinfo->client_data;
- Buffer *buffer = &vs->tight->jpeg;
+ VncTight *tight = cinfo->client_data;
+ Buffer *buffer = &tight->jpeg;
cinfo->dest->next_output_byte = (JOCTET *)buffer->buffer + buffer->offset;
cinfo->dest->free_in_buffer = (size_t)(buffer->capacity - buffer->offset);
@@ -1164,8 +1178,8 @@ static void jpeg_init_destination(j_compress_ptr cinfo)
/* This is called when we ran out of buffer (shouldn't happen!) */
static boolean jpeg_empty_output_buffer(j_compress_ptr cinfo)
{
- VncState *vs = cinfo->client_data;
- Buffer *buffer = &vs->tight->jpeg;
+ VncTight *tight = cinfo->client_data;
+ Buffer *buffer = &tight->jpeg;
buffer->offset = buffer->capacity;
buffer_reserve(buffer, 2048);
@@ -1176,13 +1190,14 @@ static boolean jpeg_empty_output_buffer(j_compress_ptr cinfo)
/* This is called when we are done processing data */
static void jpeg_term_destination(j_compress_ptr cinfo)
{
- VncState *vs = cinfo->client_data;
- Buffer *buffer = &vs->tight->jpeg;
+ VncTight *tight = cinfo->client_data;
+ Buffer *buffer = &tight->jpeg;
buffer->offset = buffer->capacity - cinfo->dest->free_in_buffer;
}
-static int send_jpeg_rect(VncState *vs, int x, int y, int w, int h, int quality)
+static int send_jpeg_rect(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h, int quality)
{
struct jpeg_compress_struct cinfo;
struct jpeg_error_mgr jerr;
@@ -1193,15 +1208,15 @@ static int send_jpeg_rect(VncState *vs, int x, int y, int w, int h, int quality)
int dy;
if (surface_bytes_per_pixel(vs->vd->ds) == 1) {
- return send_full_color_rect(vs, x, y, w, h);
+ return send_full_color_rect(vs, worker, x, y, w, h);
}
- buffer_reserve(&vs->tight->jpeg, 2048);
+ buffer_reserve(&worker->tight.jpeg, 2048);
cinfo.err = jpeg_std_error(&jerr);
jpeg_create_compress(&cinfo);
- cinfo.client_data = vs;
+ cinfo.client_data = &worker->tight;
cinfo.image_width = w;
cinfo.image_height = h;
cinfo.input_components = 3;
@@ -1231,9 +1246,9 @@ static int send_jpeg_rect(VncState *vs, int x, int y, int w, int h, int quality)
vnc_write_u8(vs, VNC_TIGHT_JPEG << 4);
- tight_send_compact_size(vs, vs->tight->jpeg.offset);
- vnc_write(vs, vs->tight->jpeg.buffer, vs->tight->jpeg.offset);
- buffer_reset(&vs->tight->jpeg);
+ tight_send_compact_size(vs, worker->tight.jpeg.offset);
+ vnc_write(vs, worker->tight.jpeg.buffer, worker->tight.jpeg.offset);
+ buffer_reset(&worker->tight.jpeg);
return 1;
}
@@ -1249,7 +1264,7 @@ static void write_png_palette(int idx, uint32_t pix, void *opaque)
VncState *vs = priv->vs;
png_colorp color = &priv->png_palette[idx];
- if (vs->tight->pixel24)
+ if (priv->tight->pixel24)
{
color->red = (pix >> vs->client_pf.rshift) & vs->client_pf.rmax;
color->green = (pix >> vs->client_pf.gshift) & vs->client_pf.gmax;
@@ -1274,12 +1289,12 @@ static void write_png_palette(int idx, uint32_t pix, void *opaque)
static void png_write_data(png_structp png_ptr, png_bytep data,
png_size_t length)
{
- VncState *vs = png_get_io_ptr(png_ptr);
+ VncWorker *worker = png_get_io_ptr(png_ptr);
- buffer_reserve(&vs->tight->png, vs->tight->png.offset + length);
- memcpy(vs->tight->png.buffer + vs->tight->png.offset, data, length);
+ buffer_reserve(&worker->tight.png, worker->tight.png.offset + length);
+ memcpy(worker->tight.png.buffer + worker->tight.png.offset, data, length);
- vs->tight->png.offset += length;
+ worker->tight.png.offset += length;
}
static void png_flush_data(png_structp png_ptr)
@@ -1296,16 +1311,16 @@ static void vnc_png_free(png_structp png_ptr, png_voidp ptr)
g_free(ptr);
}
-static int send_png_rect(VncState *vs, int x, int y, int w, int h,
- VncPalette *palette)
+static int send_png_rect(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h, VncPalette *palette)
{
png_byte color_type;
png_structp png_ptr;
png_infop info_ptr;
png_colorp png_palette = NULL;
pixman_image_t *linebuf;
- int level = tight_png_conf[vs->tight->compression].png_zlib_level;
- int filters = tight_png_conf[vs->tight->compression].png_filters;
+ int level = tight_png_conf[worker->tight.compression].png_zlib_level;
+ int filters = tight_png_conf[worker->tight.compression].png_filters;
uint8_t *buf;
int dy;
@@ -1322,7 +1337,7 @@ static int send_png_rect(VncState *vs, int x, int y, int w, int h,
return -1;
}
- png_set_write_fn(png_ptr, (void *) vs, png_write_data, png_flush_data);
+ png_set_write_fn(png_ptr, worker, png_write_data, png_flush_data);
png_set_compression_level(png_ptr, level);
png_set_filter(png_ptr, PNG_FILTER_TYPE_DEFAULT, filters);
@@ -1343,29 +1358,30 @@ static int send_png_rect(VncState *vs, int x, int y, int w, int h,
palette_size(palette));
priv.vs = vs;
+ priv.tight = &worker->tight;
priv.png_palette = png_palette;
palette_iter(palette, write_png_palette, &priv);
png_set_PLTE(png_ptr, info_ptr, png_palette, palette_size(palette));
if (vs->client_pf.bytes_per_pixel == 4) {
- tight_encode_indexed_rect32(vs->tight->tight.buffer, w * h,
+ tight_encode_indexed_rect32(worker->tight.tight.buffer, w * h,
palette);
} else {
- tight_encode_indexed_rect16(vs->tight->tight.buffer, w * h,
+ tight_encode_indexed_rect16(worker->tight.tight.buffer, w * h,
palette);
}
}
png_write_info(png_ptr, info_ptr);
- buffer_reserve(&vs->tight->png, 2048);
+ buffer_reserve(&worker->tight.png, 2048);
linebuf = qemu_pixman_linebuf_create(PIXMAN_BE_r8g8b8, w);
buf = (uint8_t *)pixman_image_get_data(linebuf);
for (dy = 0; dy < h; dy++)
{
if (color_type == PNG_COLOR_TYPE_PALETTE) {
- memcpy(buf, vs->tight->tight.buffer + (dy * w), w);
+ memcpy(buf, worker->tight.tight.buffer + (dy * w), w);
} else {
qemu_pixman_linebuf_fill(linebuf, vs->vd->server, w, x, y + dy);
}
@@ -1383,46 +1399,47 @@ static int send_png_rect(VncState *vs, int x, int y, int w, int h,
vnc_write_u8(vs, VNC_TIGHT_PNG << 4);
- tight_send_compact_size(vs, vs->tight->png.offset);
- vnc_write(vs, vs->tight->png.buffer, vs->tight->png.offset);
- buffer_reset(&vs->tight->png);
+ tight_send_compact_size(vs, worker->tight.png.offset);
+ vnc_write(vs, worker->tight.png.buffer, worker->tight.png.offset);
+ buffer_reset(&worker->tight.png);
return 1;
}
#endif /* CONFIG_PNG */
-static void vnc_tight_start(VncState *vs)
+static void vnc_tight_start(VncState *vs, VncTight *tight)
{
- buffer_reset(&vs->tight->tight);
+ buffer_reset(&tight->tight);
// make the output buffer be the zlib buffer, so we can compress it later
- vs->tight->tmp = vs->output;
- vs->output = vs->tight->tight;
+ tight->tmp = vs->output;
+ vs->output = tight->tight;
}
-static void vnc_tight_stop(VncState *vs)
+static void vnc_tight_stop(VncState *vs, VncTight *tight)
{
// switch back to normal output/zlib buffers
- vs->tight->tight = vs->output;
- vs->output = vs->tight->tmp;
+ tight->tight = vs->output;
+ vs->output = tight->tmp;
}
-static int send_sub_rect_nojpeg(VncState *vs, int x, int y, int w, int h,
+static int send_sub_rect_nojpeg(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h,
int bg, int fg, int colors, VncPalette *palette)
{
int ret;
if (colors == 0) {
- if (tight_detect_smooth_image(vs, w, h)) {
- ret = send_gradient_rect(vs, x, y, w, h);
+ if (tight_detect_smooth_image(vs, &worker->tight, w, h)) {
+ ret = send_gradient_rect(vs, worker, x, y, w, h);
} else {
- ret = send_full_color_rect(vs, x, y, w, h);
+ ret = send_full_color_rect(vs, worker, x, y, w, h);
}
} else if (colors == 1) {
- ret = send_solid_rect(vs);
+ ret = send_solid_rect(vs, worker);
} else if (colors == 2) {
- ret = send_mono_rect(vs, x, y, w, h, bg, fg);
+ ret = send_mono_rect(vs, worker, x, y, w, h, bg, fg);
} else if (colors <= 256) {
- ret = send_palette_rect(vs, x, y, w, h, palette);
+ ret = send_palette_rect(vs, worker, x, y, w, h, palette);
} else {
ret = 0;
}
@@ -1430,34 +1447,35 @@ static int send_sub_rect_nojpeg(VncState *vs, int x, int y, int w, int h,
}
#ifdef CONFIG_VNC_JPEG
-static int send_sub_rect_jpeg(VncState *vs, int x, int y, int w, int h,
+static int send_sub_rect_jpeg(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h,
int bg, int fg, int colors,
VncPalette *palette, bool force)
{
int ret;
if (colors == 0) {
- if (force || (tight_jpeg_conf[vs->tight->quality].jpeg_full &&
- tight_detect_smooth_image(vs, w, h))) {
- int quality = tight_conf[vs->tight->quality].jpeg_quality;
+ if (force || (tight_jpeg_conf[worker->tight.quality].jpeg_full &&
+ tight_detect_smooth_image(vs, &worker->tight, w, h))) {
+ int quality = tight_conf[worker->tight.quality].jpeg_quality;
- ret = send_jpeg_rect(vs, x, y, w, h, quality);
+ ret = send_jpeg_rect(vs, worker, x, y, w, h, quality);
} else {
- ret = send_full_color_rect(vs, x, y, w, h);
+ ret = send_full_color_rect(vs, worker, x, y, w, h);
}
} else if (colors == 1) {
- ret = send_solid_rect(vs);
+ ret = send_solid_rect(vs, worker);
} else if (colors == 2) {
- ret = send_mono_rect(vs, x, y, w, h, bg, fg);
+ ret = send_mono_rect(vs, worker, x, y, w, h, bg, fg);
} else if (colors <= 256) {
if (force || (colors > 96 &&
- tight_jpeg_conf[vs->tight->quality].jpeg_idx &&
- tight_detect_smooth_image(vs, w, h))) {
- int quality = tight_conf[vs->tight->quality].jpeg_quality;
+ tight_jpeg_conf[worker->tight.quality].jpeg_idx &&
+ tight_detect_smooth_image(vs, &worker->tight, w, h))) {
+ int quality = tight_conf[worker->tight.quality].jpeg_quality;
- ret = send_jpeg_rect(vs, x, y, w, h, quality);
+ ret = send_jpeg_rect(vs, worker, x, y, w, h, quality);
} else {
- ret = send_palette_rect(vs, x, y, w, h, palette);
+ ret = send_palette_rect(vs, worker, x, y, w, h, palette);
}
} else {
ret = 0;
@@ -1475,8 +1493,10 @@ static void vnc_tight_cleanup(Notifier *n, void *value)
color_count_palette = NULL;
}
-static int send_sub_rect(VncState *vs, int x, int y, int w, int h)
+static int send_sub_rect(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
+ VncTight *tight = &worker->tight;
uint32_t bg = 0, fg = 0;
int colors;
int ret = 0;
@@ -1491,57 +1511,59 @@ static int send_sub_rect(VncState *vs, int x, int y, int w, int h)
qemu_thread_atexit_add(&vnc_tight_cleanup_notifier);
}
- vnc_framebuffer_update(vs, x, y, w, h, vs->tight->type);
+ vnc_framebuffer_update(vs, x, y, w, h, tight->type);
- vnc_tight_start(vs);
+ vnc_tight_start(vs, tight);
vnc_raw_send_framebuffer_update(vs, x, y, w, h);
- vnc_tight_stop(vs);
+ vnc_tight_stop(vs, tight);
#ifdef CONFIG_VNC_JPEG
- if (!vs->vd->non_adaptive && vs->tight->quality != (uint8_t)-1) {
+ if (!vs->vd->non_adaptive && tight->quality != (uint8_t)-1) {
double freq = vnc_update_freq(vs, x, y, w, h);
- if (freq < tight_jpeg_conf[vs->tight->quality].jpeg_freq_min) {
+ if (freq < tight_jpeg_conf[tight->quality].jpeg_freq_min) {
allow_jpeg = false;
}
- if (freq >= tight_jpeg_conf[vs->tight->quality].jpeg_freq_threshold) {
+ if (freq >= tight_jpeg_conf[tight->quality].jpeg_freq_threshold) {
force_jpeg = true;
- vnc_sent_lossy_rect(vs, x, y, w, h);
+ vnc_sent_lossy_rect(worker, x, y, w, h);
}
}
#endif
- colors = tight_fill_palette(vs, x, y, w * h, &bg, &fg, color_count_palette);
+ colors = tight_fill_palette(vs, tight, x, y, w * h, &bg, &fg,
+ color_count_palette);
#ifdef CONFIG_VNC_JPEG
- if (allow_jpeg && vs->tight->quality != (uint8_t)-1) {
- ret = send_sub_rect_jpeg(vs, x, y, w, h, bg, fg, colors,
+ if (allow_jpeg && tight->quality != (uint8_t)-1) {
+ ret = send_sub_rect_jpeg(vs, worker, x, y, w, h, bg, fg, colors,
color_count_palette, force_jpeg);
} else {
- ret = send_sub_rect_nojpeg(vs, x, y, w, h, bg, fg, colors,
- color_count_palette);
+ ret = send_sub_rect_nojpeg(vs, worker, x, y, w, h, bg, fg,
+ colors, color_count_palette);
}
#else
- ret = send_sub_rect_nojpeg(vs, x, y, w, h, bg, fg, colors,
+ ret = send_sub_rect_nojpeg(vs, worker, x, y, w, h, bg, fg, colors,
color_count_palette);
#endif
return ret;
}
-static int send_sub_rect_solid(VncState *vs, int x, int y, int w, int h)
+static int send_sub_rect_solid(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
- vnc_framebuffer_update(vs, x, y, w, h, vs->tight->type);
+ vnc_framebuffer_update(vs, x, y, w, h, worker->tight.type);
- vnc_tight_start(vs);
+ vnc_tight_start(vs, &worker->tight);
vnc_raw_send_framebuffer_update(vs, x, y, w, h);
- vnc_tight_stop(vs);
+ vnc_tight_stop(vs, &worker->tight);
- return send_solid_rect(vs);
+ return send_solid_rect(vs, worker);
}
-static int send_rect_simple(VncState *vs, int x, int y, int w, int h,
- bool split)
+static int send_rect_simple(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h, bool split)
{
int max_size, max_width;
