river/river/render.zig
Isaac Freund 20d804cdb5 code: switch to custom wlroots/libwayland bindings
This is a big step up over @cImport() for ergonomics and type safety.
Nearly all void pointer casts have been eliminated!
2020-12-13 22:53:33 +01:00

363 lines
14 KiB
Zig

// This file is part of river, a dynamic tiling wayland compositor.
//
// Copyright 2020 The River Developers
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program. If not, see <https://www.gnu.org/licenses/>.
const build_options = @import("build_options");
const std = @import("std");
const os = std.os;
const wlr = @import("wlroots");
const wl = @import("wayland").server.wl;
const pixman = @import("pixman");
const log = @import("log.zig");
const util = @import("util.zig");
const Box = @import("Box.zig");
const LayerSurface = @import("LayerSurface.zig");
const Output = @import("Output.zig");
const Server = @import("Server.zig");
const View = @import("View.zig");
const ViewStack = @import("view_stack.zig").ViewStack;
const SurfaceRenderData = struct {
output: *const Output,
/// In output layout coordinates relative to the output
output_x: i32,
output_y: i32,
when: *os.timespec,
opacity: f32,
};
pub fn renderOutput(output: *Output) void {
const config = &output.root.server.config;
const renderer = output.wlr_output.backend.getRenderer().?;
var now: os.timespec = undefined;
os.clock_gettime(os.CLOCK_MONOTONIC, &now) catch unreachable;
output.wlr_output.attachRender(null) catch return;
renderer.begin(output.wlr_output.width, output.wlr_output.height);
// Find the first visible fullscreen view in the stack if there is one
var it = ViewStack(View).iter(output.views.first, .forward, output.current.tags, renderFilter);
const fullscreen_view = while (it.next()) |view| {
if (view.current.fullscreen) break view;
} else null;
// If we have a fullscreen view to render, render it.
if (fullscreen_view) |view| {
// Always clear with solid black for fullscreen
renderer.clear(&[_]f32{ 0, 0, 0, 1 });
renderView(output.*, view, &now);
if (build_options.xwayland) renderXwaylandUnmanaged(output.*, &now);
} else {
// No fullscreen view, so render normal layers/views
renderer.clear(&config.background_color);
renderLayer(output.*, output.getLayer(.background).*, &now, .toplevels);
renderLayer(output.*, output.getLayer(.bottom).*, &now, .toplevels);
// The first view in the list is "on top" so iterate in reverse.
it = ViewStack(View).iter(output.views.last, .reverse, output.current.tags, renderFilter);
while (it.next()) |view| {
// Focused views are rendered on top of normal views, skip them for now
if (view.current.focus != 0) continue;
renderView(output.*, view, &now);
if (view.draw_borders) renderBorders(output.*, view, &now);
}
// Render focused views
it = ViewStack(View).iter(output.views.last, .reverse, output.current.tags, renderFilter);
while (it.next()) |view| {
// Skip unfocused views since we already rendered them
if (view.current.focus == 0) continue;
renderView(output.*, view, &now);
if (view.draw_borders) renderBorders(output.*, view, &now);
}
if (build_options.xwayland) renderXwaylandUnmanaged(output.*, &now);
renderLayer(output.*, output.getLayer(.top).*, &now, .toplevels);
renderLayer(output.*, output.getLayer(.background).*, &now, .popups);
renderLayer(output.*, output.getLayer(.bottom).*, &now, .popups);
renderLayer(output.*, output.getLayer(.top).*, &now, .popups);
}
// The overlay layer is rendered in both fullscreen and normal cases
renderLayer(output.*, output.getLayer(.overlay).*, &now, .toplevels);
renderLayer(output.*, output.getLayer(.overlay).*, &now, .popups);
renderDragIcons(output.*, &now);
// Hardware cursors are rendered by the GPU on a separate plane, and can be
// moved around without re-rendering what's beneath them - which is more
// efficient. However, not all hardware supports hardware cursors. For this
// reason, wlroots provides a software fallback, which we ask it to render
// here. wlr_cursor handles configuring hardware vs software cursors for you,
// and this function is a no-op when hardware cursors are in use.
