wires user configured `FontFallbacks` into the cosmic text path. the
chain is resolved at font load time and stored on each `LoadedFont`.
`layout_line` splits each `FontRun` into spans by codepoint coverage and
emits one `Attrs` per slot so cosmic text shapes each span with the
correct face. inheriting codepoints (marks, zwj, zwnj, variation
selectors) stick to the current span so emoji zwj sequences and
combining marks are not torn across faces.
Closes#17254
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] Performance impact has been considered and is acceptable
- [x] Tests cover the new/changed behavior
- [] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- added support for `buffer_font_fallbacks` on linux
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)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes#32792Closes#38266Closes#54133
Release Notes:
- Fixed graphical corruption that could occur when using Wayland
----
**What**: Fixes flickering on Wayland (Sway/wlroots) under CPU load
(e.g. rust-analyzer running). The bug only reproduces in release builds
- debug builds are too slow to hit the race window.
Environment where this was reproduced: Intel GPU (both Xe KMD and i915),
Mesa 26.0.4, Sway 1.11.
**Why**: When wgpu presents a frame on Wayland+Vulkan, it calls
`vkQueuePresentKHR,` which - as required by the Vulkan spec -
synchronously issues `wl_surface.attach`, `wl_surface.damage`, and
`wl_surface.commit` to the compositor before returning. The commit also
picks up any pending frame callbacks.
Zed's `completed_frame()` independently calls `state.surface.commit()`.
This is a redundant second commit on the same surface. Under load, the
Wayland socket dispatch can be delayed enough that both commits are in
flight in close succession, and the ordering becomes timing-dependent.
When Zed's commit reaches the compositor before Mesa's attach+commit
sequence has been fully flushed, the compositor sees a commit with no
buffer attached, fires `wl_callback::Done` immediately, and Zed starts
the next frame too early - Mesa's real buffer arrives late, causing the
visible flicker.
Under no load, Mesa's synchronous commit consistently reaches the
compositor first, so the bug doesn't appear.
**Fix**: Track whether `renderer.draw()` actually called
`frame.present()`. When it did, Mesa owns the `wl_surface.commit()` for
that frame - skip Zed's commit in `completed_frame()`. Only commit
ourselves when wgpu didn't present (surface not configured, lost,
occluded, etc.) - in those cases Mesa won't commit, and we need to keep
the frame callback alive.
---------
Co-authored-by: Benjamin Laib <b5l@users.noreply.github.com>
Co-authored-by: John Tur <john-tur@outlook.com>
Updates #54349
There were two problems:
* The crash never happened, instead we'd always retry.
* When re-trying it seemed like we were picking llvmpipe. Claude's
suggestion was that immediately after wake, the real GPU isn't yet
awake, and so we pick llvm. Avoid this by disallowing llvmpipe on retry
Release Notes:
- linux: Reduced crash rate when recovering GPUs
We were assuming that `Bgra8Unorm` always works, which was causing the
panic on configurations that don't support it ("wgpu-hal invariant was
violated: Requested feature is not available on this device")
This change makes a proper fallback to `Rgba8Unorm`. In the worst case,
if a device doesn't support that as well (highly unlikely), we return a
proper error instead of a panic.
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
Closes ZED-69S
Release Notes:
- Fixed panic on devices that don't support BGRA8
Corresponding font-kit commit:
94b0f28166
Before, Zed has 3 different versions of `dirs` crate to compile, this PR
bumps the related dependencies to have one, latest, version of `dirs`
instead.
Release Notes:
- N/A
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...
#11 core::option::expect_failed
#12 gpui_wgpu::wgpu_atlas::WgpuAtlas::before_frame
#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 should prevent the "wgpu-hal invariant was violated (usage error):
Requested feature is not available on this device" panic, although I'm
not 100% sure (there's a possibility that a device reports the feature
but fails to use it). At the very least, we get more logging
I've tested this fix by patching code to emulate a situation where
device features change after the creation, but that may be not
representative of the real world failure.