int max_sub_width, max_sub_height;
@@ -1549,8 +1571,8 @@ static int send_rect_simple(VncState *vs, int x, int y, int w, int h,
int rw, rh;
int n = 0;
- max_size = tight_conf[vs->tight->compression].max_rect_size;
- max_width = tight_conf[vs->tight->compression].max_rect_width;
+ max_size = tight_conf[worker->tight.compression].max_rect_size;
+ max_width = tight_conf[worker->tight.compression].max_rect_width;
if (split && (w > max_width || w * h > max_size)) {
max_sub_width = (w > max_width) ? max_width : w;
@@ -1560,18 +1582,18 @@ static int send_rect_simple(VncState *vs, int x, int y, int w, int h,
for (dx = 0; dx < w; dx += max_width) {
rw = MIN(max_sub_width, w - dx);
rh = MIN(max_sub_height, h - dy);
- n += send_sub_rect(vs, x+dx, y+dy, rw, rh);
+ n += send_sub_rect(vs, worker, x + dx, y + dy, rw, rh);
}
}
} else {
- n += send_sub_rect(vs, x, y, w, h);
+ n += send_sub_rect(vs, worker, x, y, w, h);
}
return n;
}
-static int find_large_solid_color_rect(VncState *vs, int x, int y,
- int w, int h, int max_rows)
+static int find_large_solid_color_rect(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h, int max_rows)
{
int dx, dy, dw, dh;
int n = 0;
@@ -1583,7 +1605,7 @@ static int find_large_solid_color_rect(VncState *vs, int x, int y,
/* If a rectangle becomes too large, send its upper part now. */
if (dy - y >= max_rows) {
- n += send_rect_simple(vs, x, y, w, max_rows, true);
+ n += send_rect_simple(vs, worker, x, y, w, max_rows, true);
y += max_rows;
h -= max_rows;
}
@@ -1622,26 +1644,28 @@ static int find_large_solid_color_rect(VncState *vs, int x, int y,
/* Send rectangles at top and left to solid-color area. */
if (y_best != y) {
- n += send_rect_simple(vs, x, y, w, y_best-y, true);
+ n += send_rect_simple(vs, worker, x, y, w, y_best - y, true);
}
if (x_best != x) {
- n += tight_send_framebuffer_update(vs, x, y_best,
+ n += tight_send_framebuffer_update(vs, worker, x, y_best,
x_best-x, h_best);
}
/* Send solid-color rectangle. */
- n += send_sub_rect_solid(vs, x_best, y_best, w_best, h_best);
+ n += send_sub_rect_solid(vs, worker,
+ x_best, y_best, w_best, h_best);
/* Send remaining rectangles (at right and bottom). */
if (x_best + w_best != x + w) {
- n += tight_send_framebuffer_update(vs, x_best+w_best,
+ n += tight_send_framebuffer_update(vs, worker, x_best + w_best,
y_best,
w-(x_best-x)-w_best,
h_best);
}
if (y_best + h_best != y + h) {
- n += tight_send_framebuffer_update(vs, x, y_best+h_best,
+ n += tight_send_framebuffer_update(vs, worker,
+ x, y_best + h_best,
w, h-(y_best-y)-h_best);
}
@@ -1649,73 +1673,73 @@ static int find_large_solid_color_rect(VncState *vs, int x, int y,
return n;
}
}
- return n + send_rect_simple(vs, x, y, w, h, true);
+ return n + send_rect_simple(vs, worker, x, y, w, h, true);
}
-static int tight_send_framebuffer_update(VncState *vs, int x, int y,
- int w, int h)
+static int tight_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
int max_rows;
if (vs->client_pf.bytes_per_pixel == 4 && vs->client_pf.rmax == 0xFF &&
vs->client_pf.bmax == 0xFF && vs->client_pf.gmax == 0xFF) {
- vs->tight->pixel24 = true;
+ worker->tight.pixel24 = true;
} else {
- vs->tight->pixel24 = false;
+ worker->tight.pixel24 = false;
}
#ifdef CONFIG_VNC_JPEG
- if (vs->tight->quality != (uint8_t)-1) {
+ if (worker->tight.quality != (uint8_t)-1) {
double freq = vnc_update_freq(vs, x, y, w, h);
- if (freq > tight_jpeg_conf[vs->tight->quality].jpeg_freq_threshold) {
- return send_rect_simple(vs, x, y, w, h, false);
+ if (freq > tight_jpeg_conf[worker->tight.quality].jpeg_freq_threshold) {
+ return send_rect_simple(vs, worker, x, y, w, h, false);
}
}
#endif
if (w * h < VNC_TIGHT_MIN_SPLIT_RECT_SIZE) {
- return send_rect_simple(vs, x, y, w, h, true);
+ return send_rect_simple(vs, worker, x, y, w, h, true);
}
/* Calculate maximum number of rows in one non-solid rectangle. */
- max_rows = tight_conf[vs->tight->compression].max_rect_size;
- max_rows /= MIN(tight_conf[vs->tight->compression].max_rect_width, w);
+ max_rows = tight_conf[worker->tight.compression].max_rect_size;
+ max_rows /= MIN(tight_conf[worker->tight.compression].max_rect_width, w);
- return find_large_solid_color_rect(vs, x, y, w, h, max_rows);
+ return find_large_solid_color_rect(vs, worker, x, y, w, h, max_rows);
}
-int vnc_tight_send_framebuffer_update(VncState *vs, int x, int y,
- int w, int h)
+int vnc_tight_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
- vs->tight->type = VNC_ENCODING_TIGHT;
- return tight_send_framebuffer_update(vs, x, y, w, h);
+ worker->tight.type = VNC_ENCODING_TIGHT;
+ return tight_send_framebuffer_update(vs, worker, x, y, w, h);
}
-int vnc_tight_png_send_framebuffer_update(VncState *vs, int x, int y,
- int w, int h)
+int vnc_tight_png_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
- vs->tight->type = VNC_ENCODING_TIGHT_PNG;
- return tight_send_framebuffer_update(vs, x, y, w, h);
+ worker->tight.type = VNC_ENCODING_TIGHT_PNG;
+ return tight_send_framebuffer_update(vs, worker, x, y, w, h);
}
-void vnc_tight_clear(VncState *vs)
+void vnc_tight_clear(VncWorker *worker)
{
int i;
- for (i = 0; i < ARRAY_SIZE(vs->tight->stream); i++) {
- if (vs->tight->stream[i].opaque) {
- deflateEnd(&vs->tight->stream[i]);
+ for (i = 0; i < ARRAY_SIZE(worker->tight.stream); i++) {
+ if (worker->tight.stream[i].opaque) {
+ deflateEnd(&worker->tight.stream[i]);
}
}
- buffer_free(&vs->tight->tight);
- buffer_free(&vs->tight->zlib);
- buffer_free(&vs->tight->gradient);
+ buffer_free(&worker->tight.tight);
+ buffer_free(&worker->tight.zlib);
+ buffer_free(&worker->tight.gradient);
#ifdef CONFIG_VNC_JPEG
- buffer_free(&vs->tight->jpeg);
+ buffer_free(&worker->tight.jpeg);
#endif
#ifdef CONFIG_PNG
- buffer_free(&vs->tight->png);
+ buffer_free(&worker->tight.png);
#endif
}
diff --git a/ui/vnc-enc-zlib.c b/ui/vnc-enc-zlib.c
index 52e9193eab..a6d287118a 100644
--- a/ui/vnc-enc-zlib.c
+++ b/ui/vnc-enc-zlib.c
@@ -46,23 +46,23 @@ void vnc_zlib_zfree(void *x, void *addr)
g_free(addr);
}
-static void vnc_zlib_start(VncState *vs)
+static void vnc_zlib_start(VncState *vs, VncWorker *worker)
{
- buffer_reset(&vs->zlib->zlib);
+ buffer_reset(&worker->zlib.zlib);
// make the output buffer be the zlib buffer, so we can compress it later
- vs->zlib->tmp = vs->output;
- vs->output = vs->zlib->zlib;
+ worker->zlib.tmp = vs->output;
+ vs->output = worker->zlib.zlib;
}
-static int vnc_zlib_stop(VncState *vs)
+static int vnc_zlib_stop(VncState *vs, VncWorker *worker)
{
- z_streamp zstream = &vs->zlib->stream;
+ z_streamp zstream = &worker->zlib.stream;
int previous_out;
// switch back to normal output/zlib buffers
- vs->zlib->zlib = vs->output;
- vs->output = vs->zlib->tmp;
+ worker->zlib.zlib = vs->output;
+ vs->output = worker->zlib.tmp;
// compress the zlib buffer
@@ -76,7 +76,7 @@ static int vnc_zlib_stop(VncState *vs)
zstream->zalloc = vnc_zlib_zalloc;
zstream->zfree = vnc_zlib_zfree;
- err = deflateInit2(zstream, vs->tight->compression, Z_DEFLATED,
+ err = deflateInit2(zstream, worker->tight.compression, Z_DEFLATED,
MAX_WBITS,
MAX_MEM_LEVEL, Z_DEFAULT_STRATEGY);
@@ -85,24 +85,24 @@ static int vnc_zlib_stop(VncState *vs)
return -1;
}
- vs->zlib->level = vs->tight->compression;
+ worker->zlib.level = worker->tight.compression;
zstream->opaque = vs;
}
- if (vs->tight->compression != vs->zlib->level) {
- if (deflateParams(zstream, vs->tight->compression,
+ if (worker->tight.compression != worker->zlib.level) {
+ if (deflateParams(zstream, worker->tight.compression,
Z_DEFAULT_STRATEGY) != Z_OK) {
return -1;
}
- vs->zlib->level = vs->tight->compression;
+ worker->zlib.level = worker->tight.compression;
}
// reserve memory in output buffer
- buffer_reserve(&vs->output, vs->zlib->zlib.offset + 64);
+ buffer_reserve(&vs->output, worker->zlib.zlib.offset + 64);
// set pointers
- zstream->next_in = vs->zlib->zlib.buffer;
- zstream->avail_in = vs->zlib->zlib.offset;
+ zstream->next_in = worker->zlib.zlib.buffer;
+ zstream->avail_in = worker->zlib.zlib.offset;
zstream->next_out = vs->output.buffer + vs->output.offset;
zstream->avail_out = vs->output.capacity - vs->output.offset;
previous_out = zstream->avail_out;
@@ -118,7 +118,8 @@ static int vnc_zlib_stop(VncState *vs)
return previous_out - zstream->avail_out;
}
-int vnc_zlib_send_framebuffer_update(VncState *vs, int x, int y, int w, int h)
+int vnc_zlib_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
int old_offset, new_offset, bytes_written;
@@ -129,9 +130,9 @@ int vnc_zlib_send_framebuffer_update(VncState *vs, int x, int y, int w, int h)
vnc_write_s32(vs, 0);
// compress the stream
- vnc_zlib_start(vs);
+ vnc_zlib_start(vs, worker);
vnc_raw_send_framebuffer_update(vs, x, y, w, h);
- bytes_written = vnc_zlib_stop(vs);
+ bytes_written = vnc_zlib_stop(vs, worker);
if (bytes_written == -1)
return 0;
@@ -145,10 +146,10 @@ int vnc_zlib_send_framebuffer_update(VncState *vs, int x, int y, int w, int h)
return 1;
}
-void vnc_zlib_clear(VncState *vs)
+void vnc_zlib_clear(VncWorker *worker)
{
- if (vs->zlib->stream.opaque) {
- deflateEnd(&vs->zlib->stream);
+ if (worker->zlib.stream.opaque) {
+ deflateEnd(&worker->zlib.stream);
}
- buffer_free(&vs->zlib->zlib);
+ buffer_free(&worker->zlib.zlib);
}
diff --git a/ui/vnc-enc-zrle.c b/ui/vnc-enc-zrle.c
index 97ec6c7119..7679014c9e 100644
--- a/ui/vnc-enc-zrle.c
+++ b/ui/vnc-enc-zrle.c
@@ -35,45 +35,45 @@ static const int bits_per_packed_pixel[] = {
};
-static void vnc_zrle_start(VncState *vs)
+static void vnc_zrle_start(VncState *vs, VncZrle *zrle)
{
- buffer_reset(&vs->zrle->zrle);
+ buffer_reset(&zrle->zrle);
/* make the output buffer be the zlib buffer, so we can compress it later */
- vs->zrle->tmp = vs->output;
- vs->output = vs->zrle->zrle;
+ zrle->tmp = vs->output;
+ vs->output = zrle->zrle;
}
-static void vnc_zrle_stop(VncState *vs)
+static void vnc_zrle_stop(VncState *vs, VncZrle *zrle)
{
/* switch back to normal output/zlib buffers */
- vs->zrle->zrle = vs->output;
- vs->output = vs->zrle->tmp;
+ zrle->zrle = vs->output;
+ vs->output = zrle->tmp;
}
-static void *zrle_convert_fb(VncState *vs, int x, int y, int w, int h,
- int bpp)
+static void *zrle_convert_fb(VncState *vs, VncZrle *zrle,
+ int x, int y, int w, int h, int bpp)
{
Buffer tmp;
- buffer_reset(&vs->zrle->fb);
- buffer_reserve(&vs->zrle->fb, w * h * bpp + bpp);
+ buffer_reset(&zrle->fb);
+ buffer_reserve(&zrle->fb, w * h * bpp + bpp);
tmp = vs->output;
- vs->output = vs->zrle->fb;
+ vs->output = zrle->fb;
vnc_raw_send_framebuffer_update(vs, x, y, w, h);
- vs->zrle->fb = vs->output;
+ zrle->fb = vs->output;
vs->output = tmp;
- return vs->zrle->fb.buffer;
+ return zrle->fb.buffer;
}
-static int zrle_compress_data(VncState *vs, int level)
+static int zrle_compress_data(VncState *vs, VncZrle *zrle, int level)
{
- z_streamp zstream = &vs->zrle->stream;
+ z_streamp zstream = &zrle->stream;
- buffer_reset(&vs->zrle->zlib);
+ buffer_reset(&zrle->zlib);
if (zstream->opaque != vs) {
int err;
@@ -93,13 +93,13 @@ static int zrle_compress_data(VncState *vs, int level)
}
/* reserve memory in output buffer */
- buffer_reserve(&vs->zrle->zlib, vs->zrle->zrle.offset + 64);
+ buffer_reserve(&zrle->zlib, zrle->zrle.offset + 64);
/* set pointers */
- zstream->next_in = vs->zrle->zrle.buffer;
- zstream->avail_in = vs->zrle->zrle.offset;
- zstream->next_out = vs->zrle->zlib.buffer;
- zstream->avail_out = vs->zrle->zlib.capacity;
+ zstream->next_in = zrle->zrle.buffer;
+ zstream->avail_in = zrle->zrle.offset;
+ zstream->next_out = zrle->zlib.buffer;
+ zstream->avail_out = zrle->zlib.capacity;
zstream->data_type = Z_BINARY;
/* start encoding */
@@ -108,8 +108,8 @@ static int zrle_compress_data(VncState *vs, int level)
return -1;
}
- vs->zrle->zlib.offset = vs->zrle->zlib.capacity - zstream->avail_out;
- return vs->zrle->zlib.offset;
+ zrle->zlib.offset = zrle->zlib.capacity - zstream->avail_out;
+ return zrle->zlib.offset;
}
/* Try to work out whether to use RLE and/or a palette. We do this by
@@ -252,21 +252,21 @@ static void zrle_write_u8(VncState *vs, uint8_t value)
#undef ZRLE_COMPACT_PIXEL
#undef ZRLE_BPP
-static int zrle_send_framebuffer_update(VncState *vs, int x, int y,
- int w, int h)
+static int zrle_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
bool be = vs->client_endian == G_BIG_ENDIAN;
size_t bytes;
int zywrle_level;
- if (vs->zrle->type == VNC_ENCODING_ZYWRLE) {
- if (!vs->vd->lossy || vs->tight->quality == (uint8_t)-1
