diff options
| author | Kenneth Graunke <[email protected]> | 2019-02-07 08:41:29 -0800 |
|---|---|---|
| committer | Kenneth Graunke <[email protected]> | 2019-02-21 10:26:11 -0800 |
| commit | 18e31a9b311a6c1b35eedfae48d7512eea5be68d (patch) | |
| tree | f557d45cd0a3d8686fdc8d46514880fd2093ac91 /src/gallium/drivers | |
| parent | 3b1ac8244ec5aecf703110f8191058f3299b72af (diff) | |
iris: Fix accidental busy-looping in query waits
When switching from bo_wait to sync-points, I missed that we turned an
if (not landed) bo_wait into a while (not landed) check_syncpt(), which
has a timeout of 0. This meant, rather than sleeping until the batch
is complete, we'd busy-loop, continually asking the kernel "is the batch
done yet???". This is not what we want at all - if we wanted a busy
loop, we'd just loop on !snapshots_landed. We want to sleep.
Add an effectively infinite timeout so that we sleep.
Diffstat (limited to 'src/gallium/drivers')
| -rw-r--r-- | src/gallium/drivers/iris/iris_query.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/gallium/drivers/iris/iris_query.c b/src/gallium/drivers/iris/iris_query.c index 6e8062f3b4e..c5130bda36a 100644 --- a/src/gallium/drivers/iris/iris_query.c +++ b/src/gallium/drivers/iris/iris_query.c @@ -876,7 +876,7 @@ iris_get_query_result(struct pipe_context *ctx, while (!q->map->snapshots_landed) { if (wait) - iris_wait_syncpt(ctx->screen, q->syncpt, 0); + iris_wait_syncpt(ctx->screen, q->syncpt, INT64_MAX); else return false; } |