From b3e925a21ba693fd7423b45405b406bfcb8f088a Mon Sep 17 00:00:00 2001
From: Karol Herbst <kherbst@redhat.com>
Date: Sun, 4 Aug 2024 14:56:13 +0200
Subject: [PATCH] rusticl/kernel: properly respect device thread limits per
 dimension

Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30504>
---
 src/gallium/frontends/rusticl/core/kernel.rs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/gallium/frontends/rusticl/core/kernel.rs b/src/gallium/frontends/rusticl/core/kernel.rs
index 301a0bca289..1548ebc77be 100644
--- a/src/gallium/frontends/rusticl/core/kernel.rs
+++ b/src/gallium/frontends/rusticl/core/kernel.rs
@@ -1005,7 +1005,7 @@ impl Kernel {
         let total_threads = block.iter().take(work_dim).product::<usize>();
         if threads != 1 && total_threads < subgroups {
             for i in 0..work_dim {
-                if grid[i] * total_threads < threads {
+                if grid[i] * total_threads < threads && grid[i] * block[i] <= dim_threads[i] {
                     block[i] *= grid[i];
                     grid[i] = 1;
                     // can only do it once as nothing is cleanly divisible