[PATCH 0/7][v7] Container-init signal semantics

[Sukadev Bhattiprolu]

KAMEZAWA Hiroyuki [kamezawa.hiroyu at jp.fujitsu.com] wrote:
| On Sat, 17 Jan 2009 12:26:38 -0800
| Sukadev Bhattiprolu <sukadev at linux.vnet.ibm.com> wrote:
| 
| > 
| > Container-init must behave like global-init to processes within the
| > container and hence it must be immune to unhandled fatal signals from
| > within the container (i.e SIG_DFL signals that terminate the process).
| > 
| > But the same container-init must behave like a normal process to 
| > processes in ancestor namespaces and so if it receives the same fatal
| > signal from a process in ancestor namespace, the signal must be
| > processed.
| > 
| > Implementing these semantics requires that send_signal() determine pid
| > namespace of the sender but since signals can originate from workqueues/
| > interrupt-handlers, determining pid namespace of sender may not always
| > be possible or safe.
| > 
| 
| Is this feature is for blocking signals from children to name-space
| creator(owner) ?  And automatically used when namespace/cgroup is created ?
| IOW, Container-init is Namespace-Cgroup-init ? 

I am not sure what "Namespace-cgroup-init refers" to.

But, yes, this patchset applies to the first process in a pid namespace
i.e the child of clone(NEWPID) call.

| 
| I'm glad if you add some documentation updates about how-it-works to patch set.

Yes,  when the patchset is accepted, I am planning to add some notes to
/sbin/init man page. 

Thanks,

Sukadev