Merge pull request #2919 from bungoboingo/hack/fix-gl-cubemap-detection

[Fix] GL backend interpreting atlas texture as cube map vs 2d tex array when atlas layer depth == 6

wgpu/src/image/atlas.rs

@@ -19,6 +19,7 @@ use std::sync::Arc;
 
 #[derive(Debug)]
 pub struct Atlas {
+    backend: wgpu::Backend,
     texture: wgpu::Texture,
     texture_view: wgpu::TextureView,
     texture_bind_group: wgpu::BindGroup,
@@ -79,6 +80,7 @@ impl Atlas {
             });
 
         Atlas {
+            backend,
             texture,
             texture_view,
             texture_bind_group,
@@ -376,12 +378,22 @@ impl Atlas {
             return;
         }
 
+        // On the GL backend if layers.len() == 6 we need to help wgpu figure out that this texture
+        // is still a `GL_TEXTURE_2D_ARRAY` rather than `GL_TEXTURE_CUBE_MAP`. This will over-allocate
+        // some unused memory on GL, but it's better than not being able to grow the atlas past a depth
+        // of 6!
+        // https://github.com/gfx-rs/wgpu/blob/004e3efe84a320d9331371ed31fa50baa2414911/wgpu-hal/src/gles/mod.rs#L371
+        let depth_or_array_layers = match self.backend {
+            wgpu::Backend::Gl if self.layers.len() == 6 => 7,
+            _ => self.layers.len() as u32,
+        };
+
         let new_texture = device.create_texture(&wgpu::TextureDescriptor {
             label: Some("iced_wgpu::image texture atlas"),
             size: wgpu::Extent3d {
                 width: SIZE,
                 height: SIZE,
-                depth_or_array_layers: self.layers.len() as u32,
+                depth_or_array_layers,
             },
             mip_level_count: 1,
             sample_count: 1,