weston/tests/weston-test-client-helper.c

1938 lines
48 KiB
C
Raw Normal View History

/*
* Copyright © 2012 Intel Corporation
* Copyright © 2015 Samsung Electronics Co., Ltd
* Copyright 2016, 2017 Collabora, Ltd.
*
* Permission is hereby granted, free of charge, to any person obtaining
* a copy of this software and associated documentation files (the
* "Software"), to deal in the Software without restriction, including
* without limitation the rights to use, copy, modify, merge, publish,
* distribute, sublicense, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
*
* The above copyright notice and this permission notice (including the
* next paragraph) shall be included in all copies or substantial
* portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#include "config.h"
#include <stdlib.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include <errno.h>
#include <sys/mman.h>
#include <cairo.h>
#include "test-config.h"
#include "shared/os-compatibility.h"
#include "shared/xalloc.h"
#include <libweston/zalloc.h>
#include "weston-test-client-helper.h"
#define max(a, b) (((a) > (b)) ? (a) : (b))
#define min(a, b) (((a) > (b)) ? (b) : (a))
#define clip(x, a, b) min(max(x, a), b)
int
surface_contains(struct surface *surface, int x, int y)
{
/* test whether a global x,y point is contained in the surface */
int sx = surface->x;
int sy = surface->y;
int sw = surface->width;
int sh = surface->height;
return x >= sx && y >= sy && x < sx + sw && y < sy + sh;
}
static void
frame_callback_handler(void *data, struct wl_callback *callback, uint32_t time)
{
int *done = data;
*done = 1;
wl_callback_destroy(callback);
}
static const struct wl_callback_listener frame_listener = {
frame_callback_handler
};
struct wl_callback *
frame_callback_set(struct wl_surface *surface, int *done)
{
struct wl_callback *callback;
*done = 0;
callback = wl_surface_frame(surface);
wl_callback_add_listener(callback, &frame_listener, done);
return callback;
}
int
frame_callback_wait_nofail(struct client *client, int *done)
{
while (!*done) {
if (wl_display_dispatch(client->wl_display) < 0)
return 0;
}
return 1;
}
void
move_client(struct client *client, int x, int y)
{
struct surface *surface = client->surface;
int done;
client->surface->x = x;
client->surface->y = y;
weston_test_move_surface(client->test->weston_test, surface->wl_surface,
surface->x, surface->y);
2014-12-11 05:33:47 +03:00
/* The attach here is necessary because commit() will call configure
* only on surfaces newly attached, and the one that sets the surface
* position is the configure. */
tests: introduce struct buffer for client-helper We are growing more tests that need to handle buffers, both just images and wl_buffers. Particularly the screenshooting facility needs these. Currently everything is in struct surface, which contains more than we need. It is a bit messy. Create a new struct buffer to encapsulate the image representation, the wl_buffer, and enough information to tear it all down (munmap) so we don't have to leak everything. Some tests might start doing things in loops, and leaking would accumulate. Instead of inventing our own image representation, use pixman_image_t. It is a well-tested library worth using, and we already rely on it in other places. This makes the tests depend on Pixman, which requires the fix for building buffer-count, which would otherwise not find pixman.h. The new create_shm_buffer_a8r8g8b8() creates an image with an explicit format, and pixman_image_t keeps track of it. And stride and size and data. This implementation is still a little hacky due to calling create_shm_buffer(). A very new thing is buffer_destroy(). Previously we didn't really free any buffers. It is not a problem when the process will exit soon anyway, but it may become a problem if tests start iterating things. Manual memset() on a image is converted to a pixman action, just to show how to do it properly with pixman. Stride and pixel format assumptions still linger all around, but those are for another patch. Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk> Reviewed-by: Daniel Stone <daniels@collabora.com>
2016-05-20 17:25:38 +03:00
wl_surface_attach(surface->wl_surface, surface->buffer->proxy, 0, 0);
wl_surface_damage(surface->wl_surface, 0, 0, surface->width,
surface->height);
frame_callback_set(surface->wl_surface, &done);
wl_surface_commit(surface->wl_surface);
frame_callback_wait(client, &done);
}
static void
pointer_handle_enter(void *data, struct wl_pointer *wl_pointer,
uint32_t serial, struct wl_surface *wl_surface,
wl_fixed_t x, wl_fixed_t y)
{
struct pointer *pointer = data;
if (wl_surface)
pointer->focus = wl_surface_get_user_data(wl_surface);
else
pointer->focus = NULL;
pointer->x = wl_fixed_to_int(x);
pointer->y = wl_fixed_to_int(y);
testlog("test-client: got pointer enter %d %d, surface %p\n",
pointer->x, pointer->y, pointer->focus);
}
static void
pointer_handle_leave(void *data, struct wl_pointer *wl_pointer,
uint32_t serial, struct wl_surface *wl_surface)
{
struct pointer *pointer = data;
pointer->focus = NULL;
testlog("test-client: got pointer leave, surface %p\n",
wl_surface ? wl_surface_get_user_data(wl_surface) : NULL);
}
static void
pointer_handle_motion(void *data, struct wl_pointer *wl_pointer,
uint32_t time_msec, wl_fixed_t x, wl_fixed_t y)
{
struct pointer *pointer = data;
pointer->x = wl_fixed_to_int(x);
pointer->y = wl_fixed_to_int(y);
pointer->motion_time_msec = time_msec;
pointer->motion_time_timespec = pointer->input_timestamp;
pointer->input_timestamp = (struct timespec) { 0 };
testlog("test-client: got pointer motion %d %d\n",
pointer->x, pointer->y);
}
static void
pointer_handle_button(void *data, struct wl_pointer *wl_pointer,
uint32_t serial, uint32_t time_msec, uint32_t button,
uint32_t state)
{
struct pointer *pointer = data;
pointer->button = button;
pointer->state = state;
pointer->button_time_msec = time_msec;
pointer->button_time_timespec = pointer->input_timestamp;
pointer->input_timestamp = (struct timespec) { 0 };
testlog("test-client: got pointer button %u %u\n", button, state);
}
static void
pointer_handle_axis(void *data, struct wl_pointer *wl_pointer,
uint32_t time_msec, uint32_t axis, wl_fixed_t value)
{
struct pointer *pointer = data;
pointer->axis = axis;
pointer->axis_value = wl_fixed_to_double(value);
pointer->axis_time_msec = time_msec;
pointer->axis_time_timespec = pointer->input_timestamp;
pointer->input_timestamp = (struct timespec) { 0 };
testlog("test-client: got pointer axis %u %f\n",
axis, wl_fixed_to_double(value));
}
static void
pointer_handle_frame(void *data, struct wl_pointer *wl_pointer)
{
testlog("test-client: got pointer frame\n");
}
static void
pointer_handle_axis_source(void *data, struct wl_pointer *wl_pointer,
uint32_t source)
{
testlog("test-client: got pointer axis source %u\n", source);
}
static void
pointer_handle_axis_stop(void *data, struct wl_pointer *wl_pointer,
uint32_t time_msec, uint32_t axis)
{
struct pointer *pointer = data;
pointer->axis = axis;
pointer->axis_stop_time_msec = time_msec;
pointer->axis_stop_time_timespec = pointer->input_timestamp;
pointer->input_timestamp = (struct timespec) { 0 };
testlog("test-client: got pointer axis stop %u\n", axis);
}
static void
pointer_handle_axis_discrete(void *data, struct wl_pointer *wl_pointer,
uint32_t axis, int32_t value)
{
testlog("test-client: got pointer axis discrete %u %d\n", axis, value);
}
static const struct wl_pointer_listener pointer_listener = {
pointer_handle_enter,
pointer_handle_leave,
pointer_handle_motion,
pointer_handle_button,
pointer_handle_axis,
pointer_handle_frame,
pointer_handle_axis_source,
pointer_handle_axis_stop,
pointer_handle_axis_discrete,
};
static void
keyboard_handle_keymap(void *data, struct wl_keyboard *wl_keyboard,
uint32_t format, int fd, uint32_t size)
{
close(fd);
testlog("test-client: got keyboard keymap\n");
}
static void
keyboard_handle_enter(void *data, struct wl_keyboard *wl_keyboard,
uint32_t serial, struct wl_surface *wl_surface,
struct wl_array *keys)
{
struct keyboard *keyboard = data;
if (wl_surface)
keyboard->focus = wl_surface_get_user_data(wl_surface);
else
keyboard->focus = NULL;
testlog("test-client: got keyboard enter, surface %p\n",
