https://bugs.kde.org/show_bug.cgi?id=384732
--- Comment #4 from Mark Wielaard <m...@klomp.org> --- I wonder if we can also rely on the CLONE_VFORK mechanism. So if we see a clone with CLONE_VFORK | CLONE_VM we don't drop the CLONE_VM, but do freeze all other threads till we see the exec/exit syscall. Since till the exit/exec the host and child memory are the same and only that child thread is running it should keep the valgrind datastructures sane. -- You are receiving this mail because: You are watching all bug changes.