Skip to content

[core-remote] Remove mutexes and other simplifications #10074

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
teoxoy merged 8 commits into from
Aug 14, 2026
Merged

[core-remote] Remove mutexes and other simplifications #10074

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
39df360
[core-remote] IdentityManager only for resource alloc and always requ…
sagudev Aug 14, 2026
c7aeca7
[core-remote] remove mutex from `IdentityManager`
sagudev Aug 14, 2026
0317897
[core-remote] Global does not need to be Send or Sync
sagudev Aug 14, 2026
064a2b5
[core-remote] Use RefCell instead of RwLock in Registry
sagudev Aug 14, 2026
93a1264
[core-remote] Reuse RefCell from Registery for &mut RenderPass
sagudev Aug 14, 2026
cb5c816
[core-remote] Reuse RefCell from Registery for &mut ComputePass
sagudev Aug 14, 2026
eb65b99
[core-remote] Reuse RefCell from Registery for &mut RenderBundleEncoder
sagudev Aug 14, 2026
2c52b10
[core-remote] rm parking_lot
sagudev Aug 14, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion Cargo.lock
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

1 change: 0 additions & 1 deletion wgpu-core-remote/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,6 +32,5 @@ serde = ["dep:serde", "wgpu-types/serde", "wgpu-core/serde"]
wgpu-core = { workspace = true, default-features = false }
wgpu-hal = { workspace = true, default-features = false }
wgpu-types.workspace = true
parking_lot.workspace = true
raw-window-handle.workspace = true
serde = { workspace = true, optional = true }
72 changes: 12 additions & 60 deletions wgpu-core-remote/src/global/bundle.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,11 +9,7 @@ impl crate::global::Global {
bind_group_id: Option<id::BindGroupId>,
offsets: &[wgt::DynamicOffset],
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.set_bind_group(
index,
Expand All @@ -27,11 +23,7 @@ impl crate::global::Global {
bundle_encoder: id::RenderBundleEncoderId,
pipeline_id: id::RenderPipelineId,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.set_pipeline(self.hub.render_pipelines.get(pipeline_id))
}
Expand All @@ -44,11 +36,7 @@ impl crate::global::Global {
offset: wgt::BufferAddress,
size: Option<wgt::BufferSize>,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.set_vertex_buffer(
slot,
Expand All @@ -66,11 +54,7 @@ impl crate::global::Global {
offset: wgt::BufferAddress,
size: Option<wgt::BufferSize>,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.set_index_buffer(self.hub.buffers.get(buffer), index_format, offset, size)
}
Expand All @@ -81,11 +65,7 @@ impl crate::global::Global {
offset: u32,
data: &[u8],
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.set_immediates(offset, data)
}
Expand All @@ -98,11 +78,7 @@ impl crate::global::Global {
first_vertex: u32,
first_instance: u32,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.draw(vertex_count, instance_count, first_vertex, first_instance)
}
Expand All @@ -116,11 +92,7 @@ impl crate::global::Global {
base_vertex: i32,
first_instance: u32,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.draw_indexed(
index_count,
Expand All @@ -137,11 +109,7 @@ impl crate::global::Global {
buffer_id: id::BufferId,
offset: wgt::BufferAddress,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.draw_indirect(self.hub.buffers.get(buffer_id), offset)
}
Expand All @@ -152,11 +120,7 @@ impl crate::global::Global {
buffer_id: id::BufferId,
offset: wgt::BufferAddress,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.draw_indexed_indirect(self.hub.buffers.get(buffer_id), offset)
}
Expand All @@ -166,11 +130,7 @@ impl crate::global::Global {
bundle_encoder: id::RenderBundleEncoderId,
label: &str,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.push_debug_group(label)
}
Expand All @@ -179,11 +139,7 @@ impl crate::global::Global {
&self,
bundle_encoder: id::RenderBundleEncoderId,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.pop_debug_group()
}
Expand All @@ -193,11 +149,7 @@ impl crate::global::Global {
bundle_encoder: id::RenderBundleEncoderId,
label: &str,
) -> Result<(), PassStateError> {
let bundle_encoder = self.hub.render_bundle_encoders.get(bundle_encoder);

let mut bundle_encoder = bundle_encoder
.try_lock()
.expect("RenderBundleEncoders should not be accessed concurrently");
let mut bundle_encoder = self.hub.render_bundle_encoders.get_mut(bundle_encoder);