keyboard->focus);
}
static void
keyboard_handle_leave(void *data, struct wl_keyboard *wl_keyboard,
uint32_t serial, struct wl_surface *wl_surface)
{
struct keyboard *keyboard = data;
keyboard->focus = NULL;
testlog("test-client: got keyboard leave, surface %p\n",
wl_surface ? wl_surface_get_user_data(wl_surface) : NULL);
}
static void
keyboard_handle_key(void *data, struct wl_keyboard *wl_keyboard,
uint32_t serial, uint32_t time_msec, uint32_t key,
uint32_t state)
{
struct keyboard *keyboard = data;
keyboard->key = key;
keyboard->state = state;
keyboard->key_time_msec = time_msec;
keyboard->key_time_timespec = keyboard->input_timestamp;
keyboard->input_timestamp = (struct timespec) { 0 };
testlog("test-client: got keyboard key %u %u\n", key, state);
}
static void
keyboard_handle_modifiers(void *data, struct wl_keyboard *wl_keyboard,
uint32_t serial, uint32_t mods_depressed,
uint32_t mods_latched, uint32_t mods_locked,
uint32_t group)
{
struct keyboard *keyboard = data;
keyboard->mods_depressed = mods_depressed;
keyboard->mods_latched = mods_latched;
keyboard->mods_locked = mods_locked;
keyboard->group = group;
testlog("test-client: got keyboard modifiers %u %u %u %u\n",
mods_depressed, mods_latched, mods_locked, group);
}
static void
keyboard_handle_repeat_info(void *data, struct wl_keyboard *wl_keyboard,
int32_t rate, int32_t delay)
{
struct keyboard *keyboard = data;
keyboard->repeat_info.rate = rate;
keyboard->repeat_info.delay = delay;
testlog("test-client: got keyboard repeat_info %d %d\n", rate, delay);
}
static const struct wl_keyboard_listener keyboard_listener = {
keyboard_handle_keymap,
keyboard_handle_enter,
keyboard_handle_leave,
keyboard_handle_key,
keyboard_handle_modifiers,
keyboard_handle_repeat_info,
};
static void
touch_handle_down(void *data, struct wl_touch *wl_touch,
uint32_t serial, uint32_t time_msec,
struct wl_surface *surface, int32_t id,
wl_fixed_t x_w, wl_fixed_t y_w)
{
struct touch *touch = data;
touch->down_x = wl_fixed_to_int(x_w);
touch->down_y = wl_fixed_to_int(y_w);
touch->id = id;
touch->down_time_msec = time_msec;
touch->down_time_timespec = touch->input_timestamp;
touch->input_timestamp = (struct timespec) { 0 };
testlog("test-client: got touch down %d %d, surf: %p, id: %d\n",
touch->down_x, touch->down_y, surface, id);
}
static void
touch_handle_up(void *data, struct wl_touch *wl_touch,
uint32_t serial, uint32_t time_msec, int32_t id)
{
struct touch *touch = data;
touch->up_id = id;
touch->up_time_msec = time_msec;
touch->up_time_timespec = touch->input_timestamp;
touch->input_timestamp = (struct timespec) { 0 };
testlog("test-client: got touch up, id: %d\n", id);
}
static void
touch_handle_motion(void *data, struct wl_touch *wl_touch,
uint32_t time_msec, int32_t id,
wl_fixed_t x_w, wl_fixed_t y_w)
{
struct touch *touch = data;
touch->x = wl_fixed_to_int(x_w);
touch->y = wl_fixed_to_int(y_w);
touch->motion_time_msec = time_msec;
touch->motion_time_timespec = touch->input_timestamp;
touch->input_timestamp = (struct timespec) { 0 };
testlog("test-client: got touch motion, %d %d, id: %d\n",
touch->x, touch->y, id);
}
static void
touch_handle_frame(void *data, struct wl_touch *wl_touch)
{
struct touch *touch = data;
++touch->frame_no;
testlog("test-client: got touch frame (%d)\n", touch->frame_no);
}
static void
touch_handle_cancel(void *data, struct wl_touch *wl_touch)
{
struct touch *touch = data;
++touch->cancel_no;
testlog("test-client: got touch cancel (%d)\n", touch->cancel_no);
}
static const struct wl_touch_listener touch_listener = {
touch_handle_down,
touch_handle_up,
touch_handle_motion,
touch_handle_frame,
touch_handle_cancel,
};
static void
surface_enter(void *data,
struct wl_surface *wl_surface, struct wl_output *output)
{
struct surface *surface = data;
surface->output = wl_output_get_user_data(output);
testlog("test-client: got surface enter output %p\n", surface->output);
}
static void
surface_leave(void *data,
struct wl_surface *wl_surface, struct wl_output *output)
{
struct surface *surface = data;
surface->output = NULL;
testlog("test-client: got surface leave output %p\n",
wl_output_get_user_data(output));
}
static const struct wl_surface_listener surface_listener = {
surface_enter,
surface_leave
};
static struct buffer *
create_shm_buffer(struct client *client, int width, int height,
pixman_format_code_t format, uint32_t wlfmt)
{
struct wl_shm *shm = client->wl_shm;
struct buffer *buf;
size_t stride_bytes;
struct wl_shm_pool *pool;
int fd;
void *data;
size_t bytes_pp;
assert(width > 0);
assert(height > 0);
buf = xzalloc(sizeof *buf);
bytes_pp = PIXMAN_FORMAT_BPP(format) / 8;
stride_bytes = width * bytes_pp;
/* round up to multiple of 4 bytes for Pixman */
stride_bytes = (stride_bytes + 3) & ~3u;
assert(stride_bytes / bytes_pp >= (unsigned)width);
buf->len = stride_bytes * height;
assert(buf->len / stride_bytes == (unsigned)height);
fd = os_create_anonymous_file(buf->len);
assert(fd >= 0);
data = mmap(NULL, buf->len, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
if (data == MAP_FAILED) {
close(fd);
assert(data != MAP_FAILED);
}
pool = wl_shm_create_pool(shm, fd, buf->len);
buf->proxy = wl_shm_pool_create_buffer(pool, 0, width, height,
stride_bytes, wlfmt);
wl_shm_pool_destroy(pool);
close(fd);
buf->image = pixman_image_create_bits(format, width, height,
data, stride_bytes);
assert(buf->proxy);
assert(buf->image);
return buf;
}
tests: introduce struct buffer for client-helper We are growing more tests that need to handle buffers, both just images and wl_buffers. Particularly the screenshooting facility needs these. Currently everything is in struct surface, which contains more than we need. It is a bit messy. Create a new struct buffer to encapsulate the image representation, the wl_buffer, and enough information to tear it all down (munmap) so we don't have to leak everything. Some tests might start doing things in loops, and leaking would accumulate. Instead of inventing our own image representation, use pixman_image_t. It is a well-tested library worth using, and we already rely on it in other places. This makes the tests depend on Pixman, which requires the fix for building buffer-count, which would otherwise not find pixman.h. The new create_shm_buffer_a8r8g8b8() creates an image with an explicit format, and pixman_image_t keeps track of it. And stride and size and data. This implementation is still a little hacky due to calling create_shm_buffer(). A very new thing is buffer_destroy(). Previously we didn't really free any buffers. It is not a problem when the process will exit soon anyway, but it may become a problem if tests start iterating things. Manual memset() on a image is converted to a pixman action, just to show how to do it properly with pixman. Stride and pixel format assumptions still linger all around, but those are for another patch. Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk> Reviewed-by: Daniel Stone <daniels@collabora.com>
2016-05-20 17:25:38 +03:00
struct buffer *
create_shm_buffer_a8r8g8b8(struct client *client, int width, int height)
{
assert(client->has_argb);
tests: introduce struct buffer for client-helper We are growing more tests that need to handle buffers, both just images and wl_buffers. Particularly the screenshooting facility needs these. Currently everything is in struct surface, which contains more than we need. It is a bit messy. Create a new struct buffer to encapsulate the image representation, the wl_buffer, and enough information to tear it all down (munmap) so we don't have to leak everything. Some tests might start doing things in loops, and leaking would accumulate. Instead of inventing our own image representation, use pixman_image_t. It is a well-tested library worth using, and we already rely on it in other places. This makes the tests depend on Pixman, which requires the fix for building buffer-count, which would otherwise not find pixman.h. The new create_shm_buffer_a8r8g8b8() creates an image with an explicit format, and pixman_image_t keeps track of it. And stride and size and data. This implementation is still a little hacky due to calling create_shm_buffer(). A very new thing is buffer_destroy(). Previously we didn't really free any buffers. It is not a problem when the process will exit soon anyway, but it may become a problem if tests start iterating things. Manual memset() on a image is converted to a pixman action, just to show how to do it properly with pixman. Stride and pixel format assumptions still linger all around, but those are for another patch. Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk> Reviewed-by: Daniel Stone <daniels@collabora.com>
