[RFC,PATCH] fix /sbin/init signal handling

Oleg Nesterov oleg at tv-sign.ru
Tue Aug 21 03:30:02 PDT 2007


On 08/21, sukadev at us.ibm.com wrote:
> 
> I am still reviewing this patch and will try to plug in the multiple
> pid ns code and play with it some more in the next couple of days.

Thanks!

> But am curious why we need the in_interrupt() check and that too only
> for the container-init process.

For example, send_sigio(). We shouldn't send this signal to /sbin/init.
But what if the interrupted process is from the parent namespace?

> Also, maybe a dumb que, are the checks for SIG_IGN and SIG_DFL required
> both on sender side and the receiver side (get_signal_to_deliver())

get_signal_to_deliver() checks SIG_IGN/SIG_DFL, but I guess you meant
a special check for init.

Unless the patch is wrong, this is not needed. And, in fact we can't do
this anyway. Because get_signal_to_deliver() can't know if the signal
comes from parent namespace or not.

Oleg.



More information about the Containers mailing list