nvk: Call ensure_slm for nvk_cmd_dispatch_shader

Internal shaders can also use slm, so we need to allocate it correctly.

This fixes
dEQP-VK.dgc.ext.compute.misc.max_pc_range_256_full_preprocess_with_execution_set
with NAK_DEBUG=spill

Fixes: 105bdf2e36 ("nvk: Add a helper for dispatching compute shaders")
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35143>

src/nouveau/vulkan/nvk_cmd_dispatch.c

@@ -305,6 +305,8 @@ nvk_cmd_dispatch_shader(struct nvk_cmd_buffer *cmd,
                         uint32_t groupCountY,
                         uint32_t groupCountZ)
 {
+   struct nvk_device *dev = nvk_cmd_buffer_device(cmd);
+
    struct nvk_root_descriptor_table root = {
       .cs.group_count = {
          groupCountX,
@@ -324,6 +326,11 @@ nvk_cmd_dispatch_shader(struct nvk_cmd_buffer *cmd,
       return;
    }
 
+   if (shader != NULL) {
+      nvk_device_ensure_slm(dev, shader->info.slm_size,
+                                 shader->info.crs_size);
+   }
+
    struct nv_push *p = nvk_cmd_buffer_push(cmd, 8);
 
    /* Internal shaders don't want conditional rendering */