Crash on un-aggregated SeriesPoint for 1M series

[nikolausWest]

Repro:
data: 1M_line_series.rrd.zip

RUST_LOG=debug cargo r -p rerun-cli --no-default-features --features native_viewer --release -- -j1 1M_line_series.rrd

Setup:

marker.crash.mp4

CLI output:

[2024-02-06T08:15:16Z DEBUG re_memory::memory_limit] Setting memory limit to 24.0 GiB, which is 75% of total available memory (32.0 GiB).
[2024-02-06T08:15:16Z INFO  re_data_source::load_file] Loading "1M_line_series.rrd"…
[2024-02-06T08:15:16Z DEBUG re_data_source::data_loader::loader_rrd] Loading rrd data from filesystem… filepath="1M_line_series.rrd" loader="rerun.data_loaders.Rrd"
[2024-02-06T08:15:16Z DEBUG re_data_source::load_file] compatible loader found loader="rerun.data_loaders.Rrd" path="1M_line_series.rrd"
[2024-02-06T08:15:16Z DEBUG eframe] Using the wgpu renderer
[2024-02-06T08:15:16Z DEBUG re_data_source::load_file] compatible loader found loader="rerun.data_loaders.Archetype" path="1M_line_series.rrd"
[2024-02-06T08:15:16Z DEBUG re_data_source::data_loader::loader_external] Found duplicated data-loader in $PATH name="rerun-loader-docx" paths=["/Users/niko/.local/bin/rerun-loader-docx", "/Users/niko/.local/bin/rerun-loader-docx"]
[2024-02-06T08:15:16Z DEBUG re_data_source::data_loader::loader_external] Found duplicated data-loader in $PATH name="rerun-loader-tfrecord" paths=["/Users/niko/.local/bin/rerun-loader-tfrecord", "/Users/niko/.local/bin/rerun-loader-tfrecord"]
[2024-02-06T08:15:16Z DEBUG re_data_source::data_loader::loader_external] Loading data from filesystem using external loader… filepath="1M_line_series.rrd" loader="/Users/niko/.local/bin/rerun-loader-docx"
[2024-02-06T08:15:16Z DEBUG re_data_source::data_loader::loader_external] Loading data from filesystem using external loader… filepath="1M_line_series.rrd" loader="/Users/niko/.local/bin/rerun-loader-tfrecord"
[2024-02-06T08:15:16Z DEBUG eframe::native::wgpu_integration] Event::Resumed
[2024-02-06T08:15:16Z DEBUG eframe::native::file_storage] Loading app state from "/Users/niko/Library/Application Support/rerun/app.ron"…
[2024-02-06T08:15:16Z DEBUG egui_wgpu] Picked the only available wgpu adapter: backend: Metal, device_type: IntegratedGpu, name: "Apple M1 Pro"
[2024-02-06T08:15:16Z DEBUG re_renderer::context] wgpu adapter backend: Metal, device_type: IntegratedGpu, name: "Apple M1 Pro"
[2024-02-06T08:15:16Z DEBUG egui::context] Loading new font definitions
[2024-02-06T08:15:19Z DEBUG re_viewer] Shutting down ui_waker thread
[2024-02-06T08:15:19Z DEBUG re_viewer::app] Opening a new recording: StoreInfo { application_id: ApplicationId("rerun_example_plot_dashboard_stress"), store_id: StoreId { kind: Recording, id: "ac487959-f0b2-4dc2-9a9f-d14aeadbc578" }, is_official_example: false, started: "2024-02-05 21:04:49.408498Z", store_source: RustSdk { rustc_version: "1.74.0 (79e9716c9 2023-11-13)", llvm_version: "17.0.4" }, store_kind: Recording }
[2024-02-06T08:15:19Z DEBUG re_viewer::store_hub] Trying to load blueprint for rerun_example_plot_dashboard_stress from "/Users/niko/Library/Application Support/rerun/blueprints/rerun_example_plot_dashboard_stress.blueprint"
[2024-02-06T08:15:19Z DEBUG re_viewer::store_hub] Switching blueprint for ApplicationId("rerun_example_plot_dashboard_stress") to StoreId { kind: Blueprint, id: "rerun_example_plot_dashboard_stress" }
[2024-02-06T08:15:36Z DEBUG re_viewer::app] Data source file://1M_line_series.rrd has left
[2024-02-06T08:15:46Z DEBUG re_viewer::store_hub] Saving blueprint for <welcome screen> to "/Users/niko/Library/Application Support/rerun/blueprints/-welcome-screen--e59dfa2c766d0b8f.blueprint"
[2024-02-06T08:15:46Z DEBUG re_viewer::store_hub] Saving blueprint for rerun_example_plot_dashboard_stress to "/Users/niko/Library/Application Support/rerun/blueprints/rerun_example_plot_dashboard_stress.blueprint"
[2024-02-06T08:16:22Z DEBUG re_viewer_context::space_view::view_query] Overriding /space_view/8e4a5141-e7eb-4e72-82e0-fc12846f7abf with EntityProperties { visible: true, visible_history: ExtraQueryHistory { enabled: false, nanos: VisibleHistory { from: Infinite, to: Infinite }, sequences: VisibleHistory { from: Infinite, to: Infinite } }, interactive: true, color_mapper: Auto(Colormap(Turbo)), pinhole_image_plane_distance: Auto(0.0), backproject_depth: Auto(true), depth_from_world_scale: Auto(1.0), backproject_radius_scale: Auto(1.0), transform_3d_visible: Auto(false), transform_3d_size: Auto(1.0), show_legend: Auto(true), legend_location: None, time_series_aggregator: UserEdited(Off) }

thread 'main' panicked at 'Failed to create staging buffer for index data'
egui-wgpu/src/renderer.rs:834
stack backtrace:


Troubleshooting Rerun: https://www.rerun.io/docs/getting-started/troubleshooting
Report bugs: https://github.com/rerun-io/rerun/issues

[emilk]

Improved error message for next time: emilk/egui#3986

thread 'main' panicked at 'Failed to create staging buffer for vertex data. Vertex count: 10 011 232. Required vertex buffer size: 200 224 640. Actual size 349 450 240 and capacity: 349 450 240 (bytes)'

I guess this goes over some WebGPU min-spec limit @Wumpf ?

The crash is in: https://docs.rs/wgpu/latest/wgpu/struct.Queue.html#method.write_buffer_with

And the given size is less than the buffer size, so according to the specs of write_buffer_with, it should NOT fail, yet it does.