2016-05-20 17:25:38 +03:00
return create_shm_buffer(client, width, height,
PIXMAN_a8r8g8b8, WL_SHM_FORMAT_ARGB8888);
tests: introduce struct buffer for client-helper We are growing more tests that need to handle buffers, both just images and wl_buffers. Particularly the screenshooting facility needs these. Currently everything is in struct surface, which contains more than we need. It is a bit messy. Create a new struct buffer to encapsulate the image representation, the wl_buffer, and enough information to tear it all down (munmap) so we don't have to leak everything. Some tests might start doing things in loops, and leaking would accumulate. Instead of inventing our own image representation, use pixman_image_t. It is a well-tested library worth using, and we already rely on it in other places. This makes the tests depend on Pixman, which requires the fix for building buffer-count, which would otherwise not find pixman.h. The new create_shm_buffer_a8r8g8b8() creates an image with an explicit format, and pixman_image_t keeps track of it. And stride and size and data. This implementation is still a little hacky due to calling create_shm_buffer(). A very new thing is buffer_destroy(). Previously we didn't really free any buffers. It is not a problem when the process will exit soon anyway, but it may become a problem if tests start iterating things. Manual memset() on a image is converted to a pixman action, just to show how to do it properly with pixman. Stride and pixel format assumptions still linger all around, but those are for another patch. Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk> Reviewed-by: Daniel Stone <daniels@collabora.com>
2016-05-20 17:25:38 +03:00
}
void
buffer_destroy(struct buffer *buf)
{
void *pixels;
pixels = pixman_image_get_data(buf->image);
if (buf->proxy) {
wl_buffer_destroy(buf->proxy);
assert(munmap(pixels, buf->len) == 0);
}
assert(pixman_image_unref(buf->image));
free(buf);
}
static void
shm_format(void *data, struct wl_shm *wl_shm, uint32_t format)
{
struct client *client = data;
if (format == WL_SHM_FORMAT_ARGB8888)
client->has_argb = 1;
}
struct wl_shm_listener shm_listener = {
shm_format
};
static void
test_handle_pointer_position(void *data, struct weston_test *weston_test,
wl_fixed_t x, wl_fixed_t y)
{
struct test *test = data;
test->pointer_x = wl_fixed_to_int(x);
test->pointer_y = wl_fixed_to_int(y);
testlog("test-client: got global pointer %d %d\n",
test->pointer_x, test->pointer_y);
}
static void
test_handle_capture_screenshot_done(void *data, struct weston_screenshooter *screenshooter)
{
struct client *client = data;
testlog("Screenshot has been captured\n");
client->buffer_copy_done = true;
}
static const struct weston_screenshooter_listener screenshooter_listener = {
test_handle_capture_screenshot_done
};
static const struct weston_test_listener test_listener = {
test_handle_pointer_position,
};
static void
input_destroy(struct input *inp)
{
if (inp->pointer) {
wl_pointer_release(inp->pointer->wl_pointer);
free(inp->pointer);
}
if (inp->keyboard) {
wl_keyboard_release(inp->keyboard->wl_keyboard);
free(inp->keyboard);
}
if (inp->touch) {
wl_touch_release(inp->touch->wl_touch);
free(inp->touch);
}
wl_list_remove(&inp->link);
wl_seat_release(inp->wl_seat);
free(inp->seat_name);
free(inp);
}
static void
input_update_devices(struct input *input)
{
struct pointer *pointer;
struct keyboard *keyboard;
struct touch *touch;
struct wl_seat *seat = input->wl_seat;
enum wl_seat_capability caps = input->caps;
if ((caps & WL_SEAT_CAPABILITY_POINTER) && !input->pointer) {
pointer = xzalloc(sizeof *pointer);
pointer->wl_pointer = wl_seat_get_pointer(seat);
wl_pointer_set_user_data(pointer->wl_pointer, pointer);
wl_pointer_add_listener(pointer->wl_pointer, &pointer_listener,
pointer);
input->pointer = pointer;
} else if (!(caps & WL_SEAT_CAPABILITY_POINTER) && input->pointer) {
wl_pointer_destroy(input->pointer->wl_pointer);
free(input->pointer);
input->pointer = NULL;
}
if ((caps & WL_SEAT_CAPABILITY_KEYBOARD) && !input->keyboard) {
keyboard = xzalloc(sizeof *keyboard);
keyboard->wl_keyboard = wl_seat_get_keyboard(seat);
wl_keyboard_set_user_data(keyboard->wl_keyboard, keyboard);
wl_keyboard_add_listener(keyboard->wl_keyboard, &keyboard_listener,
keyboard);
input->keyboard = keyboard;
} else if (!(caps & WL_SEAT_CAPABILITY_KEYBOARD) && input->keyboard) {
wl_keyboard_destroy(input->keyboard->wl_keyboard);
free(input->keyboard);
input->keyboard = NULL;
}
if ((caps & WL_SEAT_CAPABILITY_TOUCH) && !input->touch) {
touch = xzalloc(sizeof *touch);
touch->wl_touch = wl_seat_get_touch(seat);
wl_touch_set_user_data(touch->wl_touch, touch);
wl_touch_add_listener(touch->wl_touch, &touch_listener,
touch);
input->touch = touch;
} else if (!(caps & WL_SEAT_CAPABILITY_TOUCH) && input->touch) {
wl_touch_destroy(input->touch->wl_touch);
free(input->touch);
input->touch = NULL;
}
}
static void
seat_handle_capabilities(void *data, struct wl_seat *seat,
enum wl_seat_capability caps)
{
struct input *input = data;
input->caps = caps;
/* we will create/update the devices only with the right (test) seat.
* If we haven't discovered which seat is the test seat, just
* store capabilities and bail out */
if (input->seat_name && strcmp(input->seat_name, "test-seat") == 0)
input_update_devices(input);
testlog("test-client: got seat %p capabilities: %x\n", input, caps);
}
static void
seat_handle_name(void *data, struct wl_seat *seat, const char *name)
{
struct input *input = data;
input->seat_name = strdup(name);
assert(input->seat_name && "No memory");
/* We only update the devices and set client input for the test seat */
if (strcmp(name, "test-seat") == 0) {
assert(!input->client->input &&
"Multiple test seats detected!");
input_update_devices(input);
input->client->input = input;
}
testlog("test-client: got seat %p name: \'%s\'\n", input, name);
}
static const struct wl_seat_listener seat_listener = {
seat_handle_capabilities,
seat_handle_name,
};
static void
output_handle_geometry(void *data,
struct wl_output *wl_output,
int x, int y,
int physical_width,
int physical_height,
int subpixel,
const char *make,
const char *model,
int32_t transform)
{
struct output *output = data;
output->x = x;
output->y = y;
}
static void
output_handle_mode(void *data,
struct wl_output *wl_output,
uint32_t flags,
int width,
int height,
int refresh)
{
struct output *output = data;
if (flags & WL_OUTPUT_MODE_CURRENT) {
output->width = width;
output->height = height;
}
}
static void
output_handle_scale(void *data,
struct wl_output *wl_output,
int scale)
{
struct output *output = data;
output->scale = scale;
}
static void
output_handle_done(void *data,
struct wl_output *wl_output)
{
struct output *output = data;
output->initialized = 1;
}
static const struct wl_output_listener output_listener = {
output_handle_geometry,
output_handle_mode,
output_handle_done,
output_handle_scale,
};
static void
output_destroy(struct output *output)
{
assert(wl_proxy_get_version((struct wl_proxy *)output->wl_output) >= 3);
wl_output_release(output->wl_output);
wl_list_remove(&output->link);
free(output);
}
static void
handle_global(void *data, struct wl_registry *registry,
uint32_t id, const char *interface, uint32_t version)
{
struct client *client = data;
struct output *output;
struct test *test;
struct global *global;
struct input *input;
global = xzalloc(sizeof *global);
global->name = id;
global->interface = strdup(interface);
assert(interface);
global->version = version;
wl_list_insert(client->global_list.prev, &global->link);
/* We deliberately bind all globals with the maximum (advertised)
* version, because this test suite must be kept up-to-date with
* Weston. We must always implement at least the version advertised
* by Weston. This is not ok for normal clients, but it is ok in
* this test suite.
*/
if (strcmp(interface, "wl_compositor") == 0) {
client->wl_compositor =
wl_registry_bind(registry, id,
&wl_compositor_interface, version);
} else if (strcmp(interface, "wl_seat") == 0) {
input = xzalloc(sizeof *input);
input->client = client;
input->global_name = global->name;
input->wl_seat =
wl_registry_bind(registry, id,
&wl_seat_interface, version);
wl_seat_add_listener(input->wl_seat, &seat_listener, input);
wl_list_insert(&client->inputs, &input->link);
} else if (strcmp(interface, "wl_shm") == 0) {
client->wl_shm =
wl_registry_bind(registry, id,
&wl_shm_interface, version);
wl_shm_add_listener(client->wl_shm, &shm_listener, client);
} else if (strcmp(interface, "wl_output") == 0) {
output = xzalloc(sizeof *output);
output->wl_output =
wl_registry_bind(registry, id,
&wl_output_interface, version);
wl_output_add_listener(output->wl_output,
&output_listener, output);
wl_list_insert(&client->output_list, &output->link);
client->output = output;
} else if (strcmp(interface, "weston_test") == 0) {
test = xzalloc(sizeof *test);
test->weston_test =
wl_registry_bind(registry, id,
&weston_test_interface, version);
weston_test_add_listener(test->weston_test, &test_listener, test);
client->test = test;
} else if (strcmp(interface, "weston_screenshooter") == 0) {
client->screenshooter =
wl_registry_bind(registry, id,
&weston_screenshooter_interface, 1);
weston_screenshooter_add_listener(client->screenshooter,
&screenshooter_listener, client);
}
}
static struct global *
client_find_global_with_name(struct client *client, uint32_t name)
{
struct global *global;
wl_list_for_each(global, &client->global_list, link) {
if (global->name == name)
return global;
}
return NULL;
}
static struct input *
client_find_input_with_name(struct client *client, uint32_t name)
{
struct input *input;
wl_list_for_each(input, &client->inputs, link) {
if (input->global_name == name)
return input;
}
return NULL;
}
static void
global_destroy(struct global *global)
{
wl_list_remove(&global->link);
free(global->interface);
free(global);
}
static void
handle_global_remove(void *data, struct wl_registry *registry, uint32_t name)
{
struct client *client = data;
struct global *global;
struct input *input;
global = client_find_global_with_name(client, name);
assert(global && "Request to remove unknown global");
if (strcmp(global->interface, "wl_seat") == 0) {
input = client_find_input_with_name(client, name);
if (input) {
if (client->input == input)
client->input = NULL;
input_destroy(input);
}
}
/* XXX: handle wl_output */
global_destroy(global);
}
static const struct wl_registry_listener registry_listener = {
handle_global,
handle_global_remove,
};
void
skip(const char *fmt, ...)
{
va_list argp;
va_start(argp, fmt);
vfprintf(stderr, fmt, argp);
va_end(argp);
/* automake tests uses exit code 77. weston-test-runner will see
* this and use it, and then weston-test's sigchld handler (in the
* weston process) will use that as an exit status, which is what
* ninja will see in the end. */
exit(77);
}
void
expect_protocol_error(struct client *client,
const struct wl_interface *intf,
uint32_t code)
{
int err;
uint32_t errcode, failed = 0;
const struct wl_interface *interface;
unsigned int id;
/* if the error has not come yet, make it happen */
wl_display_roundtrip(client->wl_display);
err = wl_display_get_error(client->wl_display);
assert(err && "Expected protocol error but nothing came");
assert(err == EPROTO && "Expected protocol error but got local error");
errcode = wl_display_get_protocol_error(client->wl_display,
&interface, &id);
/* check error */
if (errcode != code) {
testlog("Should get error code %d but got %d\n", code, errcode);
failed = 1;
}
/* this should be definitely set */
assert(interface);
if (strcmp(intf->name, interface->name) != 0) {
testlog("Should get interface '%s' but got '%s'\n",
intf->name, interface->name);
failed = 1;
}
if (failed) {
testlog("Expected other protocol error\n");
abort();
}
/* all OK */
testlog("Got expected protocol error on '%s' (object id: %d) "
"with code %d\n", interface->name, id, errcode);
}
static void
log_handler(const char *fmt, va_list args)
{
fprintf(stderr, "libwayland: ");
vfprintf(stderr, fmt, args);
}
struct client *
create_client(void)
{
struct client *client;
wl_log_set_handler_client(log_handler);
/* connect to display */
client = xzalloc(sizeof *client);
client->wl_display = wl_display_connect(NULL);
assert(client->wl_display);
wl_list_init(&client->global_list);
wl_list_init(&client->inputs);
wl_list_init(&client->output_list);
/* setup registry so we can bind to interfaces */
client->wl_registry = wl_display_get_registry(client->wl_display);
wl_registry_add_listener(client->wl_registry, &registry_listener,
client);
/* this roundtrip makes sure we have all globals and we bound to them */
client_roundtrip(client);
/* this roundtrip makes sure we got all wl_shm.format and wl_seat.*
* events */
client_roundtrip(client);
/* must have WL_SHM_FORMAT_ARGB32 */
assert(client->has_argb);
/* must have weston_test interface */
assert(client->test);
/* must have an output */
assert(client->output);
/* the output must be initialized */
assert(client->output->initialized == 1);
/* must have seat set */
assert(client->input);
return client;
}
struct surface *
create_test_surface(struct client *client)
{
struct surface *surface;
surface = xzalloc(sizeof *surface);
surface->wl_surface =
wl_compositor_create_surface(client->wl_compositor);
assert(surface->wl_surface);
wl_surface_add_listener(surface->wl_surface, &surface_listener,
surface);
wl_surface_set_user_data(surface->wl_surface, surface);
return surface;
}
void
surface_destroy(struct surface *surface)
{
if (surface->wl_surface)
wl_surface_destroy(surface->wl_surface);
if (surface->buffer)
buffer_destroy(surface->buffer);
free(surface);
}
struct client *
create_client_and_test_surface(int x, int y, int width, int height)
{
struct client *client;
struct surface *surface;
pixman_color_t color = { 16384, 16384, 16384, 16384 }; /* uint16_t */
pixman_image_t *solid;
client = create_client();
/* initialize the client surface */
surface = create_test_surface(client);
client->surface = surface;
surface->width = width;
surface->height = height;
tests: introduce struct buffer for client-helper We are growing more tests that need to handle buffers, both just images and wl_buffers. Particularly the screenshooting facility needs these. Currently everything is in struct surface, which contains more than we need. It is a bit messy. Create a new struct buffer to encapsulate the image representation, the wl_buffer, and enough information to tear it all down (munmap) so we don't have to leak everything. Some tests might start doing things in loops, and leaking would accumulate. Instead of inventing our own image representation, use pixman_image_t. It is a well-tested library worth using, and we already rely on it in other places. This makes the tests depend on Pixman, which requires the fix for building buffer-count, which would otherwise not find pixman.h. The new create_shm_buffer_a8r8g8b8() creates an image with an explicit format, and pixman_image_t keeps track of it. And stride and size and data. This implementation is still a little hacky due to calling create_shm_buffer(). A very new thing is buffer_destroy(). Previously we didn't really free any buffers. It is not a problem when the process will exit soon anyway, but it may become a problem if tests start iterating things. Manual memset() on a image is converted to a pixman action, just to show how to do it properly with pixman. Stride and pixel format assumptions still linger all around, but those are for another patch. Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk> Reviewed-by: Daniel Stone <daniels@collabora.com>
2016-05-20 17:25:38 +03:00
surface->buffer = create_shm_buffer_a8r8g8b8(client, width, height);
solid = pixman_image_create_solid_fill(&color);
pixman_image_composite32(PIXMAN_OP_SRC,
solid, /* src */
NULL, /* mask */
surface->buffer->image, /* dst */
0, 0, /* src x,y */
0, 0, /* mask x,y */
0, 0, /* dst x,y */
width, height);
pixman_image_unref(solid);
move_client(client, x, y);
return client;
}
void
client_destroy(struct client *client)
{
if (client->surface)
surface_destroy(client->surface);
while (!wl_list_empty(&client->inputs)) {
input_destroy(container_of(client->inputs.next,
struct input, link));
}
while (!wl_list_empty(&client->output_list)) {
output_destroy(container_of(client->output_list.next,
struct output, link));
}
while (!wl_list_empty(&client->global_list)) {
global_destroy(container_of(client->global_list.next,
struct global, link));
}
if (client->test) {
weston_test_destroy(client->test->weston_test);
free(client->test);
}
if (client->wl_shm)
wl_shm_destroy(client->wl_shm);
if (client->wl_compositor)
wl_compositor_destroy(client->wl_compositor);
if (client->wl_registry)
wl_registry_destroy(client->wl_registry);
client_roundtrip(client);
if (client->wl_display)
wl_display_disconnect(client->wl_display);
free(client);
}
static const char*
output_path(void)
{
char *path = getenv("WESTON_TEST_OUTPUT_PATH");
if (!path)
return ".";
return path;
}
char*
screenshot_output_filename(const char *basename, uint32_t seq)
{
char *filename;
if (asprintf(&filename, "%s/%s-%02d.png",
output_path(), basename, seq) < 0)
return NULL;
return filename;
}
static const char*
reference_path(void)
{
char *path = getenv("WESTON_TEST_REFERENCE_PATH");
if (!path)
return WESTON_TEST_REFERENCE_PATH;
return path;
}
char*
screenshot_reference_filename(const char *basename, uint32_t seq)
{
char *filename;
if (asprintf(&filename, "%s/%s-%02d.png",
reference_path(), basename, seq) < 0)
return NULL;
return filename;
}
tests: add output transform tests This goes through all output transforms with two different buffer transforms and verifies the visual output against reference images. This commit introduces a new test input image 'basic-test-card.png'. It is a small image with deliberately odd and indivisible dimensions to provoke bad assumptions about image sizes. It contains red, green and blue areas which are actually text that makes it very obvious if you have e.g. color channels swapped. It has a white thick circle to highlight aspect ratio issues, and an orange cross to show a mixed color. The white border is for contrast and a 1px wide detail. The whole design makes it clear if the image happens to be rotated or flipped in any way. The image has one pixel wide transparent border so that bilinear sampling filter near the edges of the image would produce the same colors with both Pixman- and GL-renderers which handle the out-of-image samples fundamentally differently: Pixman assumes (0, 0, 0, 0) samples outside of the image, while GL-renderer clamps sample coordinates to the edge essentially repeating the edge pixels. It would have been "easy" to create a full matrix of every output scale & transform x every buffer scale & transform, but that would have resulted in 2 renderers * 8 output transforms * 3 output scales * 8 buffer transforms * 3 buffer scales = 1152 test cases that would have all ran strictly serially because our test harness has no parallelism inside one test program. That would have been slow to run, and need a lot more reference images too. Instead, I chose to iterate separately through all output scales & transforms (this patch) and all buffer scales & transforms (next patch). This limits the number of test cases in this patch to 56, and allows the two test programs to run in parallel. I did not even pick all possible scale & transform combinations here, but just what I think is a representative sub-set to hopefully exercise all the code paths. https://gitlab.freedesktop.org/wayland/weston/issues/52 Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2020-01-21 13:00:28 +03:00
char *
image_filename(const char *basename)
{
char *filename;
if (asprintf(&filename, "%s/%s.png", reference_path(), basename) < 0)
assert(0);
return filename;
}
struct format_map_entry {
cairo_format_t cairo;
pixman_format_code_t pixman;
};
static const struct format_map_entry format_map[] = {
{ CAIRO_FORMAT_ARGB32, PIXMAN_a8r8g8b8 },
{ CAIRO_FORMAT_RGB24, PIXMAN_x8r8g8b8 },
{ CAIRO_FORMAT_A8, PIXMAN_a8 },
{ CAIRO_FORMAT_RGB16_565, PIXMAN_r5g6b5 },
};
static pixman_format_code_t
format_cairo2pixman(cairo_format_t fmt)
{
unsigned i;
for (i = 0; i < ARRAY_LENGTH(format_map); i++)
if (format_map[i].cairo == fmt)
return format_map[i].pixman;
assert(0 && "unknown Cairo pixel format");
}
static cairo_format_t
format_pixman2cairo(pixman_format_code_t fmt)
{
unsigned i;
for (i = 0; i < ARRAY_LENGTH(format_map); i++)
if (format_map[i].pixman == fmt)
return format_map[i].cairo;
assert(0 && "unknown Pixman pixel format");
}
/**
* Validate range
*
* \param r Range to validate or NULL.
* \return The given range, or {0, 0} for NULL.
*
* Will abort if range is invalid, that is a > b.
*/
static struct range
range_get(const struct range *r)
{
if (!r)
return (struct range){ 0, 0 };
assert(r->a <= r->b);
return *r;
}
/**
* Compute the ROI for image comparisons
*
* \param img_a An image.
* \param img_b Another image.
* \param clip_rect Explicit ROI, or NULL for using the whole
* image area.
*
* \return The region of interest (ROI) that is guaranteed to be inside both
* images.
*
* If clip_rect is given, it must fall inside of both images.
* If clip_rect is NULL, the images must be of the same size.
* If any precondition is violated, this function aborts with an error.
*
* The ROI is given as pixman_box32_t, where x2,y2 are non-inclusive.
*/
static pixman_box32_t
image_check_get_roi(pixman_image_t *img_a, pixman_image_t *img_b,
const struct rectangle *clip_rect)
{
int width_a;
int width_b;
int height_a;
int height_b;
pixman_box32_t box;
width_a = pixman_image_get_width(img_a);
height_a = pixman_image_get_height(img_a);
width_b = pixman_image_get_width(img_b);
height_b = pixman_image_get_height(img_b);
if (clip_rect) {
box.x1 = clip_rect->x;
box.y1 = clip_rect->y;
box.x2 = clip_rect->x + clip_rect->width;
box.y2 = clip_rect->y + clip_rect->height;
} else {
box.x1 = 0;
box.y1 = 0;
box.x2 = max(width_a, width_b);
box.y2 = max(height_a, height_b);
}
assert(box.x1 >= 0);
assert(box.y1 >= 0);
assert(box.x2 > box.x1);
assert(box.y2 > box.y1);
assert(box.x2 <= width_a);
assert(box.x2 <= width_b);
assert(box.y2 <= height_a);
assert(box.y2 <= height_b);
return box;
}
struct image_iterator {
char *data;
int stride; /* bytes */
};
static void
image_iter_init(struct image_iterator *it, pixman_image_t *image)
{
pixman_format_code_t fmt;
it->stride = pixman_image_get_stride(image);
it->data = (void *)pixman_image_get_data(image);
fmt = pixman_image_get_format(image);
assert(PIXMAN_FORMAT_BPP(fmt) == 32);
}
tests: introduce struct buffer for client-helper We are growing more tests that need to handle buffers, both just images and wl_buffers. Particularly the screenshooting facility needs these. Currently everything is in struct surface, which contains more than we need. It is a bit messy. Create a new struct buffer to encapsulate the image representation, the wl_buffer, and enough information to tear it all down (munmap) so we don't have to leak everything. Some tests might start doing things in loops, and leaking would accumulate. Instead of inventing our own image representation, use pixman_image_t. It is a well-tested library worth using, and we already rely on it in other places. This makes the tests depend on Pixman, which requires the fix for building buffer-count, which would otherwise not find pixman.h. The new create_shm_buffer_a8r8g8b8() creates an image with an explicit format, and pixman_image_t keeps track of it. And stride and size and data. This implementation is still a little hacky due to calling create_shm_buffer(). A very new thing is buffer_destroy(). Previously we didn't really free any buffers. It is not a problem when the process will exit soon anyway, but it may become a problem if tests start iterating things. Manual memset() on a image is converted to a pixman action, just to show how to do it properly with pixman. Stride and pixel format assumptions still linger all around, but those are for another patch. Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk> Reviewed-by: Daniel Stone <daniels@collabora.com>
2016-05-20 17:25:38 +03:00
static uint32_t *
image_iter_get_row(struct image_iterator *it, int y)
{
return (uint32_t *)(it->data + y * it->stride);
}
struct pixel_diff_stat {
struct pixel_diff_stat_channel {
int min_diff;
int max_diff;
} ch[4];
};
static void
testlog_pixel_diff_stat(const struct pixel_diff_stat *stat)
{
int i;
testlog("Image difference statistics:\n");
for (i = 0; i < 4; i++) {
testlog("\tch %d: [%d, %d]\n",
i, stat->ch[i].min_diff, stat->ch[i].max_diff);
}
}
static bool
fuzzy_match_pixels(uint32_t pix_a, uint32_t pix_b,
const struct range *fuzz,
struct pixel_diff_stat *stat)
{
bool ret = true;
int shift;
int i;
for (shift = 0, i = 0; i < 4; shift += 8, i++) {
int val_a = (pix_a >> shift) & 0xffu;
int val_b = (pix_b >> shift) & 0xffu;
int d = val_b - val_a;
stat->ch[i].min_diff = min(stat->ch[i].min_diff, d);
stat->ch[i].max_diff = max(stat->ch[i].max_diff, d);
if (d < fuzz->a || d > fuzz->b)
ret = false;
}
return ret;
}
/**
* Test if a given region within two images are pixel-identical
*
* Returns true if the two images pixel-wise identical, and false otherwise.
*
* \param img_a First image.
* \param img_b Second image.
* \param clip_rect The region of interest, or NULL for comparing the whole
* images.
* \param prec Per-channel allowed difference, or NULL for identical match
* required.
*
* This function hard-fails if clip_rect is not inside both images. If clip_rect
* is given, the images do not have to match in size, otherwise size mismatch
* will be a hard failure.
*
* The per-pixel, per-channel difference is computed as img_b - img_a which is
* required to be in the range [prec->a, prec->b] inclusive. The difference is
* signed. All four channels are compared the same way, without any special
* meaning on alpha channel.
*/
bool
check_images_match(pixman_image_t *img_a, pixman_image_t *img_b,
const struct rectangle *clip_rect, const struct range *prec)
{
struct range fuzz = range_get(prec);
struct pixel_diff_stat diffstat = {};
struct image_iterator it_a;
struct image_iterator it_b;
pixman_box32_t box;
int x, y;
uint32_t *pix_a;
uint32_t *pix_b;
box = image_check_get_roi(img_a, img_b, clip_rect);
image_iter_init(&it_a, img_a);
image_iter_init(&it_b, img_b);
for (y = box.y1; y < box.y2; y++) {
pix_a = image_iter_get_row(&it_a, y) + box.x1;
pix_b = image_iter_get_row(&it_b, y) + box.x1;
for (x = box.x1; x < box.x2; x++) {
if (!fuzzy_match_pixels(*pix_a, *pix_b,
&fuzz, &diffstat))
return false;
pix_a++;
pix_b++;
}
}
return true;
}
/**
* Tint a color
*
* \param src Source pixel as x8r8g8b8.
* \param add The tint as x8r8g8b8, x8 must be zero; r8, g8 and b8 must be
* no greater than 0xc0 to avoid overflow to another channel.
* \return The tinted pixel color as x8r8g8b8, x8 guaranteed to be 0xff.
*
* The source pixel RGB values are divided by 4, and then the tint is added.
* To achieve colors outside of the range of src, a tint color channel must be
* at least 0x40. (0xff / 4 = 0x3f, 0xff - 0x3f = 0xc0)
*/
static uint32_t
tint(uint32_t src, uint32_t add)
{
uint32_t v;
v = ((src & 0xfcfcfcfc) >> 2) | 0xff000000;
return v + add;
}
/**
* Create a visualization of image differences.
*
* \param img_a First image, which is used as the basis for the output.
* \param img_b Second image.
* \param clip_rect The region of interest, or NULL for comparing the whole
* images.
* \param prec Per-channel allowed difference, or NULL for identical match
* required.
* \return A new image with the differences highlighted.
*
* Regions outside of the region of interest are shaded with black, matching
* pixels are shaded with green, and differing pixels are shaded with
* bright red.
*
* This function hard-fails if clip_rect is not inside both images. If clip_rect
* is given, the images do not have to match in size, otherwise size mismatch
* will be a hard failure.
*
* The per-pixel, per-channel difference is computed as img_b - img_a which is
* required to be in the range [prec->a, prec->b] inclusive. The difference is
* signed. All four channels are compared the same way, without any special
* meaning on alpha channel.
*/
pixman_image_t *
visualize_image_difference(pixman_image_t *img_a, pixman_image_t *img_b,
const struct rectangle *clip_rect,
const struct range *prec)
{
struct range fuzz = range_get(prec);
struct pixel_diff_stat diffstat = {};
pixman_image_t *diffimg;
pixman_image_t *shade;
struct image_iterator it_a;
struct image_iterator it_b;
struct image_iterator it_d;
int width;
int height;
pixman_box32_t box;
int x, y;
uint32_t *pix_a;
uint32_t *pix_b;
uint32_t *pix_d;
pixman_color_t shade_color = { 0, 0, 0, 32768 };
width = pixman_image_get_width(img_a);
height = pixman_image_get_height(img_a);
box = image_check_get_roi(img_a, img_b, clip_rect);
diffimg = pixman_image_create_bits_no_clear(PIXMAN_x8r8g8b8,
width, height, NULL, 0);
/* Fill diffimg with a black-shaded copy of img_a, and then fill
* the clip_rect area with original img_a.
*/
shade = pixman_image_create_solid_fill(&shade_color);
pixman_image_composite32(PIXMAN_OP_SRC, img_a, shade, diffimg,
0, 0, 0, 0, 0, 0, width, height);
pixman_image_unref(shade);
pixman_image_composite32(PIXMAN_OP_SRC, img_a, NULL, diffimg,
box.x1, box.y1, 0, 0, box.x1, box.y1,
box.x2 - box.x1, box.y2 - box.y1);
image_iter_init(&it_a, img_a);
image_iter_init(&it_b, img_b);
image_iter_init(&it_d, diffimg);
for (y = box.y1; y < box.y2; y++) {
pix_a = image_iter_get_row(&it_a, y) + box.x1;
pix_b = image_iter_get_row(&it_b, y) + box.x1;
pix_d = image_iter_get_row(&it_d, y) + box.x1;
for (x = box.x1; x < box.x2; x++) {
if (fuzzy_match_pixels(*pix_a, *pix_b,
&fuzz, &diffstat))
*pix_d = tint(*pix_d, 0x00008000); /* green */
else
*pix_d = tint(*pix_d, 0x00c00000); /* red */
pix_a++;
pix_b++;
pix_d++;
}
}
testlog_pixel_diff_stat(&diffstat);
return diffimg;
}
/**
* Write an image into a PNG file.
*
* \param image The image.
* \param fname The name and path for the file.
*
* \returns true if successfully saved file; false otherwise.
*
* \note Only image formats directly supported by Cairo are accepted, not all
* Pixman formats.
*/
bool
write_image_as_png(pixman_image_t *image, const char *fname)
{
cairo_surface_t *cairo_surface;
cairo_status_t status;
cairo_format_t fmt;
fmt = format_pixman2cairo(pixman_image_get_format(image));
cairo_surface = cairo_image_surface_create_for_data(
(void *)pixman_image_get_data(image),
fmt,
pixman_image_get_width(image),
pixman_image_get_height(image),
pixman_image_get_stride(image));
status = cairo_surface_write_to_png(cairo_surface, fname);
if (status != CAIRO_STATUS_SUCCESS) {
testlog("Failed to save image '%s': %s\n", fname,
cairo_status_to_string(status));
return false;
}
cairo_surface_destroy(cairo_surface);
return true;
}
static pixman_image_t *
image_convert_to_a8r8g8b8(pixman_image_t *image)
{
pixman_image_t *ret;
int width;
int height;
if (pixman_image_get_format(image) == PIXMAN_a8r8g8b8)
return pixman_image_ref(image);
width = pixman_image_get_width(image);
height = pixman_image_get_height(image);
ret = pixman_image_create_bits_no_clear(PIXMAN_a8r8g8b8, width, height,
NULL, 0);
assert(ret);
pixman_image_composite32(PIXMAN_OP_SRC, image, NULL, ret,
0, 0, 0, 0, 0, 0, width, height);
return ret;
}
static void
destroy_cairo_surface(pixman_image_t *image, void *data)
{
cairo_surface_t *surface = data;
cairo_surface_destroy(surface);
}
/**
* Load an image from a PNG file
*
* Reads a PNG image from disk using the given filename (and path)
* and returns as a Pixman image. Use pixman_image_unref() to free it.
