[PATCH 0/2] Send a SIGCHLD to the init's pid namespace parent when reboot

[Bruno Prémont]

On Tue, 23 August 2011 Oleg Nesterov <oleg at redhat.com> wrote:
> On 08/23, Greg Kurz wrote:
> >
> > On Tue, 2011-08-23 at 15:33 +0200, Oleg Nesterov wrote:
> > > [...]
> > > At least now I understand why did you mention si_code/CLD before. You
> > > meant waitid(). I thought you were talking about the death-notifications
> > > which can't report CLD_ you need.
> > >
> > > I strongly object. We shouldn't uglify wait_task_zombie() to solve the
> > > very specific problem.
> > >
> > > And once again. sub_init->parent does wiat(&status) and sees
> > > WIFSIGNALED() && WTERMSIG(status) == SIGHUP. This can only mean that
> > > sys_reboot(LINUX_REBOOT_CMD_RESTART) was called. It _can not_ be really
> > > killed by SIGHUP, it must be CMD_RESTART.
> > >
> > > Why this can't work? Why do you want the additional complications?
> > >
> >
> > I don't see either what could go wrong with you approach.
> 
> Thanks ;)
> 
> Just in case... instead of WIFSIGNALED/WTERMSIG we can also report
> the exit code in the upper bits. I mean,
> 
> 	switch (reboot_cmd) {
> 	case LINUX_REBOOT_CMD_RESTART:
> 		code = 1 << 16;
> 		break;
> 	case LINUX_REBOOT_CMD_HALT:
> 		code = 2 << 16;
> 		break;
> 	}

That looks nice and simple!

> this can't be confused with the normal exit(code), just the parent
> should be careful, I am not sure this can't confuse WIFEXITED/WEXITSTATUS
> user-space macroses.

For recent glibc (2.12.2):
sys/wait.h and stdlib.h:
# define WIFEXITED(status) __WIFEXITED (__WAIT_INT (status))
# define WEXITSTATUS(status)       __WEXITSTATUS (__WAIT_INT (status))
bits/waitstatus.h:
#define     __WIFEXITED(status)     (__WTERMSIG(status) == 0)
#define     __WTERMSIG(status)      ((status) & 0x7f)
#define     __WEXITSTATUS(status)   (((status) & 0xff00) >> 8)


For this version of glibc on x86 and amd64 it looks fine.
Would have to look at other versions and other libcs to make sure there
are no surprises.

Bruno

> wait(&status) takes "int *", we have a room for additional info, and
> wait_task_zombie() simply copies exit_code.
> 
> Oleg.
>