On Sat, Apr 28, 2018 at 11:28:50AM +0200, Otto Moerbeek wrote:
> On Sat, Apr 28, 2018 at 10:40:01AM +0200, Philip Guenther wrote:
>
> > On Sat, 28 Apr 2018, Otto Moerbeek wrote:
> > > On Fri, Apr 27, 2018 at 09:12:36PM -0400, Rupert Gallagher wrote:
> > >
> > > > The lack of information originates from "mountd -d". When it terminates
> > > > because of an error, it should log the name of the last function and
> > > > its parameters.
> > >
> > > Wrap your lines.
> > >
> > > Well, ktrace does that for system calls. What I'm seeing is that a
> > > socket is closed and mountd tries to write to it and then gets
> > > SIGPIPE because of that. But I do not know yet what's cause and effect.
> >
> > Well, since it's handling TCP connections mountd should be ignoring
> > SIGPIPE, no?
>
> Yes, I was pondering that already.
>
> Worth a try (I cannot reproduce, not having Windows 10 Pro),
Realizing my daughter has an unused license to it, I upgraded the only
windows machine in this house to Windows 10 Education (which is almost
equivalent to Pro). I can now mount NFS volumes on Windows, but no
crashes so far, even when exercising the browse button in the dialog
muiltiple times.
So stil not able to reproduce. The windows machine has all the patches
up until today.
-Otto
>
> >
> > Index: mountd.c
> > ===================================================================
> > RCS file: /data/src/openbsd/src/sbin/mountd/mountd.c,v
> > retrieving revision 1.85
> > diff -u -p -r1.85 mountd.c
> > --- mountd.c 23 Dec 2015 21:32:52 -0000 1.85
> > +++ mountd.c 28 Apr 2018 08:39:46 -0000
> > @@ -295,6 +295,7 @@ main(int argc, char *argv[])
> >
> > signal(SIGCHLD, (void (*)(int)) check_child);
> > signal(SIGHUP, (void (*)(int)) new_exportlist);
> > + signal(SIGPIPE, SIG_IGN);
> >
> > if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, socks) == -1) {
> > syslog(LOG_ERR, "socketpair: %m");