- || vs->tight->quality == 9) {
+ if (worker->zrle.type == VNC_ENCODING_ZYWRLE) {
+ if (!vs->vd->lossy || worker->tight.quality == (uint8_t)-1
+ || worker->tight.quality == 9) {
zywrle_level = 0;
- vs->zrle->type = VNC_ENCODING_ZRLE;
- } else if (vs->tight->quality < 3) {
+ worker->zrle.type = VNC_ENCODING_ZRLE;
+ } else if (worker->tight.quality < 3) {
zywrle_level = 3;
- } else if (vs->tight->quality < 6) {
+ } else if (worker->tight.quality < 6) {
zywrle_level = 2;
} else {
zywrle_level = 1;
@@ -275,25 +275,25 @@ static int zrle_send_framebuffer_update(VncState *vs, int x, int y,
zywrle_level = 0;
}
- vnc_zrle_start(vs);
+ vnc_zrle_start(vs, &worker->zrle);
switch (vs->client_pf.bytes_per_pixel) {
case 1:
- zrle_encode_8ne(vs, x, y, w, h, zywrle_level);
+ zrle_encode_8ne(vs, &worker->zrle, x, y, w, h, zywrle_level);
break;
case 2:
if (vs->client_pf.gmax > 0x1F) {
if (be) {
- zrle_encode_16be(vs, x, y, w, h, zywrle_level);
+ zrle_encode_16be(vs, &worker->zrle, x, y, w, h, zywrle_level);
} else {
- zrle_encode_16le(vs, x, y, w, h, zywrle_level);
+ zrle_encode_16le(vs, &worker->zrle, x, y, w, h, zywrle_level);
}
} else {
if (be) {
- zrle_encode_15be(vs, x, y, w, h, zywrle_level);
+ zrle_encode_15be(vs, &worker->zrle, x, y, w, h, zywrle_level);
} else {
- zrle_encode_15le(vs, x, y, w, h, zywrle_level);
+ zrle_encode_15le(vs, &worker->zrle, x, y, w, h, zywrle_level);
}
}
break;
@@ -314,53 +314,55 @@ static int zrle_send_framebuffer_update(VncState *vs, int x, int y,
if ((fits_in_ls3bytes && !be) || (fits_in_ms3bytes && be)) {
if (be) {
- zrle_encode_24abe(vs, x, y, w, h, zywrle_level);
+ zrle_encode_24abe(vs, &worker->zrle, x, y, w, h, zywrle_level);
} else {
- zrle_encode_24ale(vs, x, y, w, h, zywrle_level);
+ zrle_encode_24ale(vs, &worker->zrle, x, y, w, h, zywrle_level);
}
} else if ((fits_in_ls3bytes && be) || (fits_in_ms3bytes && !be)) {
if (be) {
- zrle_encode_24bbe(vs, x, y, w, h, zywrle_level);
+ zrle_encode_24bbe(vs, &worker->zrle, x, y, w, h, zywrle_level);
} else {
- zrle_encode_24ble(vs, x, y, w, h, zywrle_level);
+ zrle_encode_24ble(vs, &worker->zrle, x, y, w, h, zywrle_level);
}
} else {
if (be) {
- zrle_encode_32be(vs, x, y, w, h, zywrle_level);
+ zrle_encode_32be(vs, &worker->zrle, x, y, w, h, zywrle_level);
} else {
- zrle_encode_32le(vs, x, y, w, h, zywrle_level);
+ zrle_encode_32le(vs, &worker->zrle, x, y, w, h, zywrle_level);
}
}
}
break;
}
- vnc_zrle_stop(vs);
- bytes = zrle_compress_data(vs, Z_DEFAULT_COMPRESSION);
- vnc_framebuffer_update(vs, x, y, w, h, vs->zrle->type);
+ vnc_zrle_stop(vs, &worker->zrle);
+ bytes = zrle_compress_data(vs, &worker->zrle, Z_DEFAULT_COMPRESSION);
+ vnc_framebuffer_update(vs, x, y, w, h, worker->zrle.type);
vnc_write_u32(vs, bytes);
- vnc_write(vs, vs->zrle->zlib.buffer, vs->zrle->zlib.offset);
+ vnc_write(vs, worker->zrle.zlib.buffer, worker->zrle.zlib.offset);
return 1;
}
-int vnc_zrle_send_framebuffer_update(VncState *vs, int x, int y, int w, int h)
+int vnc_zrle_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
- vs->zrle->type = VNC_ENCODING_ZRLE;
- return zrle_send_framebuffer_update(vs, x, y, w, h);
+ worker->zrle.type = VNC_ENCODING_ZRLE;
+ return zrle_send_framebuffer_update(vs, worker, x, y, w, h);
}
-int vnc_zywrle_send_framebuffer_update(VncState *vs, int x, int y, int w, int h)
+int vnc_zywrle_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
- vs->zrle->type = VNC_ENCODING_ZYWRLE;
- return zrle_send_framebuffer_update(vs, x, y, w, h);
+ worker->zrle.type = VNC_ENCODING_ZYWRLE;
+ return zrle_send_framebuffer_update(vs, worker, x, y, w, h);
}
-void vnc_zrle_clear(VncState *vs)
+void vnc_zrle_clear(VncWorker *worker)
{
- if (vs->zrle->stream.opaque) {
- deflateEnd(&vs->zrle->stream);
+ if (worker->zrle.stream.opaque) {
+ deflateEnd(&worker->zrle.stream);
}
- buffer_free(&vs->zrle->zrle);
- buffer_free(&vs->zrle->fb);
- buffer_free(&vs->zrle->zlib);
+ buffer_free(&worker->zrle.zrle);
+ buffer_free(&worker->zrle.fb);
+ buffer_free(&worker->zrle.zlib);
}
diff --git a/ui/vnc-jobs.c b/ui/vnc-jobs.c
index d3486af9e2..bed33950a8 100644
--- a/ui/vnc-jobs.c
+++ b/ui/vnc-jobs.c
@@ -185,14 +185,10 @@ static void vnc_async_encoding_start(VncState *orig, VncState *local)
local->vnc_encoding = orig->vnc_encoding;
local->features = orig->features;
local->vd = orig->vd;
- local->lossy_rect = orig->lossy_rect;
local->write_pixels = orig->write_pixels;
local->client_pf = orig->client_pf;
local->client_endian = orig->client_endian;
- local->tight = orig->tight;
- local->zlib = orig->zlib;
local->hextile = orig->hextile;
- local->zrle = orig->zrle;
local->client_width = orig->client_width;
local->client_height = orig->client_height;
}
@@ -200,11 +196,7 @@ static void vnc_async_encoding_start(VncState *orig, VncState *local)
static void vnc_async_encoding_end(VncState *orig, VncState *local)
{
buffer_free(&local->output);
- orig->tight = local->tight;
- orig->zlib = local->zlib;
orig->hextile = local->hextile;
- orig->zrle = local->zrle;
- orig->lossy_rect = local->lossy_rect;
}
static bool vnc_worker_clamp_rect(VncState *vs, VncJob *job, VncRect *rect)
@@ -237,6 +229,7 @@ static bool vnc_worker_clamp_rect(VncState *vs, VncJob *job, VncRect *rect)
static int vnc_worker_thread_loop(VncJobQueue *queue)
{
+ VncConnection *vc;
VncJob *job;
VncRectEntry *entry, *tmp;
VncState vs = {};
@@ -256,6 +249,7 @@ static int vnc_worker_thread_loop(VncJobQueue *queue)
}
assert(job->vs->magic == VNC_MAGIC);
+ vc = container_of(job->vs, VncConnection, vs);
vnc_lock_output(job->vs);
if (job->vs->ioc == NULL || job->vs->abort == true) {
@@ -295,7 +289,8 @@ static int vnc_worker_thread_loop(VncJobQueue *queue)
}
if (vnc_worker_clamp_rect(&vs, job, &entry->rect)) {
- n = vnc_send_framebuffer_update(&vs, entry->rect.x, entry->rect.y,
+ n = vnc_send_framebuffer_update(&vs, &vc->worker,
+ entry->rect.x, entry->rect.y,
entry->rect.w, entry->rect.h);
if (n >= 0) {
diff --git a/ui/vnc.c b/ui/vnc.c
index ab74154e4c..1df35832d5 100644
--- a/ui/vnc.c
+++ b/ui/vnc.c
@@ -56,11 +56,6 @@
#include "io/dns-resolver.h"
#include "monitor/monitor.h"
-typedef struct VncConnection {
- VncState vs;
- VncZlib zlib;
-} VncConnection;
-
#define VNC_REFRESH_INTERVAL_BASE GUI_REFRESH_INTERVAL_DEFAULT
#define VNC_REFRESH_INTERVAL_INC 50
#define VNC_REFRESH_INTERVAL_MAX GUI_REFRESH_INTERVAL_IDLE
@@ -951,29 +946,30 @@ int vnc_raw_send_framebuffer_update(VncState *vs, int x, int y, int w, int h)
return 1;
}
-int vnc_send_framebuffer_update(VncState *vs, int x, int y, int w, int h)
+int vnc_send_framebuffer_update(VncState *vs, VncWorker *worker,
+ int x, int y, int w, int h)
{
int n = 0;
switch(vs->vnc_encoding) {
case VNC_ENCODING_ZLIB:
- n = vnc_zlib_send_framebuffer_update(vs, x, y, w, h);
+ n = vnc_zlib_send_framebuffer_update(vs, worker, x, y, w, h);
break;
case VNC_ENCODING_HEXTILE:
vnc_framebuffer_update(vs, x, y, w, h, VNC_ENCODING_HEXTILE);
n = vnc_hextile_send_framebuffer_update(vs, x, y, w, h);
break;
case VNC_ENCODING_TIGHT:
- n = vnc_tight_send_framebuffer_update(vs, x, y, w, h);
+ n = vnc_tight_send_framebuffer_update(vs, worker, x, y, w, h);
break;
case VNC_ENCODING_TIGHT_PNG:
- n = vnc_tight_png_send_framebuffer_update(vs, x, y, w, h);
+ n = vnc_tight_png_send_framebuffer_update(vs, worker, x, y, w, h);
break;
case VNC_ENCODING_ZRLE:
- n = vnc_zrle_send_framebuffer_update(vs, x, y, w, h);
+ n = vnc_zrle_send_framebuffer_update(vs, worker, x, y, w, h);
break;
case VNC_ENCODING_ZYWRLE:
- n = vnc_zywrle_send_framebuffer_update(vs, x, y, w, h);
+ n = vnc_zywrle_send_framebuffer_update(vs, worker, x, y, w, h);
break;
default:
vnc_framebuffer_update(vs, x, y, w, h, VNC_ENCODING_RAW);
@@ -1311,7 +1307,7 @@ static void vnc_disconnect_start(VncState *vs)
void vnc_disconnect_finish(VncState *vs)
{
- int i;
+ VncConnection *vc = container_of(vs, VncConnection, vs);
trace_vnc_client_disconnect_finish(vs, vs->ioc);
@@ -1325,9 +1321,9 @@ void vnc_disconnect_finish(VncState *vs)
qapi_free_VncClientInfo(vs->info);
- vnc_zlib_clear(vs);
- vnc_tight_clear(vs);
- vnc_zrle_clear(vs);
+ vnc_zlib_clear(&vc->worker);
+ vnc_tight_clear(&vc->worker);
+ vnc_zrle_clear(&vc->worker);
#ifdef CONFIG_VNC_SASL
vnc_sasl_client_cleanup(vs);
@@ -1355,19 +1351,12 @@ void vnc_disconnect_finish(VncState *vs)
}
buffer_free(&vs->jobs_buffer);
- for (i = 0; i < VNC_STAT_ROWS; ++i) {
- g_free(vs->lossy_rect[i]);
- }
- g_free(vs->lossy_rect);
-
object_unref(OBJECT(vs->ioc));
vs->ioc = NULL;
object_unref(OBJECT(vs->sioc));
vs->sioc = NULL;
vs->magic = 0;
- g_free(vs->zrle);
- g_free(vs->tight);
- g_free(container_of(vs, VncConnection, vs));
+ g_free(vc);
}
size_t vnc_client_io_error(VncState *vs, ssize_t ret, Error *err)
@@ -2131,13 +2120,14 @@ static void send_xvp_message(VncState *vs, int code)
static void set_encodings(VncState *vs, int32_t *encodings, size_t n_encodings)
{
+ VncConnection *vc = container_of(vs, VncConnection, vs);
int i;
unsigned int enc = 0;
vs->features = 0;
vs->vnc_encoding = 0;
- vs->tight->compression = 9;
- vs->tight->quality = -1; /* Lossless by default */
+ vc->worker.tight.compression = 9;
+ vc->worker.tight.quality = -1; /* Lossless by default */
vs->absolute = -1;
/*
@@ -2225,11 +2215,11 @@ static void set_encodings(VncState *vs, int32_t *encodings, size_t n_encodings)
vnc_server_cut_text_caps(vs);
break;
case VNC_ENCODING_COMPRESSLEVEL0 ... VNC_ENCODING_COMPRESSLEVEL0 + 9:
- vs->tight->compression = (enc & 0x0F);
+ vc->worker.tight.compression = (enc & 0x0F);
break;
case VNC_ENCODING_QUALITYLEVEL0 ... VNC_ENCODING_QUALITYLEVEL0 + 9:
if (vs->vd->lossy) {
- vs->tight->quality = (enc & 0x0F);
+ vc->worker.tight.quality = (enc & 0x0F);
}
break;
default:
@@ -2958,7 +2948,7 @@ static VncRectStat *vnc_stat_rect(VncDisplay *vd, int x, int y)
return &vs->stats[y / VNC_STAT_RECT][x / VNC_STAT_RECT];
}
-void vnc_sent_lossy_rect(VncState *vs, int x, int y, int w, int h)
+void vnc_sent_lossy_rect(VncWorker *worker, int x, int y, int w, int h)
{
int i, j;
@@ -2969,7 +2959,7 @@ void vnc_sent_lossy_rect(VncState *vs, int x, int y, int w, int h)
for (j = y; j <= h; j++) {
for (i = x; i <= w; i++) {
- vs->lossy_rect[j][i] = 1;
+ worker->lossy_rect[j][i] = 1;
}
}
}
@@ -2985,6 +2975,7 @@ static int vnc_refresh_lossy_rect(VncDisplay *vd, int x, int y)
x = QEMU_ALIGN_DOWN(x, VNC_STAT_RECT);
QTAILQ_FOREACH(vs, &vd->clients, next) {
+ VncConnection *vc = container_of(vs, VncConnection, vs);
int j;
/* kernel send buffers are full -> refresh later */
@@ -2992,11 +2983,11 @@ static int vnc_refresh_lossy_rect(VncDisplay *vd, int x, int y)
continue;
}
- if (!vs->lossy_rect[sty][stx]) {
+ if (!vc->worker.lossy_rect[sty][stx]) {
continue;
}
- vs->lossy_rect[sty][stx] = 0;
+ vc->worker.lossy_rect[sty][stx] = 0;
for (j = 0; j < VNC_STAT_RECT; ++j) {
bitmap_set(vs->dirty[y + j],
x / VNC_DIRTY_PIXELS_PER_BIT,
@@ -3249,12 +3240,8 @@ static void vnc_connect(VncDisplay *vd, QIOChannelSocket *sioc,
VncConnection *vc = g_new0(VncConnection, 1);
VncState *vs = &vc->vs;
bool first_client = QTAILQ_EMPTY(&vd->clients);
- int i;
trace_vnc_client_connect(vs, sioc);
- vs->zlib = &vc->zlib;
- vs->zrle = g_new0(VncZrle, 1);
- vs->tight = g_new0(VncTight, 1);
vs->magic = VNC_MAGIC;
vs->sioc = sioc;
object_ref(OBJECT(vs->sioc));
@@ -3262,23 +3249,23 @@ static void vnc_connect(VncDisplay *vd, QIOChannelSocket *sioc,
object_ref(OBJECT(vs->ioc));
vs->vd = vd;
- buffer_init(&vs->input, "vnc-input/%p", sioc);
- buffer_init(&vs->output, "vnc-output/%p", sioc);
- buffer_init(&vs->jobs_buffer, "vnc-jobs_buffer/%p", sioc);
+ buffer_init(&vs->input, "vnc-input/%p", sioc);
+ buffer_init(&vs->output, "vnc-output/%p", sioc);
+ buffer_init(&vs->jobs_buffer, "vnc-jobs_buffer/%p", sioc);
- buffer_init(&vs->tight->tight, "vnc-tight/%p", sioc);
- buffer_init(&vs->tight->zlib, "vnc-tight-zlib/%p", sioc);
- buffer_init(&vs->tight->gradient, "vnc-tight-gradient/%p", sioc);
+ buffer_init(&vc->worker.tight.tight, "vnc-tight/%p", sioc);
+ buffer_init(&vc->worker.tight.zlib, "vnc-tight-zlib/%p", sioc);
+ buffer_init(&vc->worker.tight.gradient, "vnc-tight-gradient/%p", sioc);
#ifdef CONFIG_VNC_JPEG
- buffer_init(&vs->tight->jpeg, "vnc-tight-jpeg/%p", sioc);
+ buffer_init(&vc->worker.tight.jpeg, "vnc-tight-jpeg/%p", sioc);
#endif
#ifdef CONFIG_PNG
- buffer_init(&vs->tight->png, "vnc-tight-png/%p", sioc);
+ buffer_init(&vc->worker.tight.png, "vnc-tight-png/%p", sioc);
#endif
