Pierre Thierry, le Thu 25 Jul 2013 15:53:41 +0200, a écrit :
> On Mon, Jul 22, 2013 at 10:11:32AM +0200, Samuel Thibault wrote:
> > Mmm, but we also want to check that the user can read all components,
> > otherwise we may be revealing directory entries which the user was not
> > supposed to find o
Justus Winter, le Thu 25 Jul 2013 22:41:09 +0200, a écrit :
> - err = proc_getmsgport (procserver, 1, &startup);
> + err = proc_getmsgport (procserver, 2, &startup);
Maybe make it a #define, so it'll be easier to change next time? :)
Samuel
On Mon, Jul 22, 2013 at 10:11:32AM +0200, Samuel Thibault wrote:
> Mmm, but we also want to check that the user can read all components,
> otherwise we may be revealing directory entries which the user was not
> supposed to find out.
That's not the behaviour on Linux and I've always found it desir
---
exec/main.c |2 +-
hurd/process.defs|7 +++
init/Makefile|2 +-
init/init.c |8 ++--
libdiskfs/boot-start.c |2 +-
libdiskfs/init-startup.c |2 +-
pfinet/main.c|2 +-
proc/main.c |6
Hi,
attached is a patch that reserves pid 1 for whatever process
/hurd/init starts. runsystem.sysv for example execs /sbin/init, so
sysvinit is happily running as pid 1.
/hurd/init get's to be pid 2, so everyone who needs shutdown
notification now has to talk to pid 2. The kernel and other server
