diff --git a/Cargo.toml b/Cargo.toml
index 1fa0b42..4dd158e 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -4,6 +4,7 @@ version = "0.1.0"
 edition = "2021"
 
 [dependencies]
+bytemuck = { version = "^1.0", features = ["derive"] }
 multimap = "0.8"
 pollster = "0.2"
 rayon = "1"
diff --git a/src/lib.rs b/src/lib.rs
index 25071fd..887cd70 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -36,11 +36,11 @@ impl Renderer {
     }
 
     pub fn add_pass<T: 'static + RenderPass>(&mut self, pass: T) {
-        let pass = Arc::new(RwLock::new(pass));
+        let pass = Arc::new(pass);
         self.add_pass_arc(pass);
     }
 
-    pub fn add_pass_arc<T: 'static + RenderPass>(&mut self, pass: Arc<RwLock<T>>) {
+    pub fn add_pass_arc<T: 'static + RenderPass>(&mut self, pass: Arc<T>) {
         let pass = RenderPassBox::new(pass, self.frames_in_flight);
         self.add_pass_box(pass);
     }
@@ -99,32 +99,26 @@ impl Renderer {
             });
         }
 
+        let opaque_cmds = Mutex::new(Vec::new());
+        if let Some(opaque) = phase_passes.get_vec(&Phase::Opaque) {
+            opaque.par_iter().for_each(|pass_index| {
+                let pass = &self.render_passes[*pass_index];
+
+                let frame_data = IndexedPhaseData {
+                    phase: Phase::Opaque,
+                    frame_data: frame_index,
+                    viewport: &viewport,
+                };
+
+                if let Some(cmd) = pass.record_render(frame_data) {
+                    opaque_cmds.lock().unwrap().push(cmd);
+                }
+            })
+        }
+
+        let opaque_cmds = opaque_cmds.into_inner().unwrap();
+
         {
-            let mut opaque_cmds = Vec::new();
-            if let Some(opaque) = phase_passes.get_vec(&Phase::Opaque) {
-                opaque.iter().for_each(|pass_index| {
-                    let frame_data = IndexedPhaseData {
-                        phase: Phase::Opaque,
-                        frame_data: frame_index,
-                        viewport: &viewport,
-                    };
-
-                    let pass = &self.render_passes[*pass_index];
-                    let mut cmds = self.device.create_render_bundle_encoder(
-                        &wgpu::RenderBundleEncoderDescriptor {
-                            label: Some("Opaque Pass Render Bundle"),
-                            color_formats: &[format],
-                            depth_stencil: None,
-                            sample_count: 1,
-                            multiview: None,
-                        },
-                    );
-
-                    pass.record_render(frame_data, &mut cmds);
-                    opaque_cmds.push(cmds.finish(&wgpu::RenderBundleDescriptor::default()));
-                })
-            }
-
             let mut rp = encoder.begin_render_pass(&wgpu::RenderPassDescriptor {
                 label: Some("Render Pass"),
                 color_attachments: &[wgpu::RenderPassColorAttachment {
diff --git a/src/main.rs b/src/main.rs
index 7b8113a..2071069 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -73,7 +73,7 @@ fn main() {
     let (mut renderer, mut viewport) = pollster::block_on(make_window_renderer(&window));
 
     let device = renderer.get_device();
-    let mesh_pass = pass::mesh::MeshPass::new(device.to_owned());
+    let mesh_pass = pass::mesh::MeshPass::new(device.to_owned(), viewport.config.format);
     renderer.add_pass(mesh_pass);
 
     event_loop.run(move |event, _, control_flow| match event {
diff --git a/src/mesh.rs b/src/mesh.rs
index cef1596..e22e5fb 100644
--- a/src/mesh.rs
+++ b/src/mesh.rs
@@ -133,10 +133,10 @@ pub struct AttrAllocKey {
 /// Info about an array of attributes that has been allocated in an [AttrPool].
 #[derive(Copy, Clone, Debug, Hash, PartialEq, Eq)]
 pub struct AttrAlloc {
-    offset: usize,
-    count: usize,
-    offset_bytes: usize,
-    count_bytes: usize,
+    pub offset: usize,
+    pub count: usize,
+    pub offset_bytes: usize,
+    pub count_bytes: usize,
 }
 
 /// An unused space range in an [AttrPool].
@@ -305,17 +305,21 @@ impl AttrPool {
         Ok(())
     }
 
-    /// Gets the offset (in elements) of an allocation, passed by key.
-    pub fn get_offset(&self, key: usize) -> Result<usize, PoolError> {
-        let alloc = self.allocs.get(key).ok_or(PoolError::InvalidIndex)?;
-        Ok(alloc.offset)
+    /// Retrieves an [AttrAlloc] by key.
+    pub fn get(&self, key: usize) -> Option<AttrAlloc> {
+        self.allocs.get(key).copied()
+    }
+
+    /// Gets this pool's internal GPU buffer.
+    pub fn get_buffer(&self) -> &wgpu::Buffer {
+        &self.buffer
     }
 
     /// Gets a [CopyDest] for an allocation, by key.
     pub fn get_copy_dest(&self, key: usize) -> Result<CopyDest, PoolError> {
-        let offset = self.get_offset(key)?;
+        let offset = self.get(key).ok_or(PoolError::InvalidIndex)?.offset_bytes;
         Ok(CopyDest {
-            buffer: &self.buffer,
+            buffer: self.get_buffer(),
             offset,
         })
     }
@@ -347,17 +351,31 @@ pub type MeshLayoutDesc = smallmap::Set<AttrId>;
 #[derive(Copy, Clone, Debug, Hash, PartialEq, Eq)]
 pub struct MeshLayoutId(usize);
 
+/// Mappings of the attributes in a [MeshLayout] to specific pool indices.
+pub type MeshLayoutBindingIndices = smallmap::Map<AttrId, usize>;
+
 /// Mappings of the attributes in a [MeshLayout] to specific pools.
-pub type MeshLayoutBindings = smallmap::Map<AttrId, usize>;
+pub struct MeshLayoutBindings<'a> {
+    lock: std::sync::RwLockReadGuard<'a, HashMap<AttrId, Vec<AttrPool>>>,
+    indices: MeshLayoutBindingIndices,
+}
+
+impl<'a> MeshLayoutBindings<'a> {
+    pub fn get(&self, attr: AttrId) -> Option<&AttrPool> {
+        let pool_id = self.indices.get(&attr)?;
+        let pools = self.lock.get(&attr)?;
+        pools.get(*pool_id)
+    }
+}
 
 /// Mappings of the attributes in a [MeshAlloc] to specific pool offsets.
-pub type MeshAllocOffsets = SmallVec<[(AttrId, usize); MAX_MESH_INLINE_ATTRIBUTES]>;
+pub type MeshAllocInfos = SmallVec<[(AttrId, AttrAlloc); MAX_MESH_INLINE_ATTRIBUTES]>;
 
 /// A set of mesh instances fitting a common [MeshLayoutDesc].
 #[derive(Default)]
 pub struct MeshLayoutInstances<T> {
-    pub bindings: MeshLayoutBindings,
-    pub instances: Vec<(T, MeshAllocOffsets)>,
+    pub bindings: MeshLayoutBindingIndices,
+    pub instances: Vec<(T, MeshAllocInfos)>,
 }
 
 /// A mesh data pool.
@@ -476,7 +494,7 @@ impl MeshPool {
     ) -> Result<Vec<MeshLayoutInstances<T>>, PoolError>
     where
         I: Iterator<Item = T>,
-        F: Fn(&T) -> &'static MeshHandle,
+        F: Fn(&T) -> &MeshHandle,
     {
         let layout = self
             .mesh_layouts
@@ -511,8 +529,8 @@ impl MeshPool {
                 continue;
             }
 
-            let mut layout_bindings = MeshLayoutBindings::default();
-            let mut alloc_offsets = MeshAllocOffsets::default();
+            let mut layout_bindings = MeshLayoutBindingIndices::default();
+            let mut alloc_infos = MeshAllocInfos::default();
             for alloc in attr_allocs.iter() {
                 let pools_read = self.pools.read().unwrap();
                 let pools = pools_read
@@ -520,13 +538,13 @@ impl MeshPool {
                     .ok_or(PoolError::AttrUnregistered)?;
 
                 let pool = pools.get(alloc.pool).ok_or(PoolError::InvalidIndex)?;
-                let alloc_offset = pool.get_offset(alloc.alloc)?;
+                let alloc_info = pool.get(alloc.alloc).ok_or(PoolError::InvalidIndex)?;
 
                 layout_bindings.insert(alloc.attr, alloc.pool);
-                alloc_offsets.push((alloc.attr, alloc_offset));
+                alloc_infos.push((alloc.attr, alloc_info));
             }
 
-            let instance = (mesh, alloc_offsets);
+            let instance = (mesh, alloc_infos);
 
             match layouts
                 .iter_mut()
@@ -542,4 +560,9 @@ impl MeshPool {
 
         Ok(layouts)
     }
+
+    pub fn get_bindings<'a>(&'a self, indices: MeshLayoutBindingIndices) -> MeshLayoutBindings<'a> {
+        let lock = self.pools.read().unwrap();
+        MeshLayoutBindings { lock, indices }
+    }
 }
diff --git a/src/pass.rs b/src/pass.rs
index 90d3dcb..7ec67c2 100644
--- a/src/pass.rs
+++ b/src/pass.rs
@@ -32,24 +32,24 @@ pub trait RenderPass: Send + Sync {
     /// Sets up a new instance of [Self::FrameData].
     ///
     /// Called when a render pass is added to [crate::Renderer].
-    fn create_frame_data(&mut self) -> Self::FrameData;
+    fn create_frame_data(&self) -> Self::FrameData;
 
     /// Initializes this frame's [Self::FrameData], and queries the pass for
     /// which phases to execute.
     ///
     /// This is the only opportunity the render pass has to mutate this frame's
     /// data this frame, so all setup for future phases should be done here.
-    fn begin_frame(&mut self, data: &mut Self::FrameData, phases: &mut Vec<Phase>);
+    fn begin_frame(&self, data: &mut Self::FrameData, phases: &mut Vec<Phase>);
 
     fn record_commands(&self, data: PhaseData<&Self::FrameData>, cmds: &mut wgpu::CommandEncoder);
 
-    fn record_compute(&self, data: PhaseData<&Self::FrameData>, cmds: &mut wgpu::ComputePass);
-
-    fn record_render(
-        &self,
+    fn record_compute<'a>(
+        &'a self,
         data: PhaseData<&Self::FrameData>,
-        cmds: &mut wgpu::RenderBundleEncoder,
+        cmds: &mut wgpu::ComputePass<'a>,
     );
+
+    fn record_render(&self, data: PhaseData<&Self::FrameData>) -> Option<wgpu::RenderBundle>;
 }
 
 /// The interface trait for [RenderPassBox], allowing use of a statically-sized
@@ -62,14 +62,14 @@ pub trait RenderPassBoxTrait: Send + Sync {
 
     fn record_commands(&self, data: IndexedPhaseData, cmds: &mut wgpu::CommandEncoder);
 
-    fn record_compute(&self, data: IndexedPhaseData, cmds: &mut wgpu::ComputePass);
+    fn record_compute<'a>(&'a self, data: IndexedPhaseData, cmds: &mut wgpu::ComputePass<'a>);
 
-    fn record_render(&self, data: IndexedPhaseData, cmds: &mut wgpu::RenderBundleEncoder);
+    fn record_render(&self, data: IndexedPhaseData) -> Option<wgpu::RenderBundle>;
 }
 
 /// A container for a reference-counted render pass instance and its frame data.
 pub struct RenderPassBox<T: RenderPass> {
-    render_pass: Arc<RwLock<T>>,
+    render_pass: Arc<T>,
     frame_data: Vec<T::FrameData>,
 }
 
@@ -78,10 +78,11 @@ impl<T: 'static + RenderPass> RenderPassBox<T> {
     ///
     /// Calls to [RenderPassBoxTrait] functions with frame indices greater
     /// than or equal to `frame_num` are out-of-bounds and will panic.
-    pub fn new(render_pass: Arc<RwLock<T>>, frame_num: usize) -> Box<dyn RenderPassBoxTrait> {
+    pub fn new(render_pass: Arc<T>, frame_num: usize) -> Box<dyn RenderPassBoxTrait> {
         let frame_data = {
-            let mut lock = render_pass.write().unwrap();
-            (0..frame_num).map(|_| lock.create_frame_data()).collect()
+            (0..frame_num)
+                .map(|_| render_pass.create_frame_data())
+                .collect()
         };
 
         Box::new(Self {
@@ -115,25 +116,22 @@ impl<T: RenderPass> RenderPassBox<T> {
 impl<T: RenderPass> RenderPassBoxTrait for RenderPassBox<T> {
     fn begin_frame(&mut self, data_index: usize, phases: &mut Vec<Phase>) {
         let frame_data = &mut self.frame_data[data_index];
-        let mut render_pass = self.render_pass.write().unwrap();
-        render_pass.begin_frame(frame_data, phases)
+        self.render_pass.begin_frame(frame_data, phases)
     }
 
     fn record_commands(&self, data: IndexedPhaseData, cmds: &mut wgpu::CommandEncoder) {
         let frame_data = self.get_frame_data(data);
-        let render_pass = self.render_pass.read().unwrap();
-        render_pass.record_commands(frame_data, cmds)
+        self.render_pass.record_commands(frame_data, cmds)
     }
 
-    fn record_compute(&self, data: IndexedPhaseData, cmds: &mut wgpu::ComputePass) {
+    fn record_compute<'a>(&'a self, data: IndexedPhaseData, cmds: &mut wgpu::ComputePass<'a>) {
         let frame_data = self.get_frame_data(data);
-        let render_pass = self.render_pass.read().unwrap();
-        render_pass.record_compute(frame_data, cmds)
+        self.render_pass.record_compute(frame_data, cmds)
     }
 
-    fn record_render(&self, data: IndexedPhaseData, cmds: &mut wgpu::RenderBundleEncoder) {
+    fn record_render(&self, data: IndexedPhaseData) -> Option<wgpu::RenderBundle> {
         let frame_data = self.get_frame_data(data);
-        let render_pass = self.render_pass.read().unwrap();
-        render_pass.record_render(frame_data, cmds)
+        // let render_pass = self.render_pass.read().unwrap();
+        self.render_pass.record_render(frame_data)
     }
 }
diff --git a/src/pass/mesh.rs b/src/pass/mesh.rs
index 62dde58..bd1b471 100644
--- a/src/pass/mesh.rs
+++ b/src/pass/mesh.rs
@@ -1,21 +1,154 @@
 use super::*;
 use crate::mesh::*;
 
+#[repr(C)]
+#[derive(Copy, Clone, bytemuck::Pod, bytemuck::Zeroable)]
+pub struct Vertex {
+    pub position: [f32; 3],
+    pub color: [f32; 3],
+}
+
+impl Attribute for Vertex {
+    fn get_usages() -> wgpu::BufferUsages {
+        wgpu::BufferUsages::VERTEX
+    }
+}
+
+const VERTEX_ATTRS: &[wgpu::VertexAttribute] =
+    &wgpu::vertex_attr_array![0 => Float32x3, 1 => Float32x3];
+
+impl Vertex {
+    pub fn desc() -> wgpu::VertexBufferLayout<'static> {
+        wgpu::VertexBufferLayout {
+            array_stride: std::mem::size_of::<Self>() as wgpu::BufferAddress,
+            step_mode: wgpu::VertexStepMode::Vertex,
+            attributes: VERTEX_ATTRS,
+        }
+    }
+}
+
+pub type Index = u32;
+
 pub struct FrameData {}
 
 pub struct MeshPass {
+    device: Arc<wgpu::Device>,
     attr_store: Arc<AttrStore>,
     mesh_pool: Arc<MeshPool>,
+    vertex_attr_id: AttrId,
+    index_attr_id: AttrId,
+    mesh_layout_id: MeshLayoutId,
+    example_mesh: MeshHandle,
+    opaque_pipeline: wgpu::RenderPipeline,
+    target_format: wgpu::TextureFormat,
 }
 
 impl MeshPass {
-    pub fn new(device: Arc<wgpu::Device>) -> Self {
+    pub fn new(device: Arc<wgpu::Device>, target_format: wgpu::TextureFormat) -> Self {
         let attr_store = AttrStore::new();
-        let mesh_pool = MeshPool::new(device, attr_store.to_owned());
+        let mesh_pool = MeshPool::new(device.clone(), attr_store.to_owned());
+
+        let vertex_attr_id = attr_store.get_type::<Vertex>();
+        let index_attr_id = attr_store.add(AttrInfo {
+            layout: AttrLayout {
+                size: std::mem::size_of::<Index>(),
+            },
+            usages: wgpu::BufferUsages::INDEX,
+            default_pool_size: 1_000_000,
+        });
+
+        let mut mesh_layout = MeshLayoutDesc::new();
+        mesh_layout.insert(vertex_attr_id, ());
+        mesh_layout.insert(index_attr_id, ());
+        let mesh_layout_id = mesh_pool.add_layout(mesh_layout).unwrap();
+
+        let example_vertices = vec![
+            Vertex {
+                position: [-0.5, 0.5, 0.0],
+                color: [1.0, 0.0, 0.0],
+            },
+            Vertex {
+                position: [0.5, 0.5, 0.0],
+                color: [0.0, 1.0, 0.0],
+            },
+            Vertex {
+                position: [0.0, -0.5, 0.0],
+                color: [0.0, 0.0, 1.0],
+            },
+        ];
+
+        let example_vertices = AttrBuffer {
+            id: vertex_attr_id,
+            count: example_vertices.len(),
+            data: bytemuck::cast_slice(&example_vertices).to_vec(),
+        };
+
+        let example_indices = vec![0u32, 1u32, 2u32];
+        let example_indices = AttrBuffer {
+            id: index_attr_id,
+            count: example_indices.len(),
+            data: bytemuck::cast_slice(&example_indices).to_vec(),
+        };
+
+        let mut example_mesh = MeshBuffer::default();
+        example_mesh.attributes.push(example_vertices);
+        example_mesh.attributes.push(example_indices);
+        let example_mesh = mesh_pool.load(example_mesh).unwrap();
+
+        let render_pipeline_layout =
+            device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+                label: Some("Render Pipeline Layout"),
+                bind_group_layouts: &[],
+                push_constant_ranges: &[],
+            });
+
+        let shader = device.create_shader_module(&wgpu::include_wgsl!("mesh_shader.wgsl"));
+
+        let opaque_pipeline = device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+            label: Some("Render Pipeline"),
+            layout: Some(&render_pipeline_layout),
+            vertex: wgpu::VertexState {
+                module: &shader,
+                entry_point: "vs_main",
+                buffers: &[Vertex::desc()],
+            },
+            fragment: Some(wgpu::FragmentState {
+                module: &shader,
+                entry_point: "fs_main",
+                targets: &[wgpu::ColorTargetState {
+                    format: target_format,
+                    blend: Some(wgpu::BlendState::REPLACE),
+                    write_mask: wgpu::ColorWrites::ALL,
+                }],
+            }),
+            primitive: wgpu::PrimitiveState {
+                topology: wgpu::PrimitiveTopology::TriangleList,
+                strip_index_format: None,
+                front_face: wgpu::FrontFace::Ccw,
+                cull_mode: None, // Some(wgpu::Face::Back),
+                polygon_mode: wgpu::PolygonMode::Fill,
+                unclipped_depth: false,
+                conservative: false,
+            },
+            depth_stencil: None,
+            multisample: wgpu::MultisampleState {
+                count: 1,
+                mask: !0,
+                alpha_to_coverage_enabled: false,
+            },
+            multiview: None,
+        });
 
         Self {
+            device,
             attr_store,
             mesh_pool,
+            index_attr_id,
+            vertex_attr_id,
+            mesh_layout_id,
+            example_mesh,
+            opaque_pipeline,
+            target_format,
         }
     }
 }
@@ -23,13 +156,14 @@ impl MeshPass {
 impl RenderPass for MeshPass {
     type FrameData = FrameData;
 
-    fn create_frame_data(&mut self) -> FrameData {
+    fn create_frame_data(&self) -> FrameData {
         FrameData {}
     }
 
-    fn begin_frame(&mut self, data: &mut FrameData, phases: &mut Vec<Phase>) {
+    fn begin_frame(&self, data: &mut FrameData, phases: &mut Vec<Phase>) {
         println!("MeshPass::begin_frame()");
 
+        phases.push(Phase::Upload);
         phases.push(Phase::Depth);
         phases.push(Phase::Opaque);
         phases.push(Phase::Transparent);
@@ -37,13 +171,74 @@ impl RenderPass for MeshPass {
 
     fn record_commands(&self, data: PhaseData<&FrameData>, cmds: &mut wgpu::CommandEncoder) {
         println!("MeshPass::record_commands(phase: {:?})", data.phase);
+        match data.phase {
+            Phase::Upload => self.mesh_pool.flush(cmds),
+            _ => {}
+        }
     }
 
     fn record_compute(&self, data: PhaseData<&FrameData>, cmds: &mut wgpu::ComputePass) {
         println!("MeshPass::record_compute(phase: {:?})", data.phase);
     }
 