Addresses ZED-5G1
Release Notes:
- N/A
## Summary
- Add `unconfigure_surface()` and `replace_surface()` methods to
`WgpuRenderer` for mobile platform window lifecycle management
- Prefer `PresentMode::Mailbox` (triple-buffering) over `Fifo` to avoid
blocking during lifecycle transitions
- Early return in `draw()` when surface is unconfigured to prevent
driver hangs
## Motivation
On Android, the native window (`ANativeWindow`) is destroyed when the
app goes to the background and recreated when it returns to the
foreground. The same happens during orientation changes. Without surface
lifecycle methods, the only option is to destroy the entire
`WgpuRenderer` and create a new one on resume.
The problem: GPUI's scene cache holds `AtlasTextureId` references from
the old renderer's atlas. A new renderer has an empty atlas, so those
cached IDs cause index-out-of-bounds panics.
The fix: Keep the renderer (device, queue, atlas, pipelines) alive
across surface destruction. Only the wgpu `Surface` needs to be
replaced.
### `unconfigure_surface()`
Marks the surface as unconfigured so `draw()` skips rendering via the
existing `surface_configured` guard. Drops intermediate textures that
reference the old surface dimensions. The renderer stays fully alive.
### `replace_surface()`
Creates a new `wgpu::Surface` from fresh window handles using the
**same** `wgpu::Instance` that created the original adapter/device.
Reconfigures the surface and marks it as configured so rendering
resumes. All cached atlas textures remain valid.
### PresentMode::Mailbox
`Fifo` (VSync) blocks in `get_current_texture()` and can deadlock if the
compositor is frozen during a lifecycle transition (e.g.
`TerminateWindow` → `InitWindow` on Android). Mailbox (triple-buffering)
avoids this. Falls back to `AutoNoVsync` → `Fifo` if unsupported.
### draw() early return
Some drivers (notably Adreno) block indefinitely when acquiring a
texture from an unconfigured surface. The early return prevents this.
## Context
This is needed by
[gpui-mobile](https://github.com/itsbalamurali/gpui-mobile), a project
bringing GPUI to Android and iOS. The Android implementation needs these
methods to handle:
1. **Background/foreground transitions** — `TerminateWindow` destroys
the native window, `InitWindow` recreates it
2. **Orientation changes** — Surface is destroyed and recreated with new
dimensions
3. **Split-screen transitions** — Similar surface recreation
Without this change, we maintain a local fork of `gpui_wgpu` with just
these additions. Upstreaming them would let mobile platform
implementations use the official crate directly.
## Test plan
- [x] Tested on Android (Motorola, Adreno 720 GPU) — 3 consecutive
background/foreground cycles, zero panics, atlas textures preserved
- [x] Tested orientation changes (portrait→landscape→portrait) — surface
replacement completed in <40ms per rotation
- [x] Verified `draw()` correctly skips rendering when surface is
unconfigured
- [x] Verified no regression on desktop — methods are additive, existing
code paths unchanged
- [x] PresentMode fallback chain works on devices that don't support
Mailbox
## Release Notes
- N/A
## Summary
Implements the fix proposed by @israelrios in
https://github.com/zed-industries/zed/issues/50269#issuecomment-4049006746.
The issue comment narrowed the hang down to X11 window teardown:
- when the Settings window is closed on X11, the UI thread can stall
while `WgpuRenderer` is dropped during final `X11WindowState` teardown
- at that point, the X11 window has already been destroyed
- the proposed fix is to make `WgpuRenderer::destroy()` release
surface-bound GPU resources eagerly so the native window is destroyed
only after the `wgpu::Surface` and related resources are gone
This PR applies that change in `crates/gpui_wgpu/src/wgpu_renderer.rs`
by calling `self.resources.take()` inside `destroy()`.
