[Devel] Re: [PATCH 7/15] Helpers to obtain pid numbers

[Pavel Emelyanov]

Oleg Nesterov wrote:
> On 07/26, Pavel Emelyanov wrote:
>> --- linux-2.6.23-rc1-mm1.orig/include/linux/pid.h	2007-07-26 16:34:45.000000000 +0400
>> +++ linux-2.6.23-rc1-mm1-7/include/linux/pid.h	2007-07-26 16:36:37.000000000 +0400
>> @@ -83,12 +92,34 @@ extern void FASTCALL(detach_pid(struct t
>>
>>  extern struct pid *alloc_pid(struct pid_namespace *ns);
>>  extern void FASTCALL(free_pid(struct pid *pid));
>> +
>> +/*
>> + * the helpers to get the pid's id seen from different namespaces
>> + *
>> + * pid_nr()    : global id, i.e. the id seen from the init namespace;
> 
> This looks a bit strange to me, but perhaps this is just matter of taste.
> I think pid_nr(pid) should be pid_nr_ns(pid, current->nsproxy->pid_ns),
> this is imho much closer to the current meaning. I won't persist though.

pid_nr, find_task_by_pid and all other stuff, that existed in the kernel
before the set are intended to work with global ids only (just as it was
before the set).

>> +pid_t pid_nr_ns(struct pid *pid, struct pid_namespace *ns)
>> +{
>> +	pid_t nr = 0;
>> +	if (pid && ns->level <= pid->level)
>> +		nr = pid->numbers[ns->level].nr;
>> +	return nr;
>> +}
> 
> I am not sure I understand the "ns->level <= pid->level" check. Isn't it
> a bug to use a "wrong" namespace here? In that case BUG_ON() looks better.

Yes, that's a check for bad namespace passed.

> If ns could be wrong, "ns->level <= pid->level" is not enough, we should
> also check "pid->numbers[ns->level].ns == ns", no?

Yes, we should, and you're right in that we must have BUG_ON() here.

> Oleg.
> 
> 

Thanks,
Pavel