Repository navigation
perf(gpui): keep the profiler journal ring out of memory until it is written - #11
Draft
rubnogueira wants to merge 1 commit into
Draft
rubnogueira wants to merge 1 commit into
rubnogueira wants to merge 1 commit into
Conversation
…written The journal is installed in every App, but only Zed's hang detection reads it. An embedder that never reads it still paid for the whole 4 MiB ring: empty slots carried a non-zero sentinel sequence, so allocating the ring wrote every slot and the operating system backed every page at startup. A slot now stores its published sequence plus one, so an empty slot is all zeros and the ring is allocated with `new_zeroed_slice`. Pages become resident as entries are first written, and a process that records a few entries touches a few pages.
knoopx
pushed a commit
to knoopx/zed
that referenced
this pull request
Oct 19, 2026
) When atlas tiles are rapidly allocated and freed (e.g. watching a shared screen in Collab), a texture can become unreferenced and be removed while GPU uploads for it are still pending. On the next frame, `flush_uploads` indexes into the now-empty texture slot and panics: ``` thread 'main' panicked at crates/gpui_wgpu/src/wgpu_atlas.rs:231:40: texture must exist... remorses#11 core::option::expect_failed remorses#12 gpui_wgpu::wgpu_atlas::WgpuAtlas::before_frame remorses#13 gpui_wgpu::wgpu_renderer::WgpuRenderer::draw ``` This change drains pending uploads for a texture when it becomes unreferenced in `remove`, and skips uploads for missing textures in `flush_uploads` as a safety net. Self-Review Checklist: - [x] I've reviewed my own diff for quality, security, and reliability - [x] Unsafe blocks (if any) have justifying comments - [x] The content is consistent with the [UI/UX checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist) - [x] Tests cover the new/changed behavior - [x] Performance impact has been considered and is acceptable Release Notes: - Fixed occasional crashes when viewing a screen share
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The profiler journal is installed in every
App, but only Zed's hang detection reads it. An embedder that never reads it (GPUIX does not) still paid for the whole 4 MiB ring: empty slots carried a non-zero sentinel sequence, so allocating the ring wrote every slot and the operating system backed every page at startup.A slot now stores its published sequence plus one, so an empty slot is all zeros, and the ring is allocated with
new_zeroed_slice. Pages become resident as entries are first written, so a process that records a few entries touches a few pages.One file,
crates/gpui/src/profiler/journal.rs. The sequence numbers the rest of the module sees are unchanged: only the stored value is offset, andtry_publishandtry_readapply the offset on both sides.Notes
gpuix(3db1867, the merge of feat(gpui): list padding on all sides, averaged scroll extent, hover exit, font lookup fix #10).remorses/gpuixsubmodule pointer moves to the head of this branch in a companion PR. Merge this one first.Test plan
cargo test -p gpui --features profiler,test-support journal: 23 passed. The journal tests are gated behind those features, so without them zero tests run. This ran on the previousgpuixtip; the merge of feat(gpui): list padding on all sides, averaged scroll extent, hover exit, font lookup fix #10 did not touchcrates/gpui/src/profiler/, so the code under test is identical.3db1867itself, and not measured outside macOS.Authored with Claude (Claude Sonnet 5.5) in Cursor.