[PATCH 4/7][v7] Protect cinit from unblocked SIG_DFL signals

Sukadev Bhattiprolu sukadev at linux.vnet.ibm.com
Mon Jan 19 17:07:34 PST 2009


Oleg Nesterov [oleg at redhat.com] wrote:
| On 01/17, Sukadev Bhattiprolu wrote:
| >
| > @@ -1331,7 +1341,7 @@ int send_sigqueue(struct sigqueue *q, struct task_struct *t, int group)
| >  		goto ret;
| >
| >  	ret = 1; /* the signal is ignored */
| > -	if (!prepare_signal(sig, t))
| > +	if (!prepare_signal(sig, t, 1))
| 
| Hmm, just noticed. This looks wrong, it should be prepare_signal(sig, t, 0),
| no?

Grr. I missed that when we changed the parameter from 'same-ns' to
'from_ancestor_ns' and inverted the logic.

| 
| For example, /sbin/init can create the posix timer with sigev_signo = SIGKILL
| and it won't be killed before this patch.
| 
| This also looks wrong from the masquerading pov.
| 
| 
| Otherwise, the patches 1-6 are imho fine.

Thanks, Will resend this patch.

Sukadev


More information about the Containers mailing list