-    fn record_render(&self, data: PhaseData<&FrameData>, cmds: &mut wgpu::RenderBundleEncoder) {
+    fn record_render(&self, data: PhaseData<&FrameData>) -> Option<wgpu::RenderBundle> {
         println!("MeshPass::record_render(phase: {:?})", data.phase);
+
+        let mut cmds =
+            self.device
+                .create_render_bundle_encoder(&wgpu::RenderBundleEncoderDescriptor {
+                    label: Some("Opaque Pass Render Bundle"),
+                    color_formats: &[self.target_format],
+                    depth_stencil: None,
+                    sample_count: 1,
+                    multiview: None,
+                });
+
+        let pipeline = match data.phase {
+            Phase::Opaque => &self.opaque_pipeline,
+            _ => return None,
+        };
+
+        let meshes = &[&self.example_mesh, &self.example_mesh];
+
+        // yikes
+        let mesh_bindings: Vec<(MeshLayoutBindings, Vec<(&&MeshHandle, MeshAllocInfos)>)> = self
+            .mesh_pool
+            .iter_meshes(self.mesh_layout_id, meshes.iter(), |v| v)
+            .unwrap()
+            .into_iter()
+            .map(
+                |MeshLayoutInstances {
+                     bindings,
+                     instances,
+                 }| (self.mesh_pool.get_bindings(bindings), instances),
+            )
+            .collect();
+
+        cmds.set_pipeline(pipeline);
+
+        for (bindings, instances) in mesh_bindings.iter() {
+            let vertices_pool = bindings.get(self.vertex_attr_id).unwrap();
+            let indices_pool = bindings.get(self.index_attr_id).unwrap();
+
+            cmds.set_vertex_buffer(0, vertices_pool.get_buffer().slice(..));
+            cmds.set_index_buffer(
+                indices_pool.get_buffer().slice(..),
+                wgpu::IndexFormat::Uint32,
+            );
+
+            for (mesh, infos) in instances {
+                let vertices = infos.iter().find(|i| i.0 == self.vertex_attr_id).unwrap().1;
+                let indices = infos.iter().find(|i| i.0 == self.index_attr_id).unwrap().1;
+
+                let is_start = indices.offset as u32;
+                let is_end = is_start + indices.count as u32;
+
+                cmds.draw_indexed(is_start..is_end, vertices.offset as i32, 0..1);
+                println!("drew a mesh! {}..{} + {}", is_start, is_end, vertices.offset);
+            }
+        }
+
+        Some(cmds.finish(&wgpu::RenderBundleDescriptor::default()))
     }
 }
diff --git a/src/pass/mesh_shader.wgsl b/src/pass/mesh_shader.wgsl
new file mode 100644
index 0000000..93220be
--- /dev/null
+++ b/src/pass/mesh_shader.wgsl
@@ -0,0 +1,30 @@
+struct VertexInput {
+  [[location(0)]] position: vec3<f32>;
+  [[location(1)]] color: vec3<f32>;
+};
+
+struct VertexOutput {
+  [[builtin(position)]] clip_position: vec4<f32>;
+  [[location(0)]] position: vec3<f32>;
+  [[location(1)]] color: vec3<f32>;
+};
+
+[[stage(vertex)]]
+fn vs_main(
+  [[builtin(instance_index)]] mesh_idx: u32,
+  [[builtin(vertex_index)]] vertex_idx: u32,
+  vertex: VertexInput,
+) -> VertexOutput {
+  var out: VertexOutput;
+  out.clip_position = vec4<f32>(vertex.position, 1.0);
+  out.position = vertex.position;
+  out.color = vertex.color;
+  return out;
+}
+
+[[stage(fragment)]]
+fn fs_main(
+  frag: VertexOutput,
+) -> [[location(0)]] vec4<f32> {
+  return vec4<f32>(frag.color, 1.0);
+}
diff --git a/src/staging.rs b/src/staging.rs
index 19a694d..5bb72dd 100644
--- a/src/staging.rs
+++ b/src/staging.rs
@@ -64,6 +64,9 @@ impl<T: Clone> StagingPool<T> {
                 on_complete(copy.target);
             }
         }
+
+        drop(src_view);
+        src.unmap();
     }
 
     pub fn queue_copies(&self, copies: Vec<CopyBuffer<T>>) {