Re: next vforkup chapter: lwpctl

[Thor Lancelot Simon <tls%panix.com@localhost>]

On Mon, Feb 14, 2011 at 03:33:03PM +0200, Antti Kantee wrote:
> On Mon Feb 14 2011 at 14:27:10 +0100, Joerg Sonnenberger wrote:
> > On Mon, Feb 14, 2011 at 03:16:11PM +0200, Antti Kantee wrote:
> > > The following patch makes a vfork child update the parent's lwpctl area
> > > while the child is running.  Comments or better ideas?
> > 
> > I think this is a very good case of "don't do that". malloc after
> > vfork() is just wrong. If rumphijack internally forces that for normal
> > system calls, it should downgrade vfork() to fork().
> 
> This is not about rumphijack.  Look at e.g. sh and make.
> 
> Even if you do fix them, it's not just limited to malloc either.
> Anything that uses LWPCTL will be screwed up after vfork.

But what, that uses LWPCTL, are you actually supposed to be doing after
vfork()?  Whack some environment variables and some globals -- maybe -- then
exec*().  What else is legitimate?

If it doesn't slow down vfork, great, fix whatever.  But vfork performance
really is critical.  Look how much slower, for example, FreeBSD builds
the system than we do.  vfork is a huge part of that.

Thor