*
* The returned image is always in PIXMAN_a8r8g8b8 format.
*
* @returns Pixman image, or NULL in case of error.
*/
pixman_image_t *
load_image_from_png(const char *fname)
{
pixman_image_t *image;
pixman_image_t *converted;
cairo_format_t cairo_fmt;
pixman_format_code_t pixman_fmt;
cairo_surface_t *reference_cairo_surface;
cairo_status_t status;
int width;
int height;
int stride;
void *data;
reference_cairo_surface = cairo_image_surface_create_from_png(fname);
cairo_surface_flush(reference_cairo_surface);
status = cairo_surface_status(reference_cairo_surface);
if (status != CAIRO_STATUS_SUCCESS) {
testlog("Could not open %s: %s\n", fname,
cairo_status_to_string(status));
cairo_surface_destroy(reference_cairo_surface);
return NULL;
}
cairo_fmt = cairo_image_surface_get_format(reference_cairo_surface);
pixman_fmt = format_cairo2pixman(cairo_fmt);
width = cairo_image_surface_get_width(reference_cairo_surface);
height = cairo_image_surface_get_height(reference_cairo_surface);
stride = cairo_image_surface_get_stride(reference_cairo_surface);
data = cairo_image_surface_get_data(reference_cairo_surface);
/* The Cairo surface will own the data, so we keep it around. */
image = pixman_image_create_bits_no_clear(pixman_fmt,
width, height, data, stride);
assert(image);
tests: introduce struct buffer for client-helper We are growing more tests that need to handle buffers, both just images and wl_buffers. Particularly the screenshooting facility needs these. Currently everything is in struct surface, which contains more than we need. It is a bit messy. Create a new struct buffer to encapsulate the image representation, the wl_buffer, and enough information to tear it all down (munmap) so we don't have to leak everything. Some tests might start doing things in loops, and leaking would accumulate. Instead of inventing our own image representation, use pixman_image_t. It is a well-tested library worth using, and we already rely on it in other places. This makes the tests depend on Pixman, which requires the fix for building buffer-count, which would otherwise not find pixman.h. The new create_shm_buffer_a8r8g8b8() creates an image with an explicit format, and pixman_image_t keeps track of it. And stride and size and data. This implementation is still a little hacky due to calling create_shm_buffer(). A very new thing is buffer_destroy(). Previously we didn't really free any buffers. It is not a problem when the process will exit soon anyway, but it may become a problem if tests start iterating things. Manual memset() on a image is converted to a pixman action, just to show how to do it properly with pixman. Stride and pixel format assumptions still linger all around, but those are for another patch. Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk> Reviewed-by: Daniel Stone <daniels@collabora.com>
2016-05-20 17:25:38 +03:00
pixman_image_set_destroy_function(image, destroy_cairo_surface,
reference_cairo_surface);
tests: introduce struct buffer for client-helper We are growing more tests that need to handle buffers, both just images and wl_buffers. Particularly the screenshooting facility needs these. Currently everything is in struct surface, which contains more than we need. It is a bit messy. Create a new struct buffer to encapsulate the image representation, the wl_buffer, and enough information to tear it all down (munmap) so we don't have to leak everything. Some tests might start doing things in loops, and leaking would accumulate. Instead of inventing our own image representation, use pixman_image_t. It is a well-tested library worth using, and we already rely on it in other places. This makes the tests depend on Pixman, which requires the fix for building buffer-count, which would otherwise not find pixman.h. The new create_shm_buffer_a8r8g8b8() creates an image with an explicit format, and pixman_image_t keeps track of it. And stride and size and data. This implementation is still a little hacky due to calling create_shm_buffer(). A very new thing is buffer_destroy(). Previously we didn't really free any buffers. It is not a problem when the process will exit soon anyway, but it may become a problem if tests start iterating things. Manual memset() on a image is converted to a pixman action, just to show how to do it properly with pixman. Stride and pixel format assumptions still linger all around, but those are for another patch. Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk> Reviewed-by: Daniel Stone <daniels@collabora.com>
2016-05-20 17:25:38 +03:00
converted = image_convert_to_a8r8g8b8(image);
pixman_image_unref(image);
return converted;
}
/**
* Take screenshot of a single output
*
* Requests a screenshot from the server of the output that the
* client appears on. This implies that the compositor goes through an output
* repaint to provide the screenshot before this function returns. This
* function is therefore both a server roundtrip and a wait for a repaint.
*
* @returns A new buffer object, that should be freed with buffer_destroy().
*/
struct buffer *
capture_screenshot_of_output(struct client *client)
{
struct buffer *buffer;
assert(client->screenshooter);
buffer = create_shm_buffer_a8r8g8b8(client,
client->output->width,
client->output->height);
client->buffer_copy_done = false;
weston_screenshooter_take_shot(client->screenshooter,
client->output->wl_output,
buffer->proxy);
while (client->buffer_copy_done == false)
assert(wl_display_dispatch(client->wl_display) >= 0);
/* FIXME: Document somewhere the orientation the screenshot is taken
* and how the clip coords are interpreted, in case of scaling/transform.
* If we're using read_pixels() just make sure it is documented somewhere.
* Protocol docs in the XML, comparison function docs in Doxygen style.
*/
return buffer;
}
static void
write_visual_diff(pixman_image_t *ref_image,
struct buffer *shot,
const struct rectangle *clip,
const char *test_name,
int seq_no,
const struct range *fuzz)
{
char *fname;
char *ext_test_name;
pixman_image_t *diff;
int ret;
ret = asprintf(&ext_test_name, "%s-diff", test_name);
assert(ret >= 0);
fname = screenshot_output_filename(ext_test_name, seq_no);
diff = visualize_image_difference(ref_image, shot->image, clip, fuzz);
write_image_as_png(diff, fname);
pixman_image_unref(diff);
free(fname);
free(ext_test_name);
}
/**
* Verify image contents
*
* Compares the contents of the given shot to the given reference
* image over the given clip rectangle, reports whether they match to the
* test log, and if they do not match writes a visual diff into a PNG file
* and the screenshot into another PNG file named with get_test_name() and
* seq_no.
*
* The shot image size and the reference image size must both contain
* the clip rectangle.
*
* This function uses the pixel value allowed fuzz approriate for GL-renderer
* with 8 bits per channel data.
*
* \param shot The image to be verified, usually a screenshot.
* \param ref_image The reference image file basename, without sequence number
* and .png suffix.
* \param ref_seq_no The reference image sequence number.
* \param clip The region of interest, or NULL for comparing the whole
* images.
* \param seq_no Test sequence number, for writing output files.
* \return True if the shot matches the reference image, false otherwise.
*
* For bootstrapping, ref_image can be NULL or the file can be missing.
* In that case the screenshot file is written but no comparison is performed,
* and false is returned.
