[wasm] clear and set thread-in-wasm flag on runtime calls
This was causing GC stress failures. Garbage collections can happen during runtime calls, such was WasmStackGuard. If the collection cleans up Wasm objects, then they will have to modify the trap handler data structures, which requires taking a lock. This lock can only be taken if the thread-in-wasm flag is clear. We were getting crashes because this flag was not clear. This change fixes the issue by making sure any runtime calls from Wasm clear the thread-in-wasm flag and then restore it upon return. In addition, it cleans up the code by adding a helper function that generates the code to modify the flag. BUG= v8:6132 Change-Id: I95d43388dff60ba792c57fe13448a40a02ed4802 Reviewed-on: https://chromium-review.googlesource.com/458698 Commit-Queue: Eric Holk <eholk@chromium.org> Reviewed-by: Mircea Trofin <mtrofin@chromium.org> Reviewed-by: Brad Nelson <bradnelson@chromium.org> Cr-Commit-Position: refs/heads/master@{#44165}
Showing
Please
register
or
sign in
to comment