- buffer_init(&vc->zlib.zlib, "vnc-zlib/%p", sioc);
- buffer_init(&vs->zrle->zrle, "vnc-zrle/%p", sioc);
- buffer_init(&vs->zrle->fb, "vnc-zrle-fb/%p", sioc);
- buffer_init(&vs->zrle->zlib, "vnc-zrle-zlib/%p", sioc);
+ buffer_init(&vc->worker.zlib.zlib, "vnc-zlib/%p", sioc);
+ buffer_init(&vc->worker.zrle.zrle, "vnc-zrle/%p", sioc);
+ buffer_init(&vc->worker.zrle.fb, "vnc-zrle-fb/%p", sioc);
+ buffer_init(&vc->worker.zrle.zlib, "vnc-zrle-zlib/%p", sioc);
if (skipauth) {
vs->auth = VNC_AUTH_NONE;
@@ -3295,11 +3282,6 @@ static void vnc_connect(VncDisplay *vd, QIOChannelSocket *sioc,
VNC_DEBUG("Client sioc=%p ws=%d auth=%d subauth=%d\n",
sioc, websocket, vs->auth, vs->subauth);
- vs->lossy_rect = g_malloc0(VNC_STAT_ROWS * sizeof (*vs->lossy_rect));
- for (i = 0; i < VNC_STAT_ROWS; ++i) {
- vs->lossy_rect[i] = g_new0(uint8_t, VNC_STAT_COLS);
- }
-
VNC_DEBUG("New client on socket %p\n", vs->sioc);
update_displaychangelistener(&vd->dcl, VNC_REFRESH_INTERVAL_BASE);
qio_channel_set_blocking(vs->ioc, false, NULL);
diff --git a/ui/vnc-enc-zrle.c.inc b/ui/vnc-enc-zrle.c.inc
index 2ef7501d52..68d28f58b7 100644
--- a/ui/vnc-enc-zrle.c.inc
+++ b/ui/vnc-enc-zrle.c.inc
@@ -62,16 +62,16 @@
#define ZRLE_ENCODE_TILE ZRLE_CONCAT2(zrle_encode_tile, ZRLE_ENCODE_SUFFIX)
#define ZRLE_WRITE_PALETTE ZRLE_CONCAT2(zrle_write_palette,ZRLE_ENCODE_SUFFIX)
-static void ZRLE_ENCODE_TILE(VncState *vs, ZRLE_PIXEL *data, int w, int h,
- int zywrle_level);
+static void ZRLE_ENCODE_TILE(VncState *vs, VncZrle *zrle, ZRLE_PIXEL *data,
+ int w, int h, int zywrle_level);
#if ZRLE_BPP != 8
#include "vnc-enc-zywrle-template.c"
#endif
-static void ZRLE_ENCODE(VncState *vs, int x, int y, int w, int h,
- int zywrle_level)
+static void ZRLE_ENCODE(VncState *vs, VncZrle *zrle,
+ int x, int y, int w, int h, int zywrle_level)
{
int ty;
@@ -87,16 +87,16 @@ static void ZRLE_ENCODE(VncState *vs, int x, int y, int w, int h,
tw = MIN(VNC_ZRLE_TILE_WIDTH, x + w - tx);
- buf = zrle_convert_fb(vs, tx, ty, tw, th, ZRLE_BPP);
- ZRLE_ENCODE_TILE(vs, buf, tw, th, zywrle_level);
+ buf = zrle_convert_fb(vs, zrle, tx, ty, tw, th, ZRLE_BPP);
+ ZRLE_ENCODE_TILE(vs, zrle, buf, tw, th, zywrle_level);
}
}
}
-static void ZRLE_ENCODE_TILE(VncState *vs, ZRLE_PIXEL *data, int w, int h,
- int zywrle_level)
+static void ZRLE_ENCODE_TILE(VncState *vs, VncZrle *zrle, ZRLE_PIXEL *data,
+ int w, int h, int zywrle_level)
{
- VncPalette *palette = &vs->zrle->palette;
+ VncPalette *palette = &zrle->palette;
int runs = 0;
int single_pixels = 0;
@@ -236,7 +236,7 @@ static void ZRLE_ENCODE_TILE(VncState *vs, ZRLE_PIXEL *data, int w, int h,
#if ZRLE_BPP != 8
if (zywrle_level > 0 && !(zywrle_level & 0x80)) {
ZYWRLE_ANALYZE(data, data, w, h, w, zywrle_level, vs->zywrle.buf);
- ZRLE_ENCODE_TILE(vs, data, w, h, zywrle_level | 0x80);
+ ZRLE_ENCODE_TILE(vs, zrle, data, w, h, zywrle_level | 0x80);
}
else
#endif
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 03/13] ui/egl-helpers: Error check the fds in egl_dmabuf_export_texture()
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
2025-07-15 6:32 ` [PULL v2 01/13] ui/vnc: Do not copy z_stream marcandre.lureau
2025-07-15 6:32 ` [PULL v2 02/13] ui/vnc: Introduce the VncWorker type marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 04/13] ui/spice: Enable gl=on option for non-local or remote clients marcandre.lureau
` (11 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Vivek Kasireddy, Gerd Hoffmann, Marc-André Lureau,
Dmitry Osipenko, Frediano Ziglio, Dongwon Kim, Michael Scherle
From: Vivek Kasireddy <vivek.kasireddy@intel.com>
While trying to export and obtain fds associated with a texture, it
is possible that the fds returned after eglExportDMABUFImageMESA()
call have error values. Therefore, we need to evaluate the value of
all fds and return false if any of them are negative.
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Frediano Ziglio <freddy77@gmail.com>
Cc: Dongwon Kim <dongwon.kim@intel.com>
Cc: Michael Scherle <michael.scherle@rz.uni-freiburg.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Message-Id: <20250617043546.1022779-2-vivek.kasireddy@intel.com>
---
ui/egl-helpers.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/ui/egl-helpers.c b/ui/egl-helpers.c
index 5503a795e4..e3f2872cc1 100644
--- a/ui/egl-helpers.c
+++ b/ui/egl-helpers.c
@@ -295,6 +295,7 @@ bool egl_dmabuf_export_texture(uint32_t tex_id, int *fd, EGLint *offset,
{
EGLImageKHR image;
EGLuint64KHR modifiers[DMABUF_MAX_PLANES];
+ int i;
image = eglCreateImageKHR(qemu_egl_display, eglGetCurrentContext(),
EGL_GL_TEXTURE_2D_KHR,
@@ -314,6 +315,11 @@ bool egl_dmabuf_export_texture(uint32_t tex_id, int *fd, EGLint *offset,
*modifier = modifiers[0];
}
+ for (i = 0; i < *num_planes; i++) {
+ if (fd[i] < 0) {
+ return false;
+ }
+ }
return true;
}
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 04/13] ui/spice: Enable gl=on option for non-local or remote clients
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (2 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 03/13] ui/egl-helpers: Error check the fds in egl_dmabuf_export_texture() marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 05/13] ui/spice: Add an option for users to provide a preferred video codec marcandre.lureau
` (10 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Vivek Kasireddy, Gerd Hoffmann, Marc-André Lureau,
Dmitry Osipenko, Frediano Ziglio, Dongwon Kim, Michael Scherle
From: Vivek Kasireddy <vivek.kasireddy@intel.com>
Newer versions of Spice server should be able to accept dmabuf
fds from Qemu for clients that are connected via the network.
In other words, when this option is enabled, Qemu would share
a dmabuf fd with Spice which would encode and send the data
associated with the fd to a client that could be located on
a different machine.
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Frediano Ziglio <freddy77@gmail.com>
Cc: Dongwon Kim <dongwon.kim@intel.com>
Cc: Michael Scherle <michael.scherle@rz.uni-freiburg.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Message-Id: <20250617043546.1022779-3-vivek.kasireddy@intel.com>
---
include/ui/spice-display.h | 1 +
ui/spice-core.c | 4 ++++
ui/spice-display.c | 1 +
3 files changed, 6 insertions(+)
diff --git a/include/ui/spice-display.h b/include/ui/spice-display.h
index e1a9b36185..6c55f38c8b 100644
--- a/include/ui/spice-display.h
+++ b/include/ui/spice-display.h
@@ -151,6 +151,7 @@ struct SimpleSpiceCursor {
};
extern bool spice_opengl;
+extern bool spice_remote_client;
int qemu_spice_rect_is_empty(const QXLRect* r);
void qemu_spice_rect_union(QXLRect *dest, const QXLRect *r);
diff --git a/ui/spice-core.c b/ui/spice-core.c
index 0326c63bec..5acbdd3955 100644
--- a/ui/spice-core.c
+++ b/ui/spice-core.c
@@ -836,9 +836,13 @@ static void qemu_spice_init(void)
#ifdef HAVE_SPICE_GL
if (qemu_opt_get_bool(opts, "gl", 0)) {
if ((port != 0) || (tls_port != 0)) {
+#if SPICE_SERVER_VERSION >= 0x000f03 /* release 0.15.3 */
+ spice_remote_client = 1;
+#else
error_report("SPICE GL support is local-only for now and "
"incompatible with -spice port/tls-port");
exit(1);
+#endif
}
egl_init(qemu_opt_get(opts, "rendernode"), DISPLAY_GL_MODE_ON, &error_fatal);
spice_opengl = 1;
diff --git a/ui/spice-display.c b/ui/spice-display.c
index 9c39d2c5c8..0fb72f6d6f 100644
--- a/ui/spice-display.c
+++ b/ui/spice-display.c
@@ -31,6 +31,7 @@
#include "standard-headers/drm/drm_fourcc.h"
bool spice_opengl;
+bool spice_remote_client;
int qemu_spice_rect_is_empty(const QXLRect* r)
{
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 05/13] ui/spice: Add an option for users to provide a preferred video codec
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (3 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 04/13] ui/spice: Enable gl=on option for non-local or remote clients marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 06/13] ui/spice: Add an option to submit gl_draw requests at fixed rate marcandre.lureau
` (9 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Vivek Kasireddy, Gerd Hoffmann, Marc-André Lureau,
Dmitry Osipenko, Frediano Ziglio, Dongwon Kim, Michael Scherle,
Daniel P. Berrangé
From: Vivek Kasireddy <vivek.kasireddy@intel.com>
Giving users an option to choose a particular codec will enable
them to make an appropriate decision based on their hardware and
use-case. Note that, the Spice server would use this codec with
Gstreamer encoder and only when gl=on is specified.
If no codec is provided, then the codec gstreamer:h264 would be used
as default. And, for the case where gl=off, the default codec to be
used is determined by the Spice server.
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Frediano Ziglio <freddy77@gmail.com>
Cc: Dongwon Kim <dongwon.kim@intel.com>
Cc: Michael Scherle <michael.scherle@rz.uni-freiburg.de>
Cc: Daniel P. Berrangé <berrange@redhat.com>
[ Marc-Andre - fix unused variables warnings ]
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Message-Id: <20250617043546.1022779-4-vivek.kasireddy@intel.com>
---
ui/spice-core.c | 16 ++++++++++++++++
qemu-options.hx | 8 ++++++++
2 files changed, 24 insertions(+)
diff --git a/ui/spice-core.c b/ui/spice-core.c
index 5acbdd3955..51bdcef1af 100644
--- a/ui/spice-core.c
+++ b/ui/spice-core.c
@@ -488,6 +488,9 @@ static QemuOptsList qemu_spice_opts = {
},{
.name = "streaming-video",
.type = QEMU_OPT_STRING,
+ },{
+ .name = "video-codec",
+ .type = QEMU_OPT_STRING,
},{
.name = "agent-mouse",
.type = QEMU_OPT_BOOL,
@@ -837,7 +840,20 @@ static void qemu_spice_init(void)
if (qemu_opt_get_bool(opts, "gl", 0)) {
if ((port != 0) || (tls_port != 0)) {
#if SPICE_SERVER_VERSION >= 0x000f03 /* release 0.15.3 */
+ const char *video_codec = NULL;
+ g_autofree char *enc_codec = NULL;
+
spice_remote_client = 1;
+
+ video_codec = qemu_opt_get(opts, "video-codec");
+ if (video_codec) {
+ enc_codec = g_strconcat("gstreamer:", video_codec, NULL);
+ }
+ if (spice_server_set_video_codecs(spice_server,
+ enc_codec ?: "gstreamer:h264")) {
+ error_report("invalid video codec");
+ exit(1);
+ }
#else
error_report("SPICE GL support is local-only for now and "
"incompatible with -spice port/tls-port");
diff --git a/qemu-options.hx b/qemu-options.hx
index 1f862b19a6..8f6a228a89 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -2281,6 +2281,7 @@ DEF("spice", HAS_ARG, QEMU_OPTION_spice,
" [,streaming-video=[off|all|filter]][,disable-copy-paste=on|off]\n"
" [,disable-agent-file-xfer=on|off][,agent-mouse=[on|off]]\n"
" [,playback-compression=[on|off]][,seamless-migration=[on|off]]\n"
+ " [,video-codec=<codec>\n"
" [,gl=[on|off]][,rendernode=<file>]\n"
" enable spice\n"
" at least one of {port, tls-port} is mandatory\n",
@@ -2369,6 +2370,13 @@ SRST
``seamless-migration=[on|off]``
Enable/disable spice seamless migration. Default is off.
+ ``video-codec=<codec>``
+ Provide the preferred codec the Spice server should use with the
+ Gstreamer encoder. This option is only relevant when gl=on is
+ specified. If no codec is provided, then the codec gstreamer:h264
+ would be used as default. And, for the case where gl=off, the
+ default codec to be used is determined by the Spice server.
+
``gl=[on|off]``
Enable/disable OpenGL context. Default is off.
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 06/13] ui/spice: Add an option to submit gl_draw requests at fixed rate
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (4 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 05/13] ui/spice: Add an option for users to provide a preferred video codec marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 07/13] ui/console-gl: Add a helper to create a texture with linear memory layout marcandre.lureau
` (8 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Vivek Kasireddy, Gerd Hoffmann, Marc-André Lureau,
Dmitry Osipenko, Frediano Ziglio, Dongwon Kim, Michael Scherle
From: Vivek Kasireddy <vivek.kasireddy@intel.com>
In the specific case where the display layer (virtio-gpu) is using
dmabuf, and if remote clients are enabled (-spice gl=on,port=xxxx),
it makes sense to limit the maximum (streaming) rate (refresh rate)
to a fixed value using the GUI refresh timer. Otherwise, the updates
or gl_draw requests would be sent as soon as the Guest submits a new
frame which is not optimal as it would lead to increased network
traffic and wastage of GPU cycles if the frames get dropped.
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Frediano Ziglio <freddy77@gmail.com>
Cc: Dongwon Kim <dongwon.kim@intel.com>
Cc: Michael Scherle <michael.scherle@rz.uni-freiburg.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Message-Id: <20250617043546.1022779-5-vivek.kasireddy@intel.com>
---
include/ui/spice-display.h | 1 +
ui/spice-core.c | 12 ++++++++
ui/spice-display.c | 62 ++++++++++++++++++++++++++++++++------
qemu-options.hx | 5 +++
4 files changed, 70 insertions(+), 10 deletions(-)
diff --git a/include/ui/spice-display.h b/include/ui/spice-display.h
index 6c55f38c8b..9bdde78266 100644
--- a/include/ui/spice-display.h
+++ b/include/ui/spice-display.h
@@ -152,6 +152,7 @@ struct SimpleSpiceCursor {
extern bool spice_opengl;
extern bool spice_remote_client;
+extern int spice_max_refresh_rate;
int qemu_spice_rect_is_empty(const QXLRect* r);
void qemu_spice_rect_union(QXLRect *dest, const QXLRect *r);
diff --git a/ui/spice-core.c b/ui/spice-core.c
index 51bdcef1af..5992f9daec 100644
--- a/ui/spice-core.c
+++ b/ui/spice-core.c
@@ -56,6 +56,8 @@ struct SpiceTimer {
QEMUTimer *timer;
};
+#define DEFAULT_MAX_REFRESH_RATE 30
+
static SpiceTimer *timer_add(SpiceTimerFunc func, void *opaque)
{
SpiceTimer *timer;
@@ -491,6 +493,9 @@ static QemuOptsList qemu_spice_opts = {
},{
.name = "video-codec",
.type = QEMU_OPT_STRING,
+ },{
+ .name = "max-refresh-rate",
+ .type = QEMU_OPT_NUMBER,
},{
.name = "agent-mouse",
.type = QEMU_OPT_BOOL,
@@ -804,6 +809,13 @@ static void qemu_spice_init(void)
spice_server_set_streaming_video(spice_server, SPICE_STREAM_VIDEO_OFF);
}
+ spice_max_refresh_rate = qemu_opt_get_number(opts, "max-refresh-rate",
+ DEFAULT_MAX_REFRESH_RATE);
+ if (spice_max_refresh_rate <= 0) {
+ error_report("max refresh rate/fps is invalid");
+ exit(1);
+ }
+
spice_server_set_agent_mouse
(spice_server, qemu_opt_get_bool(opts, "agent-mouse", 1));
spice_server_set_playback_compression
diff --git a/ui/spice-display.c b/ui/spice-display.c
index 0fb72f6d6f..e409b6bdb2 100644
--- a/ui/spice-display.c
+++ b/ui/spice-display.c
@@ -32,6 +32,7 @@
bool spice_opengl;
bool spice_remote_client;
+int spice_max_refresh_rate;
int qemu_spice_rect_is_empty(const QXLRect* r)
{
@@ -844,12 +845,32 @@ static void qemu_spice_gl_block_timer(void *opaque)
warn_report("spice: no gl-draw-done within one second");
}
+static void spice_gl_draw(SimpleSpiceDisplay *ssd,
+ uint32_t x, uint32_t y, uint32_t w, uint32_t h)
+{
+ uint64_t cookie;
+
+ cookie = (uintptr_t)qxl_cookie_new(QXL_COOKIE_TYPE_GL_DRAW_DONE, 0);
+ spice_qxl_gl_draw_async(&ssd->qxl, x, y, w, h, cookie);
+}
+
static void spice_gl_refresh(DisplayChangeListener *dcl)
{
SimpleSpiceDisplay *ssd = container_of(dcl, SimpleSpiceDisplay, dcl);
- uint64_t cookie;
- if (!ssd->ds || qemu_console_is_gl_blocked(ssd->dcl.con)) {
+ if (!ssd->ds) {
+ return;
+ }
+
+ if (qemu_console_is_gl_blocked(ssd->dcl.con)) {
+ if (spice_remote_client && ssd->gl_updates && ssd->have_scanout) {
+ glFlush();
+ spice_gl_draw(ssd, 0, 0,
+ surface_width(ssd->ds), surface_height(ssd->ds));
+ ssd->gl_updates = 0;
+ /* E.g, to achieve 60 FPS, update_interval needs to be ~16.66 ms */
+ dcl->update_interval = 1000 / spice_max_refresh_rate;
+ }
return;
}
@@ -857,11 +878,8 @@ static void spice_gl_refresh(DisplayChangeListener *dcl)
if (ssd->gl_updates && ssd->have_surface) {
qemu_spice_gl_block(ssd, true);
glFlush();
- cookie = (uintptr_t)qxl_cookie_new(QXL_COOKIE_TYPE_GL_DRAW_DONE, 0);
- spice_qxl_gl_draw_async(&ssd->qxl, 0, 0,
- surface_width(ssd->ds),
- surface_height(ssd->ds),
- cookie);
+ spice_gl_draw(ssd, 0, 0,
+ surface_width(ssd->ds), surface_height(ssd->ds));
ssd->gl_updates = 0;
}
}
@@ -954,6 +972,20 @@ static void qemu_spice_gl_scanout_disable(DisplayChangeListener *dcl)
SimpleSpiceDisplay *ssd = container_of(dcl, SimpleSpiceDisplay, dcl);
trace_qemu_spice_gl_scanout_disable(ssd->qxl.id);
+
+ /*
+ * We need to check for the case of "lost" updates, where a gl_draw
+ * was not submitted because the timer did not get a chance to run.
+ * One case where this happens is when the Guest VM is getting
+ * rebooted. If the console is blocked in this situation, we need
+ * to unblock it. Otherwise, newer updates would not take effect.
+ */
+ if (qemu_console_is_gl_blocked(ssd->dcl.con)) {
+ if (spice_remote_client && ssd->gl_updates && ssd->have_scanout) {
+ ssd->gl_updates = 0;
+ qemu_spice_gl_block(ssd, false);
+ }
+ }
spice_server_gl_scanout(&ssd->qxl, NULL, 0, 0, NULL, NULL, 0, DRM_FORMAT_INVALID,
DRM_FORMAT_MOD_INVALID, false);
qemu_spice_gl_monitor_config(ssd, 0, 0, 0, 0);
@@ -1061,7 +1093,6 @@ static void qemu_spice_gl_update(DisplayChangeListener *dcl,
EGLint fourcc = 0;
bool render_cursor = false;
bool y_0_top = false; /* FIXME */
- uint64_t cookie;
uint32_t width, height, texture;
if (!ssd->have_scanout) {
@@ -1159,8 +1190,19 @@ static void qemu_spice_gl_update(DisplayChangeListener *dcl,
trace_qemu_spice_gl_update(ssd->qxl.id, w, h, x, y);
qemu_spice_gl_block(ssd, true);
glFlush();
- cookie = (uintptr_t)qxl_cookie_new(QXL_COOKIE_TYPE_GL_DRAW_DONE, 0);
- spice_qxl_gl_draw_async(&ssd->qxl, x, y, w, h, cookie);
+
+ /*
+ * In the case of remote clients, the submission of gl_draw request is
+ * deferred here, so that it can be submitted later (to spice server)
+ * from spice_gl_refresh() timer callback. This is done to ensure that
+ * Guest updates are submitted at a steady rate (e.g. 60 FPS) instead
+ * of submitting them arbitrarily.
+ */
+ if (spice_remote_client) {
+ ssd->gl_updates++;
+ } else {
+ spice_gl_draw(ssd, x, y, w, h);
+ }
}
static const DisplayChangeListenerOps display_listener_gl_ops = {
diff --git a/qemu-options.hx b/qemu-options.hx
index 8f6a228a89..ac09abfd71 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -2282,6 +2282,7 @@ DEF("spice", HAS_ARG, QEMU_OPTION_spice,
" [,disable-agent-file-xfer=on|off][,agent-mouse=[on|off]]\n"
" [,playback-compression=[on|off]][,seamless-migration=[on|off]]\n"
" [,video-codec=<codec>\n"
+ " [,max-refresh-rate=rate\n"
" [,gl=[on|off]][,rendernode=<file>]\n"
" enable spice\n"
" at least one of {port, tls-port} is mandatory\n",
@@ -2377,6 +2378,10 @@ SRST
would be used as default. And, for the case where gl=off, the
default codec to be used is determined by the Spice server.
+ ``max-refresh-rate=rate``
+ Provide the maximum refresh rate (or FPS) at which the encoding
+ requests should be sent to the Spice server. Default would be 30.
+
``gl=[on|off]``
Enable/disable OpenGL context. Default is off.
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 07/13] ui/console-gl: Add a helper to create a texture with linear memory layout
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (5 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 06/13] ui/spice: Add an option to submit gl_draw requests at fixed rate marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 08/13] ui/spice: Create a new texture with linear layout when gl=on is specified marcandre.lureau
` (7 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Vivek Kasireddy, Gerd Hoffmann, Marc-André Lureau,
Dmitry Osipenko, Frediano Ziglio, Dongwon Kim
From: Vivek Kasireddy <vivek.kasireddy@intel.com>
There are cases where we do not want the memory layout of a texture to
be tiled as the component processing the texture would not know how to
de-tile either via software or hardware. Therefore, ensuring that the
memory backing the texture has a linear layout is absolutely necessary
in these situations.
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Frediano Ziglio <freddy77@gmail.com>
Cc: Dongwon Kim <dongwon.kim@intel.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Co-developed-by: Michael Scherle <michael.scherle@rz.uni-freiburg.de>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Message-Id: <20250617043546.1022779-6-vivek.kasireddy@intel.com>
---
include/ui/console.h | 3 +++
ui/console-gl.c | 48 ++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 51 insertions(+)
diff --git a/include/ui/console.h b/include/ui/console.h
index 46b3128185..98feaa58bd 100644
--- a/include/ui/console.h
+++ b/include/ui/console.h
@@ -422,6 +422,9 @@ bool console_gl_check_format(DisplayChangeListener *dcl,
pixman_format_code_t format);
void surface_gl_create_texture(QemuGLShader *gls,
DisplaySurface *surface);
+bool surface_gl_create_texture_from_fd(DisplaySurface *surface,
+ int fd, GLuint *texture,
+ GLuint *mem_obj);
void surface_gl_update_texture(QemuGLShader *gls,
DisplaySurface *surface,
int x, int y, int w, int h);
diff --git a/ui/console-gl.c b/ui/console-gl.c
index 103b954017..afb36dba64 100644
--- a/ui/console-gl.c
+++ b/ui/console-gl.c
@@ -25,6 +25,7 @@
* THE SOFTWARE.
*/
#include "qemu/osdep.h"
+#include "qemu/error-report.h"
#include "ui/console.h"
#include "ui/shader.h"
@@ -96,6 +97,53 @@ void surface_gl_create_texture(QemuGLShader *gls,
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
}
+bool surface_gl_create_texture_from_fd(DisplaySurface *surface,
+ int fd, GLuint *texture,
+ GLuint *mem_obj)
+{
+ unsigned long size = surface_stride(surface) * surface_height(surface);
+ GLenum err = glGetError();
+ *texture = 0;
+ *mem_obj = 0;
+
+ if (!epoxy_has_gl_extension("GL_EXT_memory_object") ||
+ !epoxy_has_gl_extension("GL_EXT_memory_object_fd")) {
+ error_report("spice: required OpenGL extensions not supported: "
+ "GL_EXT_memory_object and GL_EXT_memory_object_fd");
+ return false;
+ }
+
+#ifdef GL_EXT_memory_object_fd
+ glCreateMemoryObjectsEXT(1, mem_obj);
+ glImportMemoryFdEXT(*mem_obj, size, GL_HANDLE_TYPE_OPAQUE_FD_EXT, fd);
+
+ err = glGetError();
+ if (err != GL_NO_ERROR) {
+ error_report("spice: cannot import memory object from fd");
+ goto cleanup_mem;
+ }
+
+ glGenTextures(1, texture);
+ glBindTexture(GL_TEXTURE_2D, *texture);
+ glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_TILING_EXT, GL_LINEAR_TILING_EXT);
+ glTexStorageMem2DEXT(GL_TEXTURE_2D, 1, GL_RGBA8, surface_width(surface),
+ surface_height(surface), *mem_obj, 0);
+ err = glGetError();
+ if (err != GL_NO_ERROR) {
+ error_report("spice: cannot create texture from memory object");
+ goto cleanup_tex_and_mem;
+ }
+ return true;
+
+cleanup_tex_and_mem:
+ glDeleteTextures(1, texture);
+cleanup_mem:
+ glDeleteMemoryObjectsEXT(1, mem_obj);
+
+#endif
+ return false;
+}
+
void surface_gl_update_texture(QemuGLShader *gls,
DisplaySurface *surface,
int x, int y, int w, int h)
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 08/13] ui/spice: Create a new texture with linear layout when gl=on is specified
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (6 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 07/13] ui/console-gl: Add a helper to create a texture with linear memory layout marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 09/13] ui/spice: Blit the scanout texture if its memory layout is not linear marcandre.lureau
` (6 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Vivek Kasireddy, Gerd Hoffmann, Marc-André Lureau,
Dmitry Osipenko, Frediano Ziglio, Dongwon Kim
From: Vivek Kasireddy <vivek.kasireddy@intel.com>
Since most encoders/decoders (invoked by Spice) may not work properly
with tiled memory associated with a texture, we need to create another
texture that has linear memory layout and use that instead.
Note that, there does not seem to be a direct way to indicate to the
GL implementation that a texture's backing memory needs to be linear.
Instead, we have to do it in a roundabout way where we need to first
create a tiled texture and import that as a memory object to create
a new texture that has a linear memory layout.
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Frediano Ziglio <freddy77@gmail.com>
Cc: Dongwon Kim <dongwon.kim@intel.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Co-developed-by: Michael Scherle <michael.scherle@rz.uni-freiburg.de>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Message-Id: <20250617043546.1022779-7-vivek.kasireddy@intel.com>
---
include/ui/surface.h | 1 +
ui/console-gl.c | 6 ++++
ui/spice-display.c | 82 ++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 89 insertions(+)
diff --git a/include/ui/surface.h b/include/ui/surface.h
index f16f7be8be..006b1986bb 100644
--- a/include/ui/surface.h
+++ b/include/ui/surface.h
@@ -22,6 +22,7 @@ typedef struct DisplaySurface {
GLenum glformat;
GLenum gltype;
GLuint texture;
+ GLuint mem_obj;
#endif
qemu_pixman_shareable share_handle;
uint32_t share_handle_offset;
diff --git a/ui/console-gl.c b/ui/console-gl.c
index afb36dba64..403fc36fbd 100644
--- a/ui/console-gl.c
+++ b/ui/console-gl.c
@@ -184,6 +184,12 @@ void surface_gl_destroy_texture(QemuGLShader *gls,
}
glDeleteTextures(1, &surface->texture);
surface->texture = 0;
+#ifdef GL_EXT_memory_object_fd
+ if (surface->mem_obj) {
+ glDeleteMemoryObjectsEXT(1, &surface->mem_obj);
+ surface->mem_obj = 0;
+ }
+#endif
}
void surface_gl_setup_viewport(QemuGLShader *gls,
diff --git a/ui/spice-display.c b/ui/spice-display.c
index e409b6bdb2..854a97c198 100644
--- a/ui/spice-display.c
+++ b/ui/spice-display.c
@@ -893,6 +893,81 @@ static void spice_gl_update(DisplayChangeListener *dcl,
ssd->gl_updates++;
}
+static bool spice_gl_replace_fd_texture(SimpleSpiceDisplay *ssd,
+ int *fds, uint64_t *modifier,
+ int *num_planes)
+{
+ uint32_t offsets[DMABUF_MAX_PLANES], strides[DMABUF_MAX_PLANES];
+ GLuint texture;
+ GLuint mem_obj;
+ int fourcc;
+ bool ret;
+
+ if (!spice_remote_client) {
+ return true;
+ }
+
+ if (*modifier == DRM_FORMAT_MOD_LINEAR) {
+ return true;
+ }
+
+ if (*num_planes > 1) {
+ error_report("spice: cannot replace texture with multiple planes");
+ return false;
+ }
+
+ /*
+ * We really want to ensure that the memory layout of the texture
+ * is linear; otherwise, the encoder's output may show corruption.
+ */
+ if (!surface_gl_create_texture_from_fd(ssd->ds, fds[0], &texture,
+ &mem_obj)) {
+ error_report("spice: cannot create new texture from fd");
+ return false;
+ }
+
+ /*
+ * A successful return after glImportMemoryFdEXT() means that
+ * the ownership of fd has been passed to GL. In other words,
+ * the fd we got above should not be used anymore.
