Keeley Hammond 58d6da6b16 fix: only bind ElectronAutofillDriver on primary main frames ...

The `mojom::ElectronAutofillDriver` associated interface was being
registered for every `RenderFrameHost`, and `AutofillDriverFactory`
would create a driver for any live frame. In practice the native
autofill popup is only exercised from `<datalist>` / `<select>`
elements on the primary main frame — the autofill spec even notes
that `webContents.sendInputEvent` cannot route keyboard input to an
OOPIF, so subframes have no working path to this UI.

Scope the interface to where it is actually used:

* Gate the associated registry entry in
  `ElectronBrowserClient::RegisterAssociatedInterfaceBindersForRenderFrameHost`
  on `IsInPrimaryMainFrame()`, mirroring how `ElectronApiIPC` is
  already scoped in the same function.
* Return early from `AutofillDriverFactory::DriverForFrame` for any
  non-primary-main-frame RFH so the factory never stores a driver
  for an unsupported frame.
* Re-check `IsInPrimaryMainFrame()` inside
  `AutofillDriver::ShowAutofillPopup` as defense in depth for
  messages that race with a navigation.

2026-04-11 20:06:34 -07:00

..

app

fix: use audit token instead of PID for parent code-signature check (#50907)

2026-04-11 03:42:52 -04:00

browser

fix: only bind ElectronAutofillDriver on primary main frames

2026-04-11 20:06:34 -07:00

common

fix: use audit token instead of PID for parent code-signature check (#50907)

2026-04-11 03:42:52 -04:00

renderer

feat: capture JS stack trace on renderer OOM (#50043)

2026-04-10 15:13:16 -04:00

services/node

chore: bump chromium to 148.0.7768.0 (main) (#50599)

2026-04-06 20:38:46 -07:00

utility

fix: broken OOP window.print() on macOS/Linux (#45214)

2025-01-20 10:23:44 +01:00