## Validation
I confirmed this fix locally on Linux Mint 21.3 Cinnamon by applying it
to the `v0.227.1` build:
- opening and closing Settings no longer causes the main window to hang
## References
- Refs #50269
- Original analysis and fix proposal by @israelrios:
https://github.com/zed-industries/zed/issues/50269#issuecomment-4049006746
Release Notes:
- Fixed a Linux/X11 issue where closing the Settings window could cause
Zed to hang.
This uses the compositor hints if available to pick the best GPU. If
none is
available, it tries each GPU in turn, and the first that actually works
is chosen
Release Notes:
- Linux: Select a more appropriate GPU
---------
Co-authored-by: John Tur <john-tur@outlook.com>
Fixes Zed-5AW
Fixes Zed-5AP
Claude believes this is the right fix, but would love someone who knows
more about graphics than me to take a look: @reflectronic / @zortax?
The panic is:
```
wgpu error: Validation Error
Caused by:
In Texture::create_view
Texture with 'path_intermediate' label is invalid
gpui::platform::wgpu::wgpu_renderer::WgpuRenderer::create_path_intermediate (wgpu_renderer.rs:742)
gpui::platform::wgpu::wgpu_renderer::WgpuRenderer::update_drawable_size (wgpu_renderer.rs:784)
gpui::platform::linux::x11:🪟:X11WindowStatePtr::set_bounds (window.rs:1169)
gpui::platform::linux::x11::client::X11Client::handle_event (client.rs:902)
```
or:
```
wgpu error: Validation Error
Caused by:
In Texture::create_view
Texture with 'path_intermediate' label is invalid
gpui::platform::wgpu::wgpu_renderer::WgpuRenderer::create_path_intermediate (wgpu_renderer.rs:742)
gpui::platform::wgpu::wgpu_renderer::WgpuRenderer::new (wgpu_renderer.rs:274)
gpui::platform::linux::x11:🪟:X11WindowState:🆕:{{closure}} (window.rs:698)
gpui::platform::linux::x11:🪟:X11WindowState::new (window.rs:488)
gpui::platform::linux::x11:🪟:X11Window::new (window.rs:814)
gpui::platform::linux::x11::client::X11Client::open_window (client.rs:1514)
gpui::platform::linux::platform::<T>::open_window (platform.rs:289)
gpui:🪟:Window::new (window.rs:1119)
gpui::app::App::open_window::{{closure}} (app.rs:1025)
gpui::app::App::update (app.rs:835)
gpui::app::App::open_window (app.rs:1022)
```
I haven't seen a Wayland equivalent (not sure if that's because it
doesn't happen on Wayalnd or because I havent' seen it yet)
Release Notes:
- Linux: Fixed a panic in the new WPGU renderer during resize
Closes#49435
Before you mark this PR as ready for review, make sure that you have:
- [ ] Added a solid test coverage and/or screenshots from doing manual
testing
- [ ] Done a self-review taking into account security and performance
aspects
- [ ] Aligned any UI changes with the [UI
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- Linux: Reduced GPU memory usage during resize
Implements a basic web platform for the wasm32-unknown-unknown target
for gpui
Release Notes:
- N/A *or* Added/Fixed/Improved ...
---------
Co-authored-by: John Tur <john-tur@outlook.com>
Port of #10314 to the wgpu renderer
Closes #ISSUE
Before you mark this PR as ready for review, make sure that you have:
- [ ] Added a solid test coverage and/or screenshots from doing manual
testing
- [ ] Done a self-review taking into account security and performance
aspects
- [ ] Aligned any UI changes with the [UI
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- Fixed a panic when rendering an image larger than the GPU could
support.
Fixes ZED-54X
Release Notes:
- Linux: wait to request a graphics context until we have a window so we
can (ideally) pick a better context or (less ideally) fail more
gracefully.
---------
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
#2874 on steroids
Before you mark this PR as ready for review, make sure that you have:
- [ ] Added a solid test coverage and/or screenshots from doing manual
testing
- [ ] Done a self-review taking into account security and performance
aspects
- [ ] Aligned any UI changes with the [UI
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- N/A
---------
Co-authored-by: Eric Holk <eric@zed.dev>
