zed-industries/zed · critical

Failed to get active display list. Result: {result}

Error message

Failed to get active display list. Result: {result}

What it means

On macOS, GPUI enumerates connected monitors by calling the CoreGraphics function CGGetActiveDisplayList into a fixed 32-entry buffer. If CoreGraphics returns any non-zero CGError code, gpui_macos panics because it cannot build the display list the platform needs. The panic fires inside the API that lists displays, typically during application startup or while reacting to a screen reconfiguration.

Source

Thrown at crates/gpui_macos/src/display.rs:63

    }

    /// Obtains an iterator over all currently active system displays.
    pub fn all() -> impl Iterator<Item = Self> {
        unsafe {
            // We're assuming there aren't more than 32 displays connected to the system.
            let mut displays = Vec::with_capacity(32);
            let mut display_count = 0;
            let result = CGGetActiveDisplayList(
                displays.capacity() as u32,
                displays.as_mut_ptr(),
                &mut display_count,
            );

            if result == 0 {
                displays.set_len(display_count as usize);
                displays.into_iter().map(MacDisplay)
            } else {
                panic!("Failed to get active display list. Result: {result}");
            }
        }
    }
}

#[link(name = "ApplicationServices", kind = "framework")]
unsafe extern "C" {
    fn CGDisplayCreateUUIDFromDisplayID(display: CGDirectDisplayID) -> CFUUIDRef;
}

impl PlatformDisplay for MacDisplay {
    fn id(&self) -> DisplayId {
        DisplayId::new(self.0 as u64)
    }

    fn uuid(&self) -> Result<Uuid> {
        let cfuuid = unsafe { CGDisplayCreateUUIDFromDisplayID(self.0 as CGDirectDisplayID) };
        anyhow::ensure!(

View on GitHub (pinned to f4178619ac)

Solutions

1. Run the app from a normal GUI login session (WindowServer reachable), not a bare SSH or CI shell on macOS
2. Retry launching after the display change settles: unplug/replug external monitors, or reboot to reset CoreGraphics state
3. If it repeats, reproduce with a tiny program calling CGGetActiveDisplayList and capture the returned CGError code, then report it upstream with that code
4. If you maintain gpui_macos, replace the panic with logging plus an empty iterator (or a Result) so callers can degrade gracefully

Example fix

// before (crates/gpui_macos/src/display.rs)
if result == 0 {
    displays.set_len(display_count as usize);
    displays.into_iter().map(MacDisplay)
} else {
    panic!("Failed to get active display list. Result: {result}");
}

// after: log and degrade instead of killing the process
if result == 0 {
    displays.set_len(display_count as usize);
    displays.into_iter().map(MacDisplay)
} else {
    log::error!("CGGetActiveDisplayList failed with {result}; continuing with no displays");
    Vec::new().into_iter().map(MacDisplay)
}

// probe CoreGraphics before initializing gpui on macOS
use core_graphics::display::CGDisplay;

fn displays_available() -> bool {
    CGDisplay::active_displays()
        .map(|displays| !displays.is_empty())
        .unwrap_or(false)
}

if !displays_available() {
    eprintln!("no displays via CoreGraphics; run inside a GUI session");
    std::process::exit(1);
}

// convert the startup panic into a diagnosable skip (CI/headless runs)
let boot = std::panic::catch_unwind(|| {
    // gpui Application::new().run(...) path that lists displays
});
if boot.is_err() {
    eprintln!("display initialization failed; aborting");
}

When it happens

Trigger: Calling gpui's display enumeration (the platform API backed by this code in crates/gpui_macos/src/display.rs, e.g. all_displays()) when CGGetActiveDisplayList returns a CGError: no WindowServer connection (headless SSH/CI session), a display reconfiguration in progress (plug/unplug, sleep/wake, GPU switching), or more active displays than the pre-allocated 32-slot buffer.

Common situations: Running a GPUI-based app over SSH or in CI on macOS where the WindowServer is unreachable; launching exactly while macOS switches GPUs or renegotiates displays after wake; kiosk or virtual-display setups with a very large display count.

Related errors

• RealtimeAudio priority should use spawn_realtime, not dispat
• database not initialized
• blocking sender returned without value
• Too many consecutive GPU errors. Last error: {error}
• Expected to render a table row

AI-assisted analysis of zed-industries/zed@f4178619ac (2026-08-20). Data as JSON: /api/errors/2b2fb2f241b00a83. Report an issue: GitHub.