swayfx/sway/desktop/transaction.c
Ryan Dwyer 96c8c02483 Fix flash of background when xwayland views are mapped
A flash of background was happening for two reasons:

1) We were using the xsurface's dimensions to check if the surface is
ready, but these are pending dimensions.
2) In my particular setup, the default geometry of the xsurface does not
intersect any output, which prevented it from receiving a frame done
event. This made the transaction time out and the client would only
redraw once it's been rendered.
2018-06-30 14:30:14 +10:00

416 lines
14 KiB
C

#define _POSIX_C_SOURCE 200809L
#include <stdbool.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <wlr/types/wlr_buffer.h>
#include <wlr/types/wlr_linux_dmabuf.h>
#include "sway/debug.h"
#include "sway/desktop/transaction.h"
#include "sway/output.h"
#include "sway/tree/container.h"
#include "sway/tree/view.h"
#include "sway/tree/workspace.h"
#include "list.h"
#include "log.h"
/**
* How long we should wait for views to respond to the configure before giving
* up and applying the transaction anyway.
*/
#define TIMEOUT_MS 200
/**
* If enabled, sway will always wait for the transaction timeout before
* applying it, rather than applying it when the views are ready. This allows us
* to observe the rendered state while a transaction is in progress.
*/
#define TRANSACTION_DEBUG false
struct sway_transaction {
struct wl_event_source *timer;
list_t *instructions; // struct sway_transaction_instruction *
size_t num_waiting;
size_t num_configures;
struct timespec create_time;
struct timespec commit_time;
};
struct sway_transaction_instruction {
struct sway_transaction *transaction;
struct sway_container *container;
struct sway_container_state state;
struct wlr_buffer *saved_buffer;
int saved_buffer_width, saved_buffer_height;
uint32_t serial;
bool ready;
};
struct sway_transaction *transaction_create() {
struct sway_transaction *transaction =
calloc(1, sizeof(struct sway_transaction));
transaction->instructions = create_list();
if (server.debug_txn_timings) {
clock_gettime(CLOCK_MONOTONIC, &transaction->create_time);
}
return transaction;
}
static void remove_saved_view_buffer(
struct sway_transaction_instruction *instruction) {
if (instruction->saved_buffer) {
wlr_buffer_unref(instruction->saved_buffer);
instruction->saved_buffer = NULL;
}
}
static void save_view_buffer(struct sway_view *view,
struct sway_transaction_instruction *instruction) {
if (!sway_assert(instruction->saved_buffer == NULL,
"Didn't expect instruction to have a saved buffer already")) {
remove_saved_view_buffer(instruction);
}
if (view->surface && wlr_surface_has_buffer(view->surface)) {
instruction->saved_buffer = wlr_buffer_ref(view->surface->buffer);
instruction->saved_buffer_width = view->surface->current->width;
instruction->saved_buffer_height = view->surface->current->height;
}
}
static void transaction_destroy(struct sway_transaction *transaction) {
// Free instructions
for (int i = 0; i < transaction->instructions->length; ++i) {
struct sway_transaction_instruction *instruction =
transaction->instructions->items[i];
struct sway_container *con = instruction->container;
for (int j = 0; j < con->instructions->length; ++j) {
if (con->instructions->items[j] == instruction) {
list_del(con->instructions, j);
break;
}
}
if (con->destroying && !con->instructions->length) {
container_free(con);
}
remove_saved_view_buffer(instruction);
free(instruction);
}
list_free(transaction->instructions);
if (transaction->timer) {
wl_event_source_remove(transaction->timer);
}
free(transaction);
}
static void copy_pending_state(struct sway_container *container,
struct sway_container_state *state) {
state->layout = container->layout;
state->swayc_x = container->x;
state->swayc_y = container->y;
state->swayc_width = container->width;
state->swayc_height = container->height;
state->has_gaps = container->has_gaps;
state->current_gaps = container->current_gaps;
state->gaps_inner = container->gaps_inner;
state->gaps_outer = container->gaps_outer;
state->parent = container->parent;
if (container->type == C_VIEW) {
struct sway_view *view = container->sway_view;
state->view_x = view->x;
state->view_y = view->y;
state->view_width = view->width;
state->view_height = view->height;
state->is_fullscreen = view->is_fullscreen;
state->border = view->border;
state->border_thickness = view->border_thickness;
state->border_top = view->border_top;
state->border_left = view->border_left;
state->border_right = view->border_right;
state->border_bottom = view->border_bottom;
} else if (container->type == C_WORKSPACE) {
state->ws_fullscreen = container->sway_workspace->fullscreen;
state->ws_floating = container->sway_workspace->floating;
state->children = create_list();
list_cat(state->children, container->children);
} else {
state->children = create_list();
list_cat(state->children, container->children);
}
}
static bool transaction_has_container(struct sway_transaction *transaction,
struct sway_container *container) {
for (int i = 0; i < transaction->instructions->length; ++i) {
struct sway_transaction_instruction *instruction =
transaction->instructions->items[i];
if (instruction->container == container) {
return true;
}
}
return false;
}
void transaction_add_container(struct sway_transaction *transaction,
struct sway_container *container) {
if (transaction_has_container(transaction, container)) {
return;
}
struct sway_transaction_instruction *instruction =
calloc(1, sizeof(struct sway_transaction_instruction));
instruction->transaction = transaction;
instruction->container = container;
copy_pending_state(container, &instruction->state);
if (container->type == C_VIEW) {
save_view_buffer(container->sway_view, instruction);
}
list_add(transaction->instructions, instruction);
}
/**
* Apply a transaction to the "current" state of the tree.
