Electron Releases

Description of Change

On a cold launch (after a reboot, an app update, or once the app has been evicted from the page cache), macOS pages the ~200 MB framework in with thousands of small reads that stall the main thread and every child process. This reads those files sequentially on a background thread as the browser process starts, so the faults hit the page cache instead.

• shell/app/electron_prefetch_mac.mm: reads the framework binary, libffmpeg.dylib and the framework's .pak/.dat/.bin resources on a background thread at the top of ElectronMain (browser process only).
• Skipped on warm launches (framework __TEXT already resident per mincore) and when memory pressure isn't normal or less than 512 MB is available.
• Universal builds read only the running architecture's slice and skip the other architecture's V8 snapshot.
• Runs at IOPOL_UTILITY so it yields to the main thread's own reads.

Spawn → first contentful paint of a localhost page, Apple silicon, internal SSD, release build, paired runs. Cold = app bundle evicted from the page cache before each launch.

Checklist

• I have built and tested this change
• I have filled out the PR description
• I have reviewed and verified the changes
• npm test passes

Release Notes

Notes: Improved cold-launch time on macOS by prefetching Electron's framework files.