Re: [PATCH] Possibly correct fix to strace phantom process entry

[Corinna Vinschen <corinna-cygwin at cygwin dot com>]

On Apr 24 14:19, Corinna Vinschen wrote:
> Hi Daniel,
> 
> On Apr 24 04:37, Daniel Santos wrote:
> > The root cause of problem with strace causing long delays when any
> > process enumerates the process database appears to be calling
> > myself.thisproc () from child_info_spawn::handle_spawn() when we've
> > dynamically loaded cygwin1.dll.  It definately fixes the problem, but I
> > don't konw what other processes dynamically load cygwin1.dll and, thus,
> > what other side-effects that this may have.  Please verify correctness.
> > 
> > Please see discussion here: https://cygwin.com/ml/cygwin/2017-04/msg00240.html
> > 
> > Daniel
> > 
> > Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
> > ---
> 
> I was just looking into this myself, but I was looking into the weird
> Sleep loop itself and was wondering if that makes any sense at all.
> 
> Assuming pinfo::init is called from process enumeration (winpids::add)
> then there's no good reason to handle this procinfo block at all.  It
> doesn't resolve into an existing "real" Cygwin process.  And that's
> exactly the case that hangs.
> 
> So my curent fix would have been this:

I'm going with my patch for now.  Mainly because I added some debug
output to see if we need the Sleep loop at all.  Right now I don't see
any situation which would qualify for this.


Thanks,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

Attachment: signature.asc
Description: PGP signature