*/
static void transaction_apply(struct sway_transaction *transaction) {
wlr_log(L_DEBUG, "Applying transaction %p", transaction);
if (server.debug_txn_timings) {
struct timespec now;
clock_gettime(CLOCK_MONOTONIC, &now);
struct timespec *create = &transaction->create_time;
struct timespec *commit = &transaction->commit_time;
float ms_arranging = (commit->tv_sec - create->tv_sec) * 1000 +
(commit->tv_nsec - create->tv_nsec) / 1000000.0;
float ms_waiting = (now.tv_sec - commit->tv_sec) * 1000 +
(now.tv_nsec - commit->tv_nsec) / 1000000.0;
float ms_total = ms_arranging + ms_waiting;
wlr_log(L_DEBUG, "Transaction %p: %.1fms arranging, %.1fms waiting, "
"%.1fms total (%.1f frames if 60Hz)", transaction,
ms_arranging, ms_waiting, ms_total, ms_total / (1000 / 60));
}
// Apply the instruction state to the container's current state
for (int i = 0; i < transaction->instructions->length; ++i) {
struct sway_transaction_instruction *instruction =
transaction->instructions->items[i];
struct sway_container *container = instruction->container;
// Damage the old and new locations
struct wlr_box old_box = {
.x = container->current.swayc_x,
.y = container->current.swayc_y,
.width = container->current.swayc_width,
.height = container->current.swayc_height,
};
struct wlr_box new_box = {
.x = instruction->state.swayc_x,
.y = instruction->state.swayc_y,
.width = instruction->state.swayc_width,
.height = instruction->state.swayc_height,
};
for (int j = 0; j < root_container.children->length; ++j) {
struct sway_container *output = root_container.children->items[j];
output_damage_box(output->sway_output, &old_box);
output_damage_box(output->sway_output, &new_box);
}
// There are separate children lists for each instruction state, the
// container's current state and the container's pending state
// (ie. con->children). The list itself needs to be freed here.
// Any child containers which are being deleted will be cleaned up in
// transaction_destroy().
list_free(container->current.children);
memcpy(&container->current, &instruction->state,
sizeof(struct sway_container_state));
}
}
/**
* For simplicity, we only progress the queue if it can be completely flushed.
*/
static void transaction_progress_queue() {
// We iterate this list in reverse because we're more likely to find a
// waiting transactions at the end of the list.
for (int i = server.transactions->length - 1; i >= 0; --i) {
struct sway_transaction *transaction = server.transactions->items[i];
if (transaction->num_waiting) {
return;
}
}
for (int i = 0; i < server.transactions->length; ++i) {
struct sway_transaction *transaction = server.transactions->items[i];
transaction_apply(transaction);
transaction_destroy(transaction);
}
server.transactions->length = 0;
}
static int handle_timeout(void *data) {
struct sway_transaction *transaction = data;
wlr_log(L_DEBUG, "Transaction %p timed out (%li waiting)",
transaction, transaction->num_waiting);
transaction->num_waiting = 0;
transaction_progress_queue();
return 0;
}
static bool should_configure(struct sway_container *con,
struct sway_transaction_instruction *instruction) {
if (con->type != C_VIEW) {
return false;
}
if (con->destroying) {
return false;
}
// The settled dimensions are what size the view will be once any pending
// configures have applied (excluding the one we might be configuring now).