output.wlr_output.renderSoftwareCursors(null);
// Conclude rendering and swap the buffers, showing the final frame
// on-screen.
renderer.end();
// TODO(wlroots): remove this with the next release. It is here due to
// a wlroots bug in the screencopy damage implementation
{
var w: c_int = undefined;
var h: c_int = undefined;
output.wlr_output.transformedResolution(&w, &h);
var damage: pixman.Region32 = undefined;
damage.init();
_ = damage.unionRect(&damage, 0, 0, @intCast(c_uint, w), @intCast(c_uint, h));
output.wlr_output.setDamage(&damage);
}
// TODO: handle failure
output.wlr_output.commit() catch
log.err(.render, "output commit failed for {}", .{output.wlr_output.name});
}
fn renderFilter(view: *View, filter_tags: u32) bool {
// This check prevents a race condition when a frame is requested
// between mapping of a view and the first configure being handled.
if (view.current.box.width == 0 or view.current.box.height == 0)
return false;
return view.current.tags & filter_tags != 0;
}
/// Render all surfaces on the passed layer
fn renderLayer(
output: Output,
layer: std.TailQueue(LayerSurface),
now: *os.timespec,
role: enum { toplevels, popups },
) void {
var it = layer.first;
while (it) |node| : (it = node.next) {
const layer_surface = &node.data;
var rdata = SurfaceRenderData{
.output = &output,
.output_x = layer_surface.box.x,
.output_y = layer_surface.box.y,
.when = now,
.opacity = 1.0,
};
switch (role) {
.toplevels => layer_surface.wlr_layer_surface.surface.forEachSurface(
*SurfaceRenderData,
renderSurfaceIterator,
&rdata,
),
.popups => layer_surface.wlr_layer_surface.forEachPopup(
*SurfaceRenderData,
renderSurfaceIterator,
&rdata,
),
}
}
}
fn renderView(output: Output, view: *View, now: *os.timespec) void {
// If we have saved buffers, we are in the middle of a transaction
// and need to render those buffers until the transaction is complete.
if (view.saved_buffers.items.len != 0) {
for (view.saved_buffers.items) |saved_buffer|
renderTexture(
output,
saved_buffer.client_buffer.texture orelse continue,
.{
.x = saved_buffer.box.x + view.current.box.x - view.saved_surface_box.x,
.y = saved_buffer.box.y + view.current.box.y - view.saved_surface_box.y,
.width = @intCast(c_int, saved_buffer.box.width),
.height = @intCast(c_int, saved_buffer.box.height),
},
saved_buffer.transform,
view.opacity,
);
} else {
// Since there is no stashed buffer, we are not in the middle of
// a transaction and may simply render each toplevel surface.
var rdata = SurfaceRenderData{
.output = &output,
.output_x = view.current.box.x - view.surface_box.x,
.output_y = view.current.box.y - view.surface_box.y,
.when = now,
.opacity = view.opacity,
};
view.forEachSurface(*SurfaceRenderData, renderSurfaceIterator, &rdata);
}
}
fn renderDragIcons(output: Output, now: *os.timespec) void {
const output_box = output.root.output_layout.getBox(output.wlr_output).?;
var it = output.root.drag_icons.first;
while (it) |node| : (it = node.next) {
const drag_icon = &node.data;
var rdata = SurfaceRenderData{
.output = &output,
.output_x = @floatToInt(i32, drag_icon.seat.cursor.wlr_cursor.x) +
drag_icon.wlr_drag_icon.surface.sx - output_box.x,
.output_y = @floatToInt(i32, drag_icon.seat.cursor.wlr_cursor.y) +
drag_icon.wlr_drag_icon.surface.sy - output_box.y,
.when = now,
.opacity = 1.0,
};
drag_icon.wlr_drag_icon.surface.forEachSurface(*SurfaceRenderData, renderSurfaceIterator, &rdata);
}
}
/// Render all xwayland unmanaged windows that appear on the output
fn renderXwaylandUnmanaged(output: Output, now: *os.timespec) void {
const output_box = output.root.output_layout.getBox(output.wlr_output).?;
var it = output.root.xwayland_unmanaged_views.first;
while (it) |node| : (it = node.next) {
const xwayland_surface = node.data.xwayland_surface;
var rdata = SurfaceRenderData{
.output = &output,
.output_x = xwayland_surface.x - output_box.x,
.output_y = xwayland_surface.y - output_box.y,
.when = now,
.opacity = 1.0,
};
xwayland_surface.surface.?.forEachSurface(*SurfaceRenderData, renderSurfaceIterator, &rdata);
}
}
/// This function is passed to wlroots to render each surface during iteration
fn renderSurfaceIterator(
surface: *wlr.Surface,
surface_x: c_int,
surface_y: c_int,
rdata: *SurfaceRenderData,
) callconv(.C) void {
renderTexture(
rdata.output.*,
surface.getTexture() orelse return,
.{
.x = rdata.output_x + surface_x,
.y = rdata.output_y + surface_y,
.width = surface.current.width,
.height = surface.current.height,
},
surface.current.transform,
rdata.opacity,
);
surface.sendFrameDone(rdata.when);
}
/// Render the given texture at the given box, taking the scale and transform
/// of the output into account.
