gallium/util: Fix leak in the live shader cache

When the nir backend is used, the create_shader
call is supposed to release state->ir.nir.
When the cache hits, create_shader is not called,
thus state->ir.nir should be freed.

There is nothing to be done for the TGSI case as the
tokens release is done by the caller.

This fixes a leak noticed in:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/2931

Fixes: 4bb919b0b8

Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4980>

src/gallium/auxiliary/util/u_live_shader_cache.c

@@ -129,8 +129,11 @@ util_live_shader_cache_get(struct pipe_context *ctx,
       *cache_hit = (shader != NULL);
 
    /* Return if the shader already exists. */
-   if (shader)
+   if (shader) {
+      if (state->type == PIPE_SHADER_IR_NIR)
+          ralloc_free(state->ir.nir);
       return shader;
+   }
 
    /* The cache mutex is unlocked to allow multiple create_shader
     * invocations to run simultaneously.