// If these match the dimensions that this transaction wants then we don't
// need to configure it.
int settled_width = con->current.view_width;
int settled_height = con->current.view_height;
if (con->instructions->length) {
struct sway_transaction_instruction *last_instruction =
con->instructions->items[con->instructions->length - 1];
settled_width = last_instruction->state.view_width;
settled_height = last_instruction->state.view_height;
}
if (settled_width == instruction->state.view_width &&
settled_height == instruction->state.view_height) {
return false;
}
return true;
}
void transaction_commit(struct sway_transaction *transaction) {
wlr_log(L_DEBUG, "Transaction %p committing with %i instructions",
transaction, transaction->instructions->length);
transaction->num_waiting = 0;
for (int i = 0; i < transaction->instructions->length; ++i) {
struct sway_transaction_instruction *instruction =
transaction->instructions->items[i];
struct sway_container *con = instruction->container;
if (should_configure(con, instruction)) {
instruction->serial = view_configure(con->sway_view,
instruction->state.view_x,
instruction->state.view_y,
instruction->state.view_width,
instruction->state.view_height);
++transaction->num_waiting;
// From here on we are rendering a saved buffer of the view, which
// means we can send a frame done event to make the client redraw it
// as soon as possible. Additionally, this is required if a view is
// mapping and its default geometry doesn't intersect an output.
struct timespec when;
wlr_surface_send_frame_done(con->sway_view->surface, &when);
}
list_add(con->instructions, instruction);
}
transaction->num_configures = transaction->num_waiting;
if (server.debug_txn_timings) {
clock_gettime(CLOCK_MONOTONIC, &transaction->commit_time);
}
if (server.transactions->length || transaction->num_waiting) {
list_add(server.transactions, transaction);
} else {
// There are no other transactions in progress, and this one has nothing
// to wait for, so we can skip the queue.
wlr_log(L_DEBUG, "Transaction %p has nothing to wait for", transaction);
transaction_apply(transaction);
transaction_destroy(transaction);
return;
}
if (transaction->num_waiting) {
// Set up a timer which the views must respond within
transaction->timer = wl_event_loop_add_timer(server.wl_event_loop,
handle_timeout, transaction);
wl_event_source_timer_update(transaction->timer, TIMEOUT_MS);
}
// The debug tree shows the pending/live tree. Here is a good place to
// update it, because we make a transaction every time we change the pending
// tree.
update_debug_tree();
}
static void set_instruction_ready(
struct sway_transaction_instruction *instruction) {
instruction->ready = true;
struct sway_transaction *transaction = instruction->transaction;
if (server.debug_txn_timings) {
struct timespec now;
clock_gettime(CLOCK_MONOTONIC, &now);
struct timespec *start = &transaction->commit_time;
float ms = (now.tv_sec - start->tv_sec) * 1000 +
(now.tv_nsec - start->tv_nsec) / 1000000.0;
wlr_log(L_DEBUG, "Transaction %p: %li/%li ready in %.1fms (%s)",
transaction,
transaction->num_configures - transaction->num_waiting + 1,
transaction->num_configures, ms,
instruction->container->name);
}
// If all views are ready, apply the transaction.
// If the transaction has timed out then its num_waiting will be 0 already.
if (transaction->num_waiting > 0 && --transaction->num_waiting == 0) {
#if !TRANSACTION_DEBUG
wlr_log(L_DEBUG, "Transaction %p is ready", transaction);
wl_event_source_timer_update(transaction->timer, 0);
transaction_progress_queue();
#endif
}
}
/**
* Mark all of the view's instructions as ready up to and including the
* instruction at the given index. This allows the view to skip a configure.
*/
static void set_instructions_ready(struct sway_view *view, int index) {
for (int i = 0; i <= index; ++i) {
struct sway_transaction_instruction *instruction =
view->swayc->instructions->items[i];
set_instruction_ready(instruction);
}
}
void transaction_notify_view_ready(struct sway_view *view, uint32_t serial) {
for (int i = 0; i < view->swayc->instructions->length; ++i) {
struct sway_transaction_instruction *instruction =
view->swayc->instructions->items[i];
if (instruction->serial == serial && !instruction->ready) {
set_instructions_ready(view, i);
return;
}
}
}
void transaction_notify_view_ready_by_size(struct sway_view *view,
int width, int height) {
for (int i = 0; i < view->swayc->instructions->length; ++i) {
struct sway_transaction_instruction *instruction =
view->swayc->instructions->items[i];
if (!instruction->ready && instruction->state.view_width == width &&
instruction->state.view_height == height) {
set_instructions_ready(view, i);
return;
}
}
}
struct wlr_texture *transaction_get_saved_texture(struct sway_view *view,
int *width, int *height) {
struct sway_transaction_instruction *instruction =
view->swayc->instructions->items[0];
if (!instruction->saved_buffer) {
return NULL;
}
*width = instruction->saved_buffer_width;
*height = instruction->saved_buffer_height;
return instruction->saved_buffer->texture;
}