Re: [PATCH] Workaround a FreeBSD ptrace() bug with clearing thread events.

[John Baldwin <jhb@freebsd.org>]

On Friday, March 02, 2018 12:13:20 AM Maciej W. Rozycki wrote:
> On Fri, 23 Feb 2018, John Baldwin wrote:
> 
> > diff --git a/gdb/fbsd-nat.c b/gdb/fbsd-nat.c
> > index d44950618c..9c87bfed33 100644
> > --- a/gdb/fbsd-nat.c
> > +++ b/gdb/fbsd-nat.c
> > @@ -1163,6 +1163,39 @@ fbsd_resume (struct target_ops *ops,
> >  	}
> >        ptid = inferior_ptid;
> >      }
> > +
> > +#if __FreeBSD_version < 1200052
> > +  /*
> > +   * When multiple threads within a process wish to report STOPPED
> > +   * events from wait(), the kernel picks one thread event as the
> > +   * thread event to report.  The chosen thread event is retrieved via
> > +   * PT_LWPINFO by passing the process ID as the request pid.  If
> > +   * multiple events are pending, then the subsequent wait() after
> > +   * resuming a process will report another STOPPED event after
> > +   * resuming the process to handle the next thread event and so on.
> > +   *
> > +   * A single thread event is cleared as a side effect of resuming the
> > +   * process with PT_CONTINUE, PT_STEP, etc.  In older kernels,
> > +   * however, the request pid was used to select which thread's event
> > +   * was cleared rather than always clearing the event that was just
> > +   * reported.  To avoid clearing the event of the wrong LWP, always
> > +   * pass the process ID instead of an LWP ID to PT_CONTINUE or
> > +   * PT_SYSCALL.
> 
>  Hmm, doesn't it have to be a run-time check then?  Otherwise you're 
> basing your decision on the host system GDB has been built for and not one 
> it will be run on, which I suppose does not necessarily have to be of the 
> same version.  Or am I missing anything here?

FreeBSD generally does not support forwards-compatability for binaries (newer
binary on older kernel), only backwards-compatability (older binary on newer
kernel).  In this case, using the workaround is also fine on a fixed kernel,
so it doesn't hurt if GDB is compiled on an older system (thus using the
workaround) and then run under a newer kernel.

-- 
John Baldwin