bundle_encoder.insert_debug_marker(label)
}
Expand Down
2 changes: 1 addition & 1 deletion wgpu-core-remote/src/global/command_encoder.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ impl Global {
&self,
encoder_id: CommandEncoderId,
desc: &wgt::CommandBufferDescriptor<Label>,
id_in: Option<id::CommandBufferId>,
id_in: id::CommandBufferId,
) -> (id::CommandBufferId, Option<(String, CommandEncoderError)>) {
let hub = &self.hub;
let cmd_enc = hub.command_encoders.get(encoder_id);
Expand Down
70 changes: 14 additions & 56 deletions wgpu-core-remote/src/global/compute_pass.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,8 +1,5 @@
use alloc::sync::Arc;

use alloc::borrow::Cow;

use parking_lot::Mutex;
use wgpu_core::command::{
CommandEncoderError, ComputePassDescriptor, EncoderStateError, PassStateError,
PassTimestampWrites,
Expand All @@ -27,7 +24,7 @@ impl Global {
&self,
encoder_id: id::CommandEncoderId,
desc: &ComputePassDescriptor<'_, PassTimestampWrites<id::QuerySetId>>,
id_in: Option<id::ComputePassEncoderId>,
id_in: id::ComputePassEncoderId,
) -> (id::ComputePassEncoderId, Option<CommandEncoderError>) {
let fid = self.hub.compute_passes.prepare(id_in);

Expand All @@ -47,10 +44,7 @@ impl Global {

let (pass, err) = cmd_enc.begin_compute_pass(&desc);

// no lock rank here because only one thread should be using compute pass
// and it's only used by id variants of compute pass methods on global
// so no deadlock (or concurrent lock) should happen in practise
let id = fid.assign(Arc::new(Mutex::new(pass)));
let id = fid.assign(pass);

(id, err)
}
Expand All @@ -59,10 +53,7 @@ impl Global {
&self,
pass_id: id::ComputePassEncoderId,
) -> Result<(), EncoderStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.end()
}

Expand All @@ -87,10 +78,7 @@ impl Global {
bind_group_id: Option<id::BindGroupId>,
offsets: &[DynamicOffset],
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.set_bind_group(
index,
bind_group_id.map(|bind_group_id| self.hub.bind_groups.get(bind_group_id)),
Expand All @@ -103,10 +91,7 @@ impl Global {
pass_id: id::ComputePassEncoderId,
pipeline_id: id::ComputePipelineId,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
let pipeline = self.hub.compute_pipelines.get(pipeline_id);
pass.set_pipeline(pipeline)
}
Expand All @@ -117,10 +102,7 @@ impl Global {
offset: u32,
data: &[u8],
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.set_immediates(offset, data)
}

Expand All @@ -131,10 +113,7 @@ impl Global {
groups_y: u32,
groups_z: u32,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.dispatch_workgroups(groups_x, groups_y, groups_z)
}

Expand All @@ -144,10 +123,7 @@ impl Global {
buffer_id: id::BufferId,
offset: BufferAddress,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.dispatch_workgroups_indirect(self.hub.buffers.get(buffer_id), offset)
}

Expand All @@ -157,21 +133,15 @@ impl Global {
label: &str,
color: u32,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.push_debug_group(label, color)
}

pub fn compute_pass_pop_debug_group(
&self,
pass_id: id::ComputePassEncoderId,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.pop_debug_group()
}

Expand All @@ -181,10 +151,7 @@ impl Global {
label: &str,
color: u32,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.insert_debug_marker(label, color)
}

Expand All @@ -194,10 +161,7 @@ impl Global {
query_set_id: id::QuerySetId,
query_index: u32,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
let query_set = self.hub.query_sets.get(query_set_id);
pass.write_timestamp(query_set, query_index)
}
Expand All @@ -208,10 +172,7 @@ impl Global {
query_set_id: id::QuerySetId,
query_index: u32,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
let query_set = self.hub.query_sets.get(query_set_id);
pass.begin_pipeline_statistics_query(query_set, query_index)
}
Expand All @@ -220,10 +181,7 @@ impl Global {
&self,
pass_id: id::ComputePassEncoderId,
) -> Result<(), PassStateError> {
let pass = self.hub.compute_passes.get(pass_id);
let mut pass = pass
.try_lock()
.expect("ComputePasses should not be accessed concurrently");
let mut pass = self.hub.compute_passes.get_mut(pass_id);
pass.end_pipeline_statistics_query()
}
}
Loading
Loading