fn renderTexture(
output: Output,
texture: *wlr.Texture,
wlr_box: wlr.Box,
transform: wl.Output.Transform,
opacity: f32,
) void {
var box = wlr_box;
// Scale the box to the output's current scaling factor
scaleBox(&box, output.wlr_output.scale);
// wlr_matrix_project_box is a helper which takes a box with a desired
// x, y coordinates, width and height, and an output geometry, then
// prepares an orthographic projection and multiplies the necessary
// transforms to produce a model-view-projection matrix.
var matrix: [9]f32 = undefined;
const inverted = wlr.Output.transformInvert(transform);
wlr.matrix.projectBox(&matrix, &box, inverted, 0.0, &output.wlr_output.transform_matrix);
// This takes our matrix, the texture, and an alpha, and performs the actual
// rendering on the GPU.
const renderer = output.wlr_output.backend.getRenderer().?;
renderer.renderTextureWithMatrix(texture, &matrix, opacity) catch return;
}
fn renderBorders(output: Output, view: *View, now: *os.timespec) void {
const config = &output.root.server.config;
const color = if (view.current.focus != 0) &config.border_color_focused else &config.border_color_unfocused;
const border_width = config.border_width;
const actual_box = if (view.saved_buffers.items.len != 0) view.saved_surface_box else view.surface_box;
var border: Box = undefined;
// left and right, covering the corners as well
border.y = view.current.box.y - @intCast(i32, border_width);
border.width = border_width;
border.height = actual_box.height + border_width * 2;
// left
border.x = view.current.box.x - @intCast(i32, border_width);
renderRect(output, border, color);
// right
border.x = view.current.box.x + @intCast(i32, actual_box.width);
renderRect(output, border, color);
// top and bottom
border.x = view.current.box.x;
border.width = actual_box.width;
border.height = border_width;
// top
border.y = view.current.box.y - @intCast(i32, border_width);
renderRect(output, border, color);
// bottom border
border.y = view.current.box.y + @intCast(i32, actual_box.height);
renderRect(output, border, color);
}
fn renderRect(output: Output, box: Box, color: *const [4]f32) void {
var wlr_box = box.toWlrBox();
scaleBox(&wlr_box, output.wlr_output.scale);
output.wlr_output.backend.getRenderer().?.renderRect(
&wlr_box,
color,
&output.wlr_output.transform_matrix,
);
}
/// Scale a wlr_box, taking the possibility of fractional scaling into account.
fn scaleBox(box: *wlr.Box, scale: f64) void {
box.x = @floatToInt(c_int, @round(@intToFloat(f64, box.x) * scale));
box.y = @floatToInt(c_int, @round(@intToFloat(f64, box.y) * scale));
box.width = scaleLength(box.width, box.x, scale);
box.height = scaleLength(box.height, box.x, scale);
}
/// Scales a width/height.
///
/// This might seem overly complex, but it needs to work for fractional scaling.
fn scaleLength(length: c_int, offset: c_int, scale: f64) c_int {
return @floatToInt(c_int, @round(@intToFloat(f64, offset + length) * scale) -
@round(@intToFloat(f64, offset) * scale));
}