gpu/shaders: ensure dynamically uniform barriers when malloc fails

GPU APIs require that barrier() calls are dynamically uniform, that is
for every barrier in the code, every shader invocation in a workgroup
must all call it, or all not call it.

Signed-off-by: Elias Naur <mail@eliasnaur.com>
This commit is contained in:
Elias Naur
2021-04-11 16:34:33 +02:00
parent f930e3f3d7
commit ee519351f7
9 changed files with 564 additions and 580 deletions
+4 -7
View File
@@ -83,10 +83,6 @@ vec4[CHUNK] fillImage(uvec2 xy, CmdImage cmd_img) {
}
void main() {
if (mem_error != NO_ERROR) {
return;
}
uint tile_ix = gl_WorkGroupID.y * conf.width_in_tiles + gl_WorkGroupID.x;
Alloc cmd_alloc = slice_mem(conf.ptcl_alloc, tile_ix * PTCL_INITIAL_ALLOC, PTCL_INITIAL_ALLOC);
CmdRef cmd_ref = CmdRef(cmd_alloc.offset);
@@ -109,7 +105,8 @@ void main() {
float area[CHUNK];
uint clip_depth = 0;
while (true) {
bool mem_ok = mem_error == NO_ERROR;
while (mem_ok) {
uint tag = Cmd_tag(cmd_alloc, cmd_ref).tag;
if (tag == Cmd_End) {
break;
@@ -122,7 +119,7 @@ void main() {
for (uint k = 0; k < CHUNK; k++) df[k] = 1e9;
TileSegRef tile_seg_ref = TileSegRef(stroke.tile_ref);
do {
TileSeg seg = TileSeg_read(new_alloc(tile_seg_ref.offset, TileSeg_size), tile_seg_ref);
TileSeg seg = TileSeg_read(new_alloc(tile_seg_ref.offset, TileSeg_size, mem_ok), tile_seg_ref);
vec2 line_vec = seg.vector;
for (uint k = 0; k < CHUNK; k++) {
vec2 dpos = xy + vec2(0.5, 0.5) - seg.origin;
@@ -143,7 +140,7 @@ void main() {
tile_seg_ref = TileSegRef(fill.tile_ref);
// Calculate coverage based on backdrop + coverage of each line segment
do {
TileSeg seg = TileSeg_read(new_alloc(tile_seg_ref.offset, TileSeg_size), tile_seg_ref);
TileSeg seg = TileSeg_read(new_alloc(tile_seg_ref.offset, TileSeg_size, mem_ok), tile_seg_ref);
for (uint k = 0; k < CHUNK; k++) {
vec2 my_xy = xy + vec2(chunk_offset(k));
vec2 start = seg.origin - my_xy;