+ */
+ ret = egl_dmabuf_export_texture(texture,
+ fds,
+ (EGLint *)offsets,
+ (EGLint *)strides,
+ &fourcc,
+ num_planes,
+ modifier);
+ if (!ret) {
+ glDeleteTextures(1, &texture);
+#ifdef GL_EXT_memory_object_fd
+ glDeleteMemoryObjectsEXT(1, &mem_obj);
+#endif
+
+ /*
+ * Since we couldn't export our newly create texture (or create,
+ * an fd associated with it) we need to backtrack and try to
+ * recreate the fd associated with the original texture.
+ */
+ ret = egl_dmabuf_export_texture(ssd->ds->texture,
+ fds,
+ (EGLint *)offsets,
+ (EGLint *)strides,
+ &fourcc,
+ num_planes,
+ modifier);
+ if (!ret) {
+ surface_gl_destroy_texture(ssd->gls, ssd->ds);
+ warn_report("spice: no texture available to display");
+ }
+ } else {
+ surface_gl_destroy_texture(ssd->gls, ssd->ds);
+ ssd->ds->texture = texture;
+ ssd->ds->mem_obj = mem_obj;
+ }
+ return ret;
+}
+
static void spice_server_gl_scanout(QXLInstance *qxl,
const int *fd,
uint32_t width, uint32_t height,
@@ -917,6 +992,7 @@ static void spice_gl_switch(DisplayChangeListener *dcl,
struct DisplaySurface *new_surface)
{
SimpleSpiceDisplay *ssd = container_of(dcl, SimpleSpiceDisplay, dcl);
+ bool ret;
if (ssd->ds) {
surface_gl_destroy_texture(ssd->gls, ssd->ds);
@@ -939,6 +1015,12 @@ static void spice_gl_switch(DisplayChangeListener *dcl,
return;
}
+ ret = spice_gl_replace_fd_texture(ssd, fd, &modifier, &num_planes);
+ if (!ret) {
+ surface_gl_destroy_texture(ssd->gls, ssd->ds);
+ return;
+ }
+
trace_qemu_spice_gl_surface(ssd->qxl.id,
surface_width(ssd->ds),
surface_height(ssd->ds),
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 09/13] ui/spice: Blit the scanout texture if its memory layout is not linear
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (7 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 08/13] ui/spice: Create a new texture with linear layout when gl=on is specified marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-17 9:23 ` Peter Maydell
2025-07-15 6:32 ` [PULL v2 10/13] hw/display: Allow injection of virtio-gpu EDID name marcandre.lureau
` (5 subsequent siblings)
14 siblings, 1 reply; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Vivek Kasireddy, Gerd Hoffmann, Marc-André Lureau,
Dmitry Osipenko, Frediano Ziglio, Dongwon Kim, Michael Scherle
From: Vivek Kasireddy <vivek.kasireddy@intel.com>
In cases where the scanout buffer is provided as a texture (e.g. Virgl)
we need to check to see if it has a linear memory layout or not. If
it doesn't have a linear layout, then blitting it onto the texture
associated with the display surface (which already has a linear layout)
seems to ensure that there is no corruption seen regardless of which
encoder or decoder is used.
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Frediano Ziglio <freddy77@gmail.com>
Cc: Dongwon Kim <dongwon.kim@intel.com>
Cc: Michael Scherle <michael.scherle@rz.uni-freiburg.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
Message-Id: <20250617043546.1022779-8-vivek.kasireddy@intel.com>
---
include/ui/spice-display.h | 3 ++
ui/spice-display.c | 81 +++++++++++++++++++++++++++++++++++---
2 files changed, 78 insertions(+), 6 deletions(-)
diff --git a/include/ui/spice-display.h b/include/ui/spice-display.h
index 9bdde78266..690ece7380 100644
--- a/include/ui/spice-display.h
+++ b/include/ui/spice-display.h
@@ -132,6 +132,9 @@ struct SimpleSpiceDisplay {
egl_fb guest_fb;
egl_fb blit_fb;
egl_fb cursor_fb;
+ bool backing_y_0_top;
+ bool blit_scanout_texture;
+ bool new_scanout_texture;
bool have_hot;
#endif
};
diff --git a/ui/spice-display.c b/ui/spice-display.c
index 854a97c198..9ce622cefc 100644
--- a/ui/spice-display.c
+++ b/ui/spice-display.c
@@ -1086,7 +1086,7 @@ static void qemu_spice_gl_scanout_texture(DisplayChangeListener *dcl,
{
SimpleSpiceDisplay *ssd = container_of(dcl, SimpleSpiceDisplay, dcl);
EGLint offset[DMABUF_MAX_PLANES], stride[DMABUF_MAX_PLANES], fourcc = 0;
- int fd[DMABUF_MAX_PLANES], num_planes;
+ int fd[DMABUF_MAX_PLANES], num_planes, i;
uint64_t modifier;
assert(tex_id);
@@ -1098,11 +1098,26 @@ static void qemu_spice_gl_scanout_texture(DisplayChangeListener *dcl,
trace_qemu_spice_gl_scanout_texture(ssd->qxl.id, w, h, fourcc);
- /* note: spice server will close the fd */
- spice_server_gl_scanout(&ssd->qxl, fd, backing_width, backing_height,
- (uint32_t *)offset, (uint32_t *)stride, num_planes,
- fourcc, modifier, y_0_top);
- qemu_spice_gl_monitor_config(ssd, x, y, w, h);
+ if (spice_remote_client && modifier != DRM_FORMAT_MOD_LINEAR) {
+ egl_fb_destroy(&ssd->guest_fb);
+ egl_fb_setup_for_tex(&ssd->guest_fb,
+ backing_width, backing_height,
+ tex_id, false);
+ ssd->backing_y_0_top = y_0_top;
+ ssd->blit_scanout_texture = true;
+ ssd->new_scanout_texture = true;
+
+ for (i = 0; i < num_planes; i++) {
+ close(fd[i]);
+ }
+ } else {
+ /* note: spice server will close the fd */
+ spice_server_gl_scanout(&ssd->qxl, fd, backing_width, backing_height,
+ (uint32_t *)offset, (uint32_t *)stride,
+ num_planes, fourcc, modifier, y_0_top);
+ qemu_spice_gl_monitor_config(ssd, x, y, w, h);
+ }
+
ssd->have_surface = false;
ssd->have_scanout = true;
}
@@ -1168,6 +1183,50 @@ static void qemu_spice_gl_release_dmabuf(DisplayChangeListener *dcl,
egl_dmabuf_release_texture(dmabuf);
}
+static bool spice_gl_blit_scanout_texture(SimpleSpiceDisplay *ssd,
+ egl_fb *scanout_tex_fb)
+{
+ uint32_t offsets[DMABUF_MAX_PLANES], strides[DMABUF_MAX_PLANES];
+ int fds[DMABUF_MAX_PLANES], num_planes, fourcc;
+ uint64_t modifier;
+ bool ret;
+
+ egl_fb_destroy(scanout_tex_fb);
+ egl_fb_setup_for_tex(scanout_tex_fb,
+ surface_width(ssd->ds), surface_height(ssd->ds),
+ ssd->ds->texture, false);
+ egl_fb_blit(scanout_tex_fb, &ssd->guest_fb, false);
+ glFlush();
+
+ if (!ssd->new_scanout_texture) {
+ return true;
+ }
+
+ ret = egl_dmabuf_export_texture(ssd->ds->texture,
+ fds,
+ (EGLint *)offsets,
+ (EGLint *)strides,
+ &fourcc,
+ &num_planes,
+ &modifier);
+ if (!ret) {
+ error_report("spice: failed to get fd for texture");
+ return false;
+ }
+
+ spice_server_gl_scanout(&ssd->qxl, fds,
+ surface_width(ssd->ds),
+ surface_height(ssd->ds),
+ (uint32_t *)offsets, (uint32_t *)strides,
+ num_planes, fourcc, modifier,
+ ssd->backing_y_0_top);
+ qemu_spice_gl_monitor_config(ssd, 0, 0,
+ surface_width(ssd->ds),
+ surface_height(ssd->ds));
+ ssd->new_scanout_texture = false;
+ return true;
+}
+
static void qemu_spice_gl_update(DisplayChangeListener *dcl,
uint32_t x, uint32_t y, uint32_t w, uint32_t h)
{
@@ -1175,6 +1234,7 @@ static void qemu_spice_gl_update(DisplayChangeListener *dcl,
EGLint fourcc = 0;
bool render_cursor = false;
bool y_0_top = false; /* FIXME */
+ bool ret;
uint32_t width, height, texture;
if (!ssd->have_scanout) {
@@ -1269,6 +1329,15 @@ static void qemu_spice_gl_update(DisplayChangeListener *dcl,
glFlush();
}
+ if (spice_remote_client && ssd->blit_scanout_texture) {
+ egl_fb scanout_tex_fb;
+
+ ret = spice_gl_blit_scanout_texture(ssd, &scanout_tex_fb);
+ if (!ret) {
+ return;
+ }
+ }
+
trace_qemu_spice_gl_update(ssd->qxl.id, w, h, x, y);
qemu_spice_gl_block(ssd, true);
glFlush();
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 10/13] hw/display: Allow injection of virtio-gpu EDID name
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (8 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 09/13] ui/spice: Blit the scanout texture if its memory layout is not linear marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 11/13] ui/gtk: Add keep-aspect-ratio option marcandre.lureau
` (4 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Andrew Keesler, Paolo Bonzini, Daniel P. Berrangé,
Eduardo Habkost, Alex Bennée, Akihiko Odaki, Dmitry Osipenko,
Michael S. Tsirkin, Gerd Hoffmann, Eric Blake, Markus Armbruster
From: Andrew Keesler <ankeesler@google.com>
Thanks to 72d277a7, 1ed2cb32, and others, EDID (Extended Display
Identification Data) is propagated by QEMU such that a virtual display
presents legitimate metadata (e.g., name, serial number, preferred
resolutions, etc.) to its connected guest.
This change adds the ability to specify the EDID name for a particular
virtio-vga display. Previously, every virtual display would have the same
name: "QEMU Monitor". Now, we can inject names of displays in order to test
guest behavior that is specific to display names. We provide the ability to
inject the display name from the frontend since this is guest visible
data. Furthermore, this makes it clear where N potential display outputs
would get their name from (which will be added in a future change).
Note that we have elected to use a struct here for output data for
extensibility - we intend to add per-output fields like resolution in a
future change.
It should be noted that EDID names longer than 12 bytes will be truncated
per spec (I think?).
Testing: verified that when I specified 2 outputs for a virtio-gpu with
edid_name set, the names matched those that I configured with my vnc
display.
-display vnc=localhost:0,id=aaa,display=vga,head=0 \
-display vnc=localhost:1,id=bbb,display=vga,head=1 \
-device '{"driver":"virtio-vga",
"max_outputs":2,
"id":"vga",
"outputs":[
{
"name":"AAA"
},
{
"name":"BBB"
}
]}'
Signed-off-by: Andrew Keesler <ankeesler@google.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20250709121126.2946088-2-ankeesler@google.com>
---
qapi/virtio.json | 18 ++++++++++--
include/hw/display/edid.h | 2 ++
include/hw/qdev-properties-system.h | 5 ++++
include/hw/virtio/virtio-gpu.h | 3 ++
hw/core/qdev-properties-system.c | 44 +++++++++++++++++++++++++++++
hw/display/virtio-gpu-base.c | 27 ++++++++++++++++++
6 files changed, 97 insertions(+), 2 deletions(-)
diff --git a/qapi/virtio.json b/qapi/virtio.json
index 73df718a26..5e658a7033 100644
--- a/qapi/virtio.json
+++ b/qapi/virtio.json
@@ -963,17 +963,31 @@
{ 'struct': 'IOThreadVirtQueueMapping',
'data': { 'iothread': 'str', '*vqs': ['uint16'] } }
+##
+# @VirtIOGPUOutput:
+#
+# Describes configuration of a VirtIO GPU output.
