Re: ifupdown: Add support for GNU/Hurd (for review)
Svante Signell, le Mon 21 May 2012 23:28:12 +0200, a écrit :
> On Mon, 2012-05-21 at 23:15 +0200, Samuel Thibault wrote:
> > Svante Signell, le Mon 21 May 2012 23:11:07 +0200, a écrit :
> > > On Mon, 2012-05-21 at 21:09 +0200, Samuel Thibault wrote:
> > > > You are looking for "non invasive", not intrusive.
> > > >
> > > > ./hurd/debugging/gdb/noninvasive_debugging.mdwn
> > > >
> > > > in the wiki, pointing at
> > > >
> > > > http://sourceware.org/gdb/current/onlinedocs/gdb_19.html#Hurd%20Native
> > > >
> > > > That said, for inspecting a crash usually you don't need that. Just
> > > > attach, let it continue, and on crash you get control again.
> > > >
> > > Does the pfinet process respawn after crashing, it has the same PID
> > > before as well as after:
> > > inetutils-ifconfig --interface lo --up
> > > inetutils-ifconfig: SIOCSIFFLAGS failed: Computer bought the farm
> >
> > Ah??
> >
> > Well, yes, pfinet respawns, but it's supposed to get another PID. I
> > wonder which port ifconfig gets EIEIO from, then. An rpctrace might
> > telle.
>
> Attached! Note the file contains non-printable characters.
>
> 57->dir_lookup ("servers/socket/2" 0 0) = 0 1 "" 73
> 73->socket_create (1 0) = 0 75
> 75->iioctl_siocgifflags ("lo" 0) = 0 "lo" 73
> 75->iioctl_siocsifflags ("lo" 73) = 0x10000003 ((ipc/send) invalid destination port)
> inetutils-ifconfig: 71->io_write_request ("inetutils-ifconfig: " -1) = 0 20
> SIOCSIFFLAGS failed 71->io_write_request ("SIOCSIFFLAGS failed" -1) = 0 19
> : Computer bought the farm 71->io_write_request (": Computer bought the farm" -1) = 0 26
Ok, so it's indeed not crashing at all, the 0x10000003 error
(EMACH_SEND_INVALID_DEST) is converted by glibc into EIEIO.
To investigate further, you can put printfs in pfinet in the
SIOCSIFFLAGS handling code, to check there what returns this error
(especially since here no flag is actually changed...)
Samuel
Reply to: