Re: [RFC][PATCH 3/7][v4] Define siginfo_from_ancestor_ns()

[Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>]

Oleg Nesterov [oleg@redhat.com] wrote:
| Small nit... siginfo_from_user() is only called by siginfo_from_ancestor_ns().
| The first helper depends on CONFIG_PID_NS, the second is not. A bit strange.

| 
| Isn't it cleaner to do
| 
| 	#ifdef CONFIG_PID_NS
| 	static inline int siginfo_from_user(siginfo_t *info)
| 	{
| 		...
| 	}
| 	static inline int siginfo_from_ancestor_ns(...)
| 	{
| 		...
| 	}
| 	#else
| 	static inline int siginfo_from_ancestor_ns(...)
| 	{
| 		return 0;
| 	}
| 	#endif
| 
| ?

Yes, it was that way in the earlier version, but I thought we introduced
CONFIG_PID_NS only to hide the ugliness resulting from pid-ns. Ok. I
will revert.

| 
| > +#ifdef CONFIG_PID_NS
| > +/*
| > + * siginfo_from_user() assumes that si_code SI_ASYNCIO comes only from
| > + * within the kernel. If an application is passing in SI_ASYNCIO we 
| > + * want to know about it.
| > + */
| > +static void warn_on_asyncio(siginfo_t *info)
| > +{
| > +	WARN_ON_ONCE(info->si_code == SI_ASYNCIO);
| > +}
| > +#else
| > +#define warn_on_asyncio(info)	{}
| > +#endif
| > +
| >  asmlinkage long
| >  sys_rt_sigqueueinfo(pid_t pid, int sig, siginfo_t __user *uinfo)
| >  {
| > @@ -2324,6 +2388,9 @@ sys_rt_sigqueueinfo(pid_t pid, int sig, siginfo_t __user *uinfo)
| >  	   Nor can they impersonate a kill(), which adds source info.  */
| >  	if (info.si_code >= 0)
| >  		return -EPERM;
| > +
| > +	warn_on_asyncio(&info);
| 
| Hmm... why do you want this? The user-space can use any si_code >= 0,
| why should we uglify the code?

I thought losing a SIGKILL, however twisted the path, was serious enough
to justify the ugliness. Again, I am not particular.

| 
| And, SI_ASYNCIO only matters when we send the signal to the subnamespace,
| and in that case we will probably mangle .si_pid. So why don't we warn
| when .si_code == SI_USER?

I was wondering if I should there too :-) But what do you think ?