+#
+# @name: the name of the output
+#
+# Since: 10.1
+##
+
+{ 'struct': 'VirtIOGPUOutput',
+ 'data': { 'name': 'str' } }
+
##
# @DummyVirtioForceArrays:
#
# Not used by QMP; hack to let us use IOThreadVirtQueueMappingList
-# internally
+# and VirtIOGPUOutputList internally
#
# Since: 9.0
##
{ 'struct': 'DummyVirtioForceArrays',
- 'data': { 'unused-iothread-vq-mapping': ['IOThreadVirtQueueMapping'] } }
+ 'data': { 'unused-iothread-vq-mapping': ['IOThreadVirtQueueMapping'],
+ 'unused-virtio-gpu-output': ['VirtIOGPUOutput'] } }
##
# @GranuleMode:
diff --git a/include/hw/display/edid.h b/include/hw/display/edid.h
index 520f8ec202..91c0a428af 100644
--- a/include/hw/display/edid.h
+++ b/include/hw/display/edid.h
@@ -1,6 +1,8 @@
#ifndef EDID_H
#define EDID_H
+#define EDID_NAME_MAX_LENGTH 12
+
typedef struct qemu_edid_info {
const char *vendor; /* http://www.uefi.org/pnp_id_list */
const char *name;
diff --git a/include/hw/qdev-properties-system.h b/include/hw/qdev-properties-system.h
index b921392c52..9601a11a09 100644
--- a/include/hw/qdev-properties-system.h
+++ b/include/hw/qdev-properties-system.h
@@ -32,6 +32,7 @@ extern const PropertyInfo qdev_prop_cpus390entitlement;
extern const PropertyInfo qdev_prop_iothread_vq_mapping_list;
extern const PropertyInfo qdev_prop_endian_mode;
extern const PropertyInfo qdev_prop_vmapple_virtio_blk_variant;
+extern const PropertyInfo qdev_prop_virtio_gpu_output_list;
#define DEFINE_PROP_PCI_DEVFN(_n, _s, _f, _d) \
DEFINE_PROP_SIGNED(_n, _s, _f, _d, qdev_prop_pci_devfn, int32_t)
@@ -110,4 +111,8 @@ extern const PropertyInfo qdev_prop_vmapple_virtio_blk_variant;
qdev_prop_vmapple_virtio_blk_variant, \
VMAppleVirtioBlkVariant)
+#define DEFINE_PROP_VIRTIO_GPU_OUTPUT_LIST(_name, _state, _field) \
+ DEFINE_PROP(_name, _state, _field, qdev_prop_virtio_gpu_output_list, \
+ VirtIOGPUOutputList *)
+
#endif
diff --git a/include/hw/virtio/virtio-gpu.h b/include/hw/virtio/virtio-gpu.h
index a42957c4e2..9f16f89a36 100644
--- a/include/hw/virtio/virtio-gpu.h
+++ b/include/hw/virtio/virtio-gpu.h
@@ -20,6 +20,7 @@
#include "hw/virtio/virtio.h"
#include "qemu/log.h"
#include "system/vhost-user-backend.h"
+#include "qapi/qapi-types-virtio.h"
#include "standard-headers/linux/virtio_gpu.h"
#include "standard-headers/linux/virtio_ids.h"
@@ -128,6 +129,7 @@ struct virtio_gpu_base_conf {
uint32_t xres;
uint32_t yres;
uint64_t hostmem;
+ VirtIOGPUOutputList *outputs;
};
struct virtio_gpu_ctrl_command {
@@ -167,6 +169,7 @@ struct VirtIOGPUBaseClass {
#define VIRTIO_GPU_BASE_PROPERTIES(_state, _conf) \
DEFINE_PROP_UINT32("max_outputs", _state, _conf.max_outputs, 1), \
+ DEFINE_PROP_VIRTIO_GPU_OUTPUT_LIST("outputs", _state, _conf.outputs), \
DEFINE_PROP_BIT("edid", _state, _conf.flags, \
VIRTIO_GPU_FLAG_EDID_ENABLED, true), \
DEFINE_PROP_UINT32("xres", _state, _conf.xres, 1280), \
diff --git a/hw/core/qdev-properties-system.c b/hw/core/qdev-properties-system.c
index 24e145d870..1f810b7ddf 100644
--- a/hw/core/qdev-properties-system.c
+++ b/hw/core/qdev-properties-system.c
@@ -1299,3 +1299,47 @@ const PropertyInfo qdev_prop_vmapple_virtio_blk_variant = {
.set = qdev_propinfo_set_enum,
.set_default_value = qdev_propinfo_set_default_value_enum,
};
+
+/* --- VirtIOGPUOutputList --- */
+
+static void get_virtio_gpu_output_list(Object *obj, Visitor *v,
+ const char *name, void *opaque, Error **errp)
+{
+ VirtIOGPUOutputList **prop_ptr =
+ object_field_prop_ptr(obj, opaque);
+
+ visit_type_VirtIOGPUOutputList(v, name, prop_ptr, errp);
+}
+
+static void set_virtio_gpu_output_list(Object *obj, Visitor *v,
+ const char *name, void *opaque, Error **errp)
+{
+ VirtIOGPUOutputList **prop_ptr =
+ object_field_prop_ptr(obj, opaque);
+ VirtIOGPUOutputList *list;
+
+ if (!visit_type_VirtIOGPUOutputList(v, name, &list, errp)) {
+ return;
+ }
+
+ qapi_free_VirtIOGPUOutputList(*prop_ptr);
+ *prop_ptr = list;
+}
+
+static void release_virtio_gpu_output_list(Object *obj,
+ const char *name, void *opaque)
+{
+ VirtIOGPUOutputList **prop_ptr =
+ object_field_prop_ptr(obj, opaque);
+
+ qapi_free_VirtIOGPUOutputList(*prop_ptr);
+ *prop_ptr = NULL;
+}
+
+const PropertyInfo qdev_prop_virtio_gpu_output_list = {
+ .type = "VirtIOGPUOutputList",
+ .description = "VirtIO GPU output list [{\"name\":\"<name>\"},...]",
+ .get = get_virtio_gpu_output_list,
+ .set = set_virtio_gpu_output_list,
+ .release = release_virtio_gpu_output_list,
+};
diff --git a/hw/display/virtio-gpu-base.c b/hw/display/virtio-gpu-base.c
index 9eb806b71f..7269477a1c 100644
--- a/hw/display/virtio-gpu-base.c
+++ b/hw/display/virtio-gpu-base.c
@@ -19,6 +19,7 @@
#include "qemu/error-report.h"
#include "hw/display/edid.h"
#include "trace.h"
+#include "qapi/qapi-types-virtio.h"
void
virtio_gpu_base_reset(VirtIOGPUBase *g)
@@ -56,6 +57,8 @@ void
virtio_gpu_base_generate_edid(VirtIOGPUBase *g, int scanout,
struct virtio_gpu_resp_edid *edid)
{
+ size_t output_idx;
+ VirtIOGPUOutputList *node;
qemu_edid_info info = {
.width_mm = g->req_state[scanout].width_mm,
.height_mm = g->req_state[scanout].height_mm,
@@ -64,6 +67,14 @@ virtio_gpu_base_generate_edid(VirtIOGPUBase *g, int scanout,
.refresh_rate = g->req_state[scanout].refresh_rate,
};
+ for (output_idx = 0, node = g->conf.outputs;
+ output_idx <= scanout && node; output_idx++, node = node->next) {
+ if (output_idx == scanout && node->value && node->value->name) {
+ info.name = node->value->name;
+ break;
+ }
+ }
+
edid->size = cpu_to_le32(sizeof(edid->edid));
qemu_edid_generate(edid->edid, sizeof(edid->edid), &info);
}
@@ -172,6 +183,8 @@ virtio_gpu_base_device_realize(DeviceState *qdev,
VirtIOHandleOutput cursor_cb,
Error **errp)
{
+ size_t output_idx;
+ VirtIOGPUOutputList *node;
VirtIODevice *vdev = VIRTIO_DEVICE(qdev);
VirtIOGPUBase *g = VIRTIO_GPU_BASE(qdev);
int i;
@@ -181,6 +194,20 @@ virtio_gpu_base_device_realize(DeviceState *qdev,
return false;
}
+ for (output_idx = 0, node = g->conf.outputs;
+ node; output_idx++, node = node->next) {
+ if (output_idx == g->conf.max_outputs) {
+ error_setg(errp, "invalid outputs > %d", g->conf.max_outputs);
+ return false;
+ }
+ if (node->value && node->value->name &&
+ strlen(node->value->name) > EDID_NAME_MAX_LENGTH) {
+ error_setg(errp, "invalid output name '%s' > %d",
+ node->value->name, EDID_NAME_MAX_LENGTH);
+ return false;
+ }
+ }
+
if (virtio_gpu_virgl_enabled(g->conf)) {
error_setg(&g->migration_blocker, "virgl is not yet migratable");
if (migrate_add_blocker(&g->migration_blocker, errp) < 0) {
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 11/13] ui/gtk: Add keep-aspect-ratio option
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (9 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 10/13] hw/display: Allow injection of virtio-gpu EDID name marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 12/13] ui/gtk: Add scale option marcandre.lureau
` (3 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Weifeng Liu, Marc-André Lureau, Eric Blake,
Markus Armbruster
From: Weifeng Liu <weifeng.liu.z@gmail.com>
When aspect ratio of host window and that of guest display are not
aligned, we can either zoom the guest content to fill the whole host
window or add padding to respect aspect ratio of the guest. Add an
option keep-aspect-ratio to allow users to select their preferred
behavior in this case.
Suggested-by: BALATON Zoltan <balaton@eik.bme.hu>
Suggested-by: Kim, Dongwon <dongwon.kim@intel.com>
Signed-off-by: Weifeng Liu <weifeng.liu.z@gmail.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20250601045245.36778-2-weifeng.liu.z@gmail.com>
---
qapi/ui.json | 12 ++++++++----
include/ui/gtk.h | 1 +
ui/gtk.c | 12 ++++++++++--
3 files changed, 19 insertions(+), 6 deletions(-)
diff --git a/qapi/ui.json b/qapi/ui.json
index 514fa159b1..9e496b4835 100644
--- a/qapi/ui.json
+++ b/qapi/ui.json
@@ -1335,13 +1335,17 @@
# @show-menubar: Display the main window menubar. Defaults to "on".
# (Since 8.0)
#
+# @keep-aspect-ratio: Keep width/height aspect ratio of guest content when
+# resizing host window. Defaults to "on". (Since 10.1)
+#
# Since: 2.12
##
{ 'struct' : 'DisplayGTK',
- 'data' : { '*grab-on-hover' : 'bool',
- '*zoom-to-fit' : 'bool',
- '*show-tabs' : 'bool',
- '*show-menubar' : 'bool' } }
+ 'data' : { '*grab-on-hover' : 'bool',
+ '*zoom-to-fit' : 'bool',
+ '*show-tabs' : 'bool',
+ '*show-menubar' : 'bool',
+ '*keep-aspect-ratio' : 'bool' } }
##
# @DisplayEGLHeadless:
diff --git a/include/ui/gtk.h b/include/ui/gtk.h
index d3944046db..b7cfbf218e 100644
--- a/include/ui/gtk.h
+++ b/include/ui/gtk.h
@@ -140,6 +140,7 @@ struct GtkDisplayState {
GdkCursor *null_cursor;
Notifier mouse_mode_notifier;
gboolean free_scale;
+ gboolean keep_aspect_ratio;
bool external_pause_update;
diff --git a/ui/gtk.c b/ui/gtk.c
index 8c4a94c8f6..9104509ee1 100644
--- a/ui/gtk.c
+++ b/ui/gtk.c
@@ -828,8 +828,12 @@ void gd_update_scale(VirtualConsole *vc, int ww, int wh, int fbw, int fbh)
sx = (double)ww / fbw;
sy = (double)wh / fbh;
-
- vc->gfx.scale_x = vc->gfx.scale_y = MIN(sx, sy);
+ if (vc->s->keep_aspect_ratio) {
+ vc->gfx.scale_x = vc->gfx.scale_y = MIN(sx, sy);
+ } else {
+ vc->gfx.scale_x = sx;
+ vc->gfx.scale_y = sy;
+ }
}
}
/**
@@ -2328,6 +2332,10 @@ static GSList *gd_vc_gfx_init(GtkDisplayState *s, VirtualConsole *vc,
s->free_scale = true;
}
+ s->keep_aspect_ratio = true;
+ if (s->opts->u.gtk.has_keep_aspect_ratio)
+ s->keep_aspect_ratio = s->opts->u.gtk.keep_aspect_ratio;
+
for (i = 0; i < INPUT_EVENT_SLOTS_MAX; i++) {
struct touch_slot *slot = &touch_slots[i];
slot->tracking_id = -1;
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 12/13] ui/gtk: Add scale option
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (10 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 11/13] ui/gtk: Add keep-aspect-ratio option marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:32 ` [PULL v2 13/13] tpm: "qemu -tpmdev help" should return success marcandre.lureau
` (2 subsequent siblings)
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel
Cc: stefanha, Weifeng Liu, Marc-André Lureau, Eric Blake,
Markus Armbruster
From: Weifeng Liu <weifeng.liu.z@gmail.com>
Allow user to set a preferred scale (defaulting to 1) of the virtual
display. Along with zoom-to-fix=false, this would be helpful for users
running QEMU on hi-dpi host desktop to achieve pixel to pixel display --
e.g., if the scale factor of a user's host desktop is set to 200%, then
they can set a 0.5 scale for the virtual display to avoid magnification
that might cause blurriness.
Signed-off-by: Weifeng Liu <weifeng.liu.z@gmail.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20250601045245.36778-3-weifeng.liu.z@gmail.com>
---
qapi/ui.json | 5 ++++-
include/ui/gtk.h | 1 +
ui/gtk.c | 46 +++++++++++++++++++++++++++++-----------------
3 files changed, 34 insertions(+), 18 deletions(-)
diff --git a/qapi/ui.json b/qapi/ui.json
index 9e496b4835..a465d671c7 100644
--- a/qapi/ui.json
+++ b/qapi/ui.json
@@ -1338,6 +1338,8 @@
# @keep-aspect-ratio: Keep width/height aspect ratio of guest content when
# resizing host window. Defaults to "on". (Since 10.1)
#
+# @scale: Set preferred scale of the display. Defaults to 1.0. (Since 10.1)
+#
# Since: 2.12
##
{ 'struct' : 'DisplayGTK',
@@ -1345,7 +1347,8 @@
'*zoom-to-fit' : 'bool',
'*show-tabs' : 'bool',
'*show-menubar' : 'bool',
- '*keep-aspect-ratio' : 'bool' } }
+ '*keep-aspect-ratio' : 'bool',
+ '*scale' : 'number' } }
##
# @DisplayEGLHeadless:
diff --git a/include/ui/gtk.h b/include/ui/gtk.h
index b7cfbf218e..3e6ce3cb48 100644
--- a/include/ui/gtk.h
+++ b/include/ui/gtk.h
@@ -41,6 +41,7 @@ typedef struct VirtualGfxConsole {
DisplaySurface *ds;
pixman_image_t *convert;
cairo_surface_t *surface;
+ double preferred_scale;
double scale_x;
double scale_y;
#if defined(CONFIG_OPENGL)
diff --git a/ui/gtk.c b/ui/gtk.c
index 9104509ee1..e91d093a49 100644
--- a/ui/gtk.c
+++ b/ui/gtk.c
@@ -67,6 +67,7 @@
#define VC_TERM_X_MIN 80
#define VC_TERM_Y_MIN 25
#define VC_SCALE_MIN 0.25
+#define VC_SCALE_MAX 4
#define VC_SCALE_STEP 0.25
#ifdef GDK_WINDOWING_X11
@@ -272,15 +273,11 @@ static void gd_update_geometry_hints(VirtualConsole *vc)
if (!vc->gfx.ds) {
return;
}
- if (s->free_scale) {
- geo.min_width = surface_width(vc->gfx.ds) * VC_SCALE_MIN;
- geo.min_height = surface_height(vc->gfx.ds) * VC_SCALE_MIN;
- mask |= GDK_HINT_MIN_SIZE;
- } else {
- geo.min_width = surface_width(vc->gfx.ds) * vc->gfx.scale_x;
- geo.min_height = surface_height(vc->gfx.ds) * vc->gfx.scale_y;
- mask |= GDK_HINT_MIN_SIZE;
- }
+ double scale_x = s->free_scale ? VC_SCALE_MIN : vc->gfx.scale_x;
+ double scale_y = s->free_scale ? VC_SCALE_MIN : vc->gfx.scale_y;
+ geo.min_width = surface_width(vc->gfx.ds) * scale_x;
+ geo.min_height = surface_height(vc->gfx.ds) * scale_y;
+ mask |= GDK_HINT_MIN_SIZE;
geo_widget = vc->gfx.drawing_area;
gtk_widget_set_size_request(geo_widget, geo.min_width, geo.min_height);
@@ -1579,8 +1576,8 @@ static void gd_menu_full_screen(GtkMenuItem *item, void *opaque)
}
s->full_screen = FALSE;
if (vc->type == GD_VC_GFX) {
- vc->gfx.scale_x = 1.0;
- vc->gfx.scale_y = 1.0;
+ vc->gfx.scale_x = vc->gfx.preferred_scale;
+ vc->gfx.scale_y = vc->gfx.preferred_scale;
gd_update_windowsize(vc);
}
}
@@ -1636,8 +1633,8 @@ static void gd_menu_zoom_fixed(GtkMenuItem *item, void *opaque)
GtkDisplayState *s = opaque;
VirtualConsole *vc = gd_vc_find_current(s);
- vc->gfx.scale_x = 1.0;
- vc->gfx.scale_y = 1.0;
+ vc->gfx.scale_x = vc->gfx.preferred_scale;
+ vc->gfx.scale_y = vc->gfx.preferred_scale;
gd_update_windowsize(vc);
}
@@ -1651,8 +1648,8 @@ static void gd_menu_zoom_fit(GtkMenuItem *item, void *opaque)
s->free_scale = TRUE;
} else {
s->free_scale = FALSE;
- vc->gfx.scale_x = 1.0;
- vc->gfx.scale_y = 1.0;
+ vc->gfx.scale_x = vc->gfx.preferred_scale;
+ vc->gfx.scale_y = vc->gfx.preferred_scale;
}
gd_update_windowsize(vc);
@@ -2243,6 +2240,11 @@ static void gl_area_realize(GtkGLArea *area, VirtualConsole *vc)
}
#endif
+static bool gd_scale_valid(double scale)
+{
+ return scale >= VC_SCALE_MIN && scale <= VC_SCALE_MAX;
+}
+
static GSList *gd_vc_gfx_init(GtkDisplayState *s, VirtualConsole *vc,
QemuConsole *con, int idx,
GSList *group, GtkWidget *view_menu)
@@ -2252,8 +2254,18 @@ static GSList *gd_vc_gfx_init(GtkDisplayState *s, VirtualConsole *vc,
vc->label = qemu_console_get_label(con);
vc->s = s;
- vc->gfx.scale_x = 1.0;
- vc->gfx.scale_y = 1.0;
+ vc->gfx.preferred_scale = 1.0;
+ if (s->opts->u.gtk.has_scale) {
+ if (gd_scale_valid(s->opts->u.gtk.scale)) {
+ vc->gfx.preferred_scale = s->opts->u.gtk.scale;
+ } else {
+ error_report("Invalid scale value %lf given, being ignored",
+ s->opts->u.gtk.scale);
+ s->opts->u.gtk.has_scale = false;
+ }
+ }
+ vc->gfx.scale_x = vc->gfx.preferred_scale;
+ vc->gfx.scale_y = vc->gfx.preferred_scale;
#if defined(CONFIG_OPENGL)
if (display_opengl) {
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PULL v2 13/13] tpm: "qemu -tpmdev help" should return success
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (11 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 12/13] ui/gtk: Add scale option marcandre.lureau
@ 2025-07-15 6:32 ` marcandre.lureau
2025-07-15 6:38 ` [PULL v2 00/13] Ui patches Marc-André Lureau
2025-07-16 12:40 ` Stefan Hajnoczi
14 siblings, 0 replies; 20+ messages in thread
From: marcandre.lureau @ 2025-07-15 6:32 UTC (permalink / raw)
To: qemu-devel; +Cc: stefanha, Marc-André Lureau, Stefan Berger
From: Marc-André Lureau <marcandre.lureau@redhat.com>
Like other "-foo help" CLI, the qemu process should return 0 for
"-tpmdev help".
