Re: [Xen-devel] [PATCH 5/7] libxl: fork: Provide libxl_childproc_sigchld_occurred

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Fri, 2014-01-17 at 11:27 +0000, Ian Jackson wrote:
> Ian Campbell writes ("Re: [PATCH 5/7] libxl: fork: Provide 
> libxl_childproc_sigchld_occurred"):
> > On Thu, 2014-01-16 at 17:22 +0000, Ian Jackson wrote:
> > > +            /* it must have finished but we don't know its status */
> > > +            status = 255<<8; /* no wait.h macro for this! */
> > > +            assert(WIFEXITED(status));
> > > +            assert(WEXITSTATUS(status)==255);
> > > +            assert(!WIFSIGNALED(status));
> > > +            assert(!WIFSTOPPED(status));
> > 
> > This is quite exciting! How can this happen? Kernel bug or similar?
> 
> In principle it would be possible (and standards-compliant) for a
> system to encode its wait status in a different way to usual.  I don't
> think any such system really exists - at least, not one we'll be
> running libxl on.
> 
> Of course this code is only reached if DISASTER returns...

Sorry, I meant how can we get here at all, i.e. with a waitpid returning
-1/ECHLD when we think there is actually a child.

Thinking about it this way I now realise this could be a bug in libxl or
the application which caused the process to be reaped elsewhere.

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel