mtrace && fork

[Ulrich Drepper]

Andreas Jaeger <aj@suse.de> writes:

> What do you think of the patch?  Is it ok to apply?  Or how shall we
> handle the situation?

I don't like this.  The correct solution is to move the atfork
handling from the libpthread to the libc level.  mtrace() then can
register a handler for the child to disable mtrace.

If somebody wants to come up with a patch for this, fine.  I'll not
have any time for this in the next days and it'll probably not make
today's test release.  What should be done now is to mark the
descriptor as close on exec.

-- 
---------------.                          ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------