While touching this, switch to is_help_option() utility function as
suggested by Peter Maydell.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20250707101412.2055581-1-marcandre.lureau@redhat.com>
---
system/tpm.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/system/tpm.c b/system/tpm.c
index 8df0f6e72b..903b29c043 100644
--- a/system/tpm.c
+++ b/system/tpm.c
@@ -21,6 +21,7 @@
#include "system/tpm.h"
#include "qemu/config-file.h"
#include "qemu/error-report.h"
+#include "qemu/help_option.h"
static QLIST_HEAD(, TPMBackend) tpm_backends =
QLIST_HEAD_INITIALIZER(tpm_backends);
@@ -179,9 +180,9 @@ int tpm_config_parse(QemuOptsList *opts_list, const char *optstr)
{
QemuOpts *opts;
- if (!strcmp(optstr, "help")) {
+ if (is_help_option(optstr)) {
tpm_display_backend_drivers();
- return -1;
+ exit(EXIT_SUCCESS);
}
opts = qemu_opts_parse_noisily(opts_list, optstr, true);
if (!opts) {
--
2.50.0
^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PULL v2 00/13] Ui patches
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (12 preceding siblings ...)
2025-07-15 6:32 ` [PULL v2 13/13] tpm: "qemu -tpmdev help" should return success marcandre.lureau
@ 2025-07-15 6:38 ` Marc-André Lureau
2025-07-16 12:40 ` Stefan Hajnoczi
14 siblings, 0 replies; 20+ messages in thread
From: Marc-André Lureau @ 2025-07-15 6:38 UTC (permalink / raw)
To: qemu-devel; +Cc: stefanha
Hi
On Tue, Jul 15, 2025 at 10:36 AM <marcandre.lureau@redhat.com> wrote:
>
> From: Marc-André Lureau <marcandre.lureau@redhat.com>
>
> The following changes since commit 9a4e273ddec3927920c5958d2226c6b38b543336:
>
> Merge tag 'pull-tcg-20250711' of https://gitlab.com/rth7680/qemu into staging (2025-07-13 01:46:04 -0400)
>
> are available in the Git repository at:
>
> https://gitlab.com/marcandre.lureau/qemu.git tags/ui-pull-request
>
> for you to fetch changes up to df892b3954e5b2782165e6c59e5ffd55c2f7ec5a:
>
> tpm: "qemu -tpmdev help" should return success (2025-07-15 10:22:33 +0400)
>
> ----------------------------------------------------------------
> UI-related for 10.1
>
> - [PATCH v3 0/2] ui/vnc: Do not copy z_stream
> - [PATCH v6 0/7] ui/spice: Enable gl=on option for non-local or remote clients
> - [PATCH v6 0/1] Allow injection of virtio-gpu EDID name
> - [PATCH 0/2] ui/gtk: Add keep-aspect-ratio and scale option
>
> ----------------------------------------------------------------
>
> Akihiko Odaki (2):
> ui/vnc: Do not copy z_stream
> ui/vnc: Introduce the VncWorker type
Checkpatch complains with a false positive here:
ERROR: spaces required around that '*' (ctx:WxB)
#250: FILE: ui/vnc-enc-tight.c:356:
+ data = (uint##bpp##_t *)tight->tight.buffer; \
^
ERROR: spaces required around that '*' (ctx:WxV)
#327: FILE: ui/vnc-enc-tight.c:614:
+ uint##bpp##_t *buf, int w, int h) { \
^
total: 2 errors, 0 warnings, 1790 lines checked
Please ignore
>
> Andrew Keesler (1):
> hw/display: Allow injection of virtio-gpu EDID name
>
> Marc-André Lureau (1):
> tpm: "qemu -tpmdev help" should return success
>
> Vivek Kasireddy (7):
> ui/egl-helpers: Error check the fds in egl_dmabuf_export_texture()
> ui/spice: Enable gl=on option for non-local or remote clients
> ui/spice: Add an option for users to provide a preferred video codec
> ui/spice: Add an option to submit gl_draw requests at fixed rate
> ui/console-gl: Add a helper to create a texture with linear memory
> layout
> ui/spice: Create a new texture with linear layout when gl=on is
> specified
> ui/spice: Blit the scanout texture if its memory layout is not linear
>
> Weifeng Liu (2):
> ui/gtk: Add keep-aspect-ratio option
> ui/gtk: Add scale option
>
> qapi/ui.json | 15 +-
> qapi/virtio.json | 18 +-
> include/hw/display/edid.h | 2 +
> include/hw/qdev-properties-system.h | 5 +
> include/hw/virtio/virtio-gpu.h | 3 +
> include/ui/console.h | 3 +
> include/ui/gtk.h | 2 +
> include/ui/spice-display.h | 5 +
> include/ui/surface.h | 1 +
> ui/vnc.h | 49 +--
> hw/core/qdev-properties-system.c | 44 +++
> hw/display/virtio-gpu-base.c | 27 ++
> system/tpm.c | 5 +-
> ui/console-gl.c | 54 ++++
> ui/egl-helpers.c | 6 +
> ui/gtk.c | 58 ++--
> ui/spice-core.c | 32 ++
> ui/spice-display.c | 226 +++++++++++++-
> ui/vnc-enc-tight.c | 456 +++++++++++++++-------------
> ui/vnc-enc-zlib.c | 47 +--
> ui/vnc-enc-zrle.c | 122 ++++----
> ui/vnc-jobs.c | 13 +-
> ui/vnc.c | 83 +++--
> ui/vnc-enc-zrle.c.inc | 20 +-
> qemu-options.hx | 13 +
> 25 files changed, 883 insertions(+), 426 deletions(-)
>
> --
> 2.50.0
>
>
--
Marc-André Lureau
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PULL v2 00/13] Ui patches
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
` (13 preceding siblings ...)
2025-07-15 6:38 ` [PULL v2 00/13] Ui patches Marc-André Lureau
@ 2025-07-16 12:40 ` Stefan Hajnoczi
14 siblings, 0 replies; 20+ messages in thread
From: Stefan Hajnoczi @ 2025-07-16 12:40 UTC (permalink / raw)
To: marcandre.lureau; +Cc: qemu-devel, stefanha, Marc-André Lureau
[-- Attachment #1: Type: text/plain, Size: 116 bytes --]
Applied, thanks.
Please update the changelog at https://wiki.qemu.org/ChangeLog/10.1 for any user-visible changes.
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PULL v2 01/13] ui/vnc: Do not copy z_stream
2025-07-15 6:32 ` [PULL v2 01/13] ui/vnc: Do not copy z_stream marcandre.lureau
@ 2025-07-16 15:07 ` Michael Tokarev
2025-07-16 15:49 ` Daniel P. Berrangé
0 siblings, 1 reply; 20+ messages in thread
From: Michael Tokarev @ 2025-07-16 15:07 UTC (permalink / raw)
To: marcandre.lureau, qemu-devel; +Cc: stefanha, Akihiko Odaki, qemu-stable
On 15.07.2025 09:32, marcandre.lureau@redhat.com wrote:
> From: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
>
> vnc_worker_thread_loop() copies z_stream stored in its local VncState to
> the persistent VncState, and the copied one is freed with deflateEnd()
> later. However, deflateEnd() refuses to operate with a copied z_stream
> and returns Z_STREAM_ERROR, leaking the allocated memory.
>
> Avoid copying the zlib state to fix the memory leak.
>
> Fixes: bd023f953e5e ("vnc: threaded VNC server")
> Signed-off-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
> Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> Message-Id: <20250603-zlib-v3-1-20b857bd8d05@rsg.ci.i.u-tokyo.ac.jp>
> ---
> ui/vnc.h | 2 +-
> ui/vnc-enc-zlib.c | 30 +++++++++++++++---------------
> ui/vnc.c | 13 ++++++++++---
> 3 files changed, 26 insertions(+), 19 deletions(-)
This looks like a qemu-stable material, is it not?
Thanks,
/mjt
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PULL v2 01/13] ui/vnc: Do not copy z_stream
2025-07-16 15:07 ` Michael Tokarev
@ 2025-07-16 15:49 ` Daniel P. Berrangé
0 siblings, 0 replies; 20+ messages in thread
From: Daniel P. Berrangé @ 2025-07-16 15:49 UTC (permalink / raw)
To: Michael Tokarev
Cc: marcandre.lureau, qemu-devel, stefanha, Akihiko Odaki,
qemu-stable
On Wed, Jul 16, 2025 at 06:07:31PM +0300, Michael Tokarev wrote:
> On 15.07.2025 09:32, marcandre.lureau@redhat.com wrote:
> > From: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
> >
> > vnc_worker_thread_loop() copies z_stream stored in its local VncState to
> > the persistent VncState, and the copied one is freed with deflateEnd()
> > later. However, deflateEnd() refuses to operate with a copied z_stream
> > and returns Z_STREAM_ERROR, leaking the allocated memory.
> >
> > Avoid copying the zlib state to fix the memory leak.
> >
> > Fixes: bd023f953e5e ("vnc: threaded VNC server")
> > Signed-off-by: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
> > Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
> > Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> > Message-Id: <20250603-zlib-v3-1-20b857bd8d05@rsg.ci.i.u-tokyo.ac.jp>
> > ---
> > ui/vnc.h | 2 +-
> > ui/vnc-enc-zlib.c | 30 +++++++++++++++---------------
> > ui/vnc.c | 13 ++++++++++---
> > 3 files changed, 26 insertions(+), 19 deletions(-)
>
> This looks like a qemu-stable material, is it not?
Yes, please cherry-pick this.
With regards,
Daniel
--
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org -o- https://fstop138.berrange.com :|
|: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PULL v2 09/13] ui/spice: Blit the scanout texture if its memory layout is not linear
2025-07-15 6:32 ` [PULL v2 09/13] ui/spice: Blit the scanout texture if its memory layout is not linear marcandre.lureau
@ 2025-07-17 9:23 ` Peter Maydell
2025-07-17 18:17 ` Kasireddy, Vivek
0 siblings, 1 reply; 20+ messages in thread
From: Peter Maydell @ 2025-07-17 9:23 UTC (permalink / raw)
To: marcandre.lureau
Cc: qemu-devel, stefanha, Vivek Kasireddy, Gerd Hoffmann,
Dmitry Osipenko, Frediano Ziglio, Dongwon Kim, Michael Scherle
On Tue, 15 Jul 2025 at 07:43, <marcandre.lureau@redhat.com> wrote:
>
> From: Vivek Kasireddy <vivek.kasireddy@intel.com>
>
> In cases where the scanout buffer is provided as a texture (e.g. Virgl)
> we need to check to see if it has a linear memory layout or not. If
> it doesn't have a linear layout, then blitting it onto the texture
> associated with the display surface (which already has a linear layout)
> seems to ensure that there is no corruption seen regardless of which
> encoder or decoder is used.
Hi; Coverity points out (CID 1612367) that this change
introduces a use of an uninitialized variable:
> @@ -1269,6 +1329,15 @@ static void qemu_spice_gl_update(DisplayChangeListener *dcl,
> glFlush();
> }
>
> + if (spice_remote_client && ssd->blit_scanout_texture) {
> + egl_fb scanout_tex_fb;
We don't initialize scanout_tex_fb...
> +
> + ret = spice_gl_blit_scanout_texture(ssd, &scanout_tex_fb);
...and we immediately pass it to spice_gl_blit_scanout_texture(),
which will unconditionally call egl_fb_destroy() on it.
> + if (!ret) {
> + return;
> + }
> + }
thanks
-- PMM
^ permalink raw reply [flat|nested] 20+ messages in thread
* RE: [PULL v2 09/13] ui/spice: Blit the scanout texture if its memory layout is not linear
2025-07-17 9:23 ` Peter Maydell
@ 2025-07-17 18:17 ` Kasireddy, Vivek
0 siblings, 0 replies; 20+ messages in thread
From: Kasireddy, Vivek @ 2025-07-17 18:17 UTC (permalink / raw)
To: Peter Maydell, marcandre.lureau@redhat.com
Cc: qemu-devel@nongnu.org, stefanha@redhat.com, Gerd Hoffmann,
Dmitry Osipenko, Frediano Ziglio, Kim, Dongwon, Michael Scherle
Hi Peter, Marc-Andre,
> Subject: Re: [PULL v2 09/13] ui/spice: Blit the scanout texture if its memory
> layout is not linear
>
> On Tue, 15 Jul 2025 at 07:43, <marcandre.lureau@redhat.com> wrote:
> >
> > From: Vivek Kasireddy <vivek.kasireddy@intel.com>
> >
> > In cases where the scanout buffer is provided as a texture (e.g. Virgl)
> > we need to check to see if it has a linear memory layout or not. If
> > it doesn't have a linear layout, then blitting it onto the texture
> > associated with the display surface (which already has a linear layout)
> > seems to ensure that there is no corruption seen regardless of which
> > encoder or decoder is used.
>
> Hi; Coverity points out (CID 1612367) that this change
> introduces a use of an uninitialized variable:
>
> > @@ -1269,6 +1329,15 @@ static void
> qemu_spice_gl_update(DisplayChangeListener *dcl,
> > glFlush();
> > }
> >
> > + if (spice_remote_client && ssd->blit_scanout_texture) {
> > + egl_fb scanout_tex_fb;
>
> We don't initialize scanout_tex_fb...
>
> > +
> > + ret = spice_gl_blit_scanout_texture(ssd, &scanout_tex_fb);
>
> ...and we immediately pass it to spice_gl_blit_scanout_texture(),
> which will unconditionally call egl_fb_destroy() on it.
I'll send out a patch in a day or two to fix this issue.
Thanks,
Vivek
>
> > + if (!ret) {
> > + return;
> > + }
> > + }
>
> thanks
> -- PMM
^ permalink raw reply [flat|nested] 20+ messages in thread
end of thread, other threads:[~2025-07-17 20:39 UTC | newest]
Thread overview: 20+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-07-15 6:32 [PULL v2 00/13] Ui patches marcandre.lureau
2025-07-15 6:32 ` [PULL v2 01/13] ui/vnc: Do not copy z_stream marcandre.lureau
2025-07-16 15:07 ` Michael Tokarev
2025-07-16 15:49 ` Daniel P. Berrangé
2025-07-15 6:32 ` [PULL v2 02/13] ui/vnc: Introduce the VncWorker type marcandre.lureau
2025-07-15 6:32 ` [PULL v2 03/13] ui/egl-helpers: Error check the fds in egl_dmabuf_export_texture() marcandre.lureau
2025-07-15 6:32 ` [PULL v2 04/13] ui/spice: Enable gl=on option for non-local or remote clients marcandre.lureau
2025-07-15 6:32 ` [PULL v2 05/13] ui/spice: Add an option for users to provide a preferred video codec marcandre.lureau
2025-07-15 6:32 ` [PULL v2 06/13] ui/spice: Add an option to submit gl_draw requests at fixed rate marcandre.lureau
2025-07-15 6:32 ` [PULL v2 07/13] ui/console-gl: Add a helper to create a texture with linear memory layout marcandre.lureau
2025-07-15 6:32 ` [PULL v2 08/13] ui/spice: Create a new texture with linear layout when gl=on is specified marcandre.lureau
2025-07-15 6:32 ` [PULL v2 09/13] ui/spice: Blit the scanout texture if its memory layout is not linear marcandre.lureau
2025-07-17 9:23 ` Peter Maydell
2025-07-17 18:17 ` Kasireddy, Vivek
2025-07-15 6:32 ` [PULL v2 10/13] hw/display: Allow injection of virtio-gpu EDID name marcandre.lureau
2025-07-15 6:32 ` [PULL v2 11/13] ui/gtk: Add keep-aspect-ratio option marcandre.lureau
2025-07-15 6:32 ` [PULL v2 12/13] ui/gtk: Add scale option marcandre.lureau
2025-07-15 6:32 ` [PULL v2 13/13] tpm: "qemu -tpmdev help" should return success marcandre.lureau
2025-07-15 6:38 ` [PULL v2 00/13] Ui patches Marc-André Lureau
2025-07-16 12:40 ` Stefan Hajnoczi
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).