*
* \sa verify_screen_content
*/
bool
verify_image(struct buffer *shot,
const char *ref_image,
int ref_seq_no,
const struct rectangle *clip,
int seq_no)
{
const char *test_name = get_test_name();
const struct range gl_fuzz = { -3, 4 };
pixman_image_t *ref = NULL;
char *ref_fname = NULL;
char *shot_fname;
bool match = false;
shot_fname = screenshot_output_filename(test_name, seq_no);
if (ref_image) {
ref_fname = screenshot_reference_filename(ref_image, ref_seq_no);
ref = load_image_from_png(ref_fname);
}
if (ref) {
match = check_images_match(ref, shot->image, clip, &gl_fuzz);
testlog("Verify reference image %s vs. shot %s: %s\n",
ref_fname, shot_fname, match ? "PASS" : "FAIL");
if (!match) {
write_visual_diff(ref, shot, clip,
test_name, seq_no, &gl_fuzz);
}
pixman_image_unref(ref);
} else {
testlog("No reference image, shot %s: FAIL\n", shot_fname);
}
if (!match)
write_image_as_png(shot->image, shot_fname);
free(ref_fname);
free(shot_fname);
return match;
}
tests: add output transform tests This goes through all output transforms with two different buffer transforms and verifies the visual output against reference images. This commit introduces a new test input image 'basic-test-card.png'. It is a small image with deliberately odd and indivisible dimensions to provoke bad assumptions about image sizes. It contains red, green and blue areas which are actually text that makes it very obvious if you have e.g. color channels swapped. It has a white thick circle to highlight aspect ratio issues, and an orange cross to show a mixed color. The white border is for contrast and a 1px wide detail. The whole design makes it clear if the image happens to be rotated or flipped in any way. The image has one pixel wide transparent border so that bilinear sampling filter near the edges of the image would produce the same colors with both Pixman- and GL-renderers which handle the out-of-image samples fundamentally differently: Pixman assumes (0, 0, 0, 0) samples outside of the image, while GL-renderer clamps sample coordinates to the edge essentially repeating the edge pixels. It would have been "easy" to create a full matrix of every output scale & transform x every buffer scale & transform, but that would have resulted in 2 renderers * 8 output transforms * 3 output scales * 8 buffer transforms * 3 buffer scales = 1152 test cases that would have all ran strictly serially because our test harness has no parallelism inside one test program. That would have been slow to run, and need a lot more reference images too. Instead, I chose to iterate separately through all output scales & transforms (this patch) and all buffer scales & transforms (next patch). This limits the number of test cases in this patch to 56, and allows the two test programs to run in parallel. I did not even pick all possible scale & transform combinations here, but just what I think is a representative sub-set to hopefully exercise all the code paths. https://gitlab.freedesktop.org/wayland/weston/issues/52 Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2020-01-21 13:00:28 +03:00
/**
* Take a screenshot and verify its contents
*
* Takes a screenshot and calls verify_image() with it.
*
* \param client The client, for connecting to the compositor.
* \param ref_image See verify_image().
* \param ref_seq_no See verify_image().
* \param clip See verify_image().
* \param seq_no See verify_image().
* \return True if the screen contents matches the reference image,
* false otherwise.
*/
bool
verify_screen_content(struct client *client,
const char *ref_image,
int ref_seq_no,
const struct rectangle *clip,
int seq_no)
{
struct buffer *shot;
bool match;
shot = capture_screenshot_of_output(client);
assert(shot);
match = verify_image(shot, ref_image, ref_seq_no, clip, seq_no);
buffer_destroy(shot);
return match;
}
tests: add output transform tests This goes through all output transforms with two different buffer transforms and verifies the visual output against reference images. This commit introduces a new test input image 'basic-test-card.png'. It is a small image with deliberately odd and indivisible dimensions to provoke bad assumptions about image sizes. It contains red, green and blue areas which are actually text that makes it very obvious if you have e.g. color channels swapped. It has a white thick circle to highlight aspect ratio issues, and an orange cross to show a mixed color. The white border is for contrast and a 1px wide detail. The whole design makes it clear if the image happens to be rotated or flipped in any way. The image has one pixel wide transparent border so that bilinear sampling filter near the edges of the image would produce the same colors with both Pixman- and GL-renderers which handle the out-of-image samples fundamentally differently: Pixman assumes (0, 0, 0, 0) samples outside of the image, while GL-renderer clamps sample coordinates to the edge essentially repeating the edge pixels. It would have been "easy" to create a full matrix of every output scale & transform x every buffer scale & transform, but that would have resulted in 2 renderers * 8 output transforms * 3 output scales * 8 buffer transforms * 3 buffer scales = 1152 test cases that would have all ran strictly serially because our test harness has no parallelism inside one test program. That would have been slow to run, and need a lot more reference images too. Instead, I chose to iterate separately through all output scales & transforms (this patch) and all buffer scales & transforms (next patch). This limits the number of test cases in this patch to 56, and allows the two test programs to run in parallel. I did not even pick all possible scale & transform combinations here, but just what I think is a representative sub-set to hopefully exercise all the code paths. https://gitlab.freedesktop.org/wayland/weston/issues/52 Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
2020-01-21 13:00:28 +03:00
/**
* Create a wl_buffer from a PNG file
*
* Loads the named PNG file from the directory of reference images,
* creates a wl_buffer with scale times the image dimensions in pixels,
* and copies the image content into the buffer using nearest-neighbor filter.
*
* \param client The client, for the Wayland connection.
* \param basename The PNG file name without .png suffix.
* \param scale Upscaling factor >= 1.
*/
struct buffer *
client_buffer_from_image_file(struct client *client,
const char *basename,
int scale)
{
struct buffer *buf;
char *fname;
pixman_image_t *img;
int buf_w, buf_h;
pixman_transform_t scaling;
assert(scale >= 1);
fname = image_filename(basename);
img = load_image_from_png(fname);
free(fname);
assert(img);
buf_w = scale * pixman_image_get_width(img);
buf_h = scale * pixman_image_get_height(img);
buf = create_shm_buffer_a8r8g8b8(client, buf_w, buf_h);
pixman_transform_init_scale(&scaling,
pixman_fixed_1 / scale,
pixman_fixed_1 / scale);
pixman_image_set_transform(img, &scaling);
pixman_image_set_filter(img, PIXMAN_FILTER_NEAREST, NULL, 0);
pixman_image_composite32(PIXMAN_OP_SRC,
img, /* src */
NULL, /* mask */
buf->image, /* dst */
0, 0, /* src x,y */
0, 0, /* mask x,y */
0, 0, /* dst x,y */
buf_w, buf_h);
pixman_image_unref(img);
return buf;
}
/**
* Bind to a singleton global in wl_registry
*
* \param client Client whose registry and globals to use.
* \param iface The Wayland interface to look for.
* \param version The version to bind the interface with.
* \return A struct wl_proxy, which you need to cast to the proper type.
*
* Asserts that the global being searched for is a singleton and is found.
*
* Binds with the exact version given, does not take compositor interface
* version into account.
*/
void *
bind_to_singleton_global(struct client *client,
const struct wl_interface *iface,
int version)
{
struct global *tmp;
struct global *g = NULL;
struct wl_proxy *proxy;
wl_list_for_each(tmp, &client->global_list, link) {
if (strcmp(tmp->interface, iface->name))
continue;
assert(!g && "multiple singleton objects");
g = tmp;
}
assert(g && "singleton not found");
proxy = wl_registry_bind(client->wl_registry, g->name, iface, version);
assert(proxy);
return proxy;
}
/**
* Create a wp_viewport for the client surface
*
* \param client The client->surface to use.
* \return A fresh viewport object.
*/
struct wp_viewport *
client_create_viewport(struct client *client)
{
struct wp_viewporter *viewporter;
struct wp_viewport *viewport;
viewporter = bind_to_singleton_global(client,
&wp_viewporter_interface, 1);
viewport = wp_viewporter_get_viewport(viewporter,
client->surface->wl_surface);
assert(viewport);
wp_viewporter_destroy(viewporter);
return viewport;
}
/**
* Fill the image with the given color
*
* \param image The image to write to.
* \param color The color to use.
*/
void
fill_image_with_color(pixman_image_t *image, const pixman_color_t *color)
{
pixman_image_t *solid;
int width;
int height;
width = pixman_image_get_width(image);
height = pixman_image_get_height(image);
solid = pixman_image_create_solid_fill(color);
pixman_image_composite32(PIXMAN_OP_SRC,
solid, /* src */
NULL, /* mask */
image, /* dst */
0, 0, /* src x,y */
0, 0, /* mask x,y */
0, 0, /* dst x,y */
width, height);
pixman_image_unref(solid);
}
/**
* Convert 8-bit RGB to opaque Pixman color
*
* \param tmp Pixman color struct to fill in.
* \param r Red value, 0 - 255.
* \param g Green value, 0 - 255.
* \param b Blue value, 0 - 255.
* \return tmp
*/
pixman_color_t *
color_rgb888(pixman_color_t *tmp, uint8_t r, uint8_t g, uint8_t b)
{
tmp->alpha = 65535;
tmp->red = (r << 8) + r;
tmp->green = (g << 8) + g;
tmp->blue = (b << 8) + b;
return tmp;
}