6850720: Allow POSIX_SPAWN to be used for ProcessImpl on Linux
Martin Buchholz
martinrb at google.com
Mon Oct 22 00:25:24 UTC 2018
As author of the vfork strategy ...
I'm supportive of the directions in this thread.
David's patch seems like clear progress (although maybe now that we have
configure, we can make the spawn strategy conditional on HAVE_SPAWN_H)
vfork is even (!) less in favor than it used to be, so migrating off of it
seems good. Just make sure we don't bring back the OOM problems or
unacceptable overhead from e.g. an second exec. Create a 100GB heap, then
run /bin/true 1000 times.
That big internal comment I wrote years ago is suffering from bitrot.
IIRC vfork used to call a special vfork system call, NOT clone, years ago.
More information about the core-libs-dev
mailing list