From: ebiederm@xmission.com (Eric W. Biederman)
To: Hari Bathini <hbathini@linux.vnet.ibm.com>
Cc: ast@fb.com, peterz@infradead.org, lkml <linux-kernel@vger.kernel.org>,
        acme@kernel.org, alexander.shishkin@linux.intel.com, mingo@redhat.com,
        daniel@iogearbox.net, rostedt@goodmis.org,
        Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>,
        sargun@sargun.me, Aravinda Prasad <aravinda@linux.vnet.ibm.com>,
        brendan.d.gregg@gmail.com
References: <148156671980.4651.13017595439842763683.stgit@hbathini.in.ibm.com>
        <148156676528.4651.9323274278676831770.stgit@hbathini.in.ibm.com>
        <871sxdf1ew.fsf@xmission.com>
        <51be4a2e-ebec-9047-8b3b-e011335bdd4d@linux.vnet.ibm.com>
Date: Wed, 14 Dec 2016 08:58:42 +1300
In-Reply-To: <51be4a2e-ebec-9047-8b3b-e011335bdd4d@linux.vnet.ibm.com> (Hari
        Bathini's message of "Wed, 14 Dec 2016 00:17:26 +0530")
Message-ID: <87a8bzfvnx.fsf@xmission.com>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/25.1 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain
Subject: Re: [PATCH v3 1/3] perf: add PERF_RECORD_NAMESPACES to include namespaces related info
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2311
Lines: 68

Hari Bathini <hbathini@linux.vnet.ibm.com> writes:

> Hi Eric,
>
>
> On Monday 12 December 2016 11:57 PM, Eric W. Biederman wrote:
>> Hari Bathini <hbathini@linux.vnet.ibm.com> writes:
>>
>>> With the advert of container technologies like docker, that depend
>>> on namespaces for isolation, there is a need for tracing support for
>>> namespaces. This patch introduces new PERF_RECORD_NAMESPACES event
>>> for tracing based on namespaces related info.
>>> diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
>>> index c66a485..2a48fc6 100644
>>> --- a/include/uapi/linux/perf_event.h
>>> +++ b/include/uapi/linux/perf_event.h
>>> @@ -344,7 +344,8 @@ struct perf_event_attr {
>>>   				use_clockid    :  1, /* use @clockid for time fields */
>>>   				context_switch :  1, /* context switch data */
>>>   				write_backward :  1, /* Write ring buffer from end to beginning */
>>> -				__reserved_1   : 36;
>>> +				namespaces     :  1, /* include namespaces data */
>>> +				__reserved_1   : 35;
>>>     	union {
>>>   		__u32		wakeup_events;	  /* wakeup every n events */
>>> @@ -610,6 +611,18 @@ struct perf_event_header {
>>>   	__u16	size;
>>>   };
>>>   +enum {
>>> +	NET_NS_INDEX		= 0,
>>> +	UTS_NS_INDEX		= 1,
>>> +	IPC_NS_INDEX		= 2,
>>> +	PID_NS_INDEX		= 3,
>>> +	USER_NS_INDEX		= 4,
>>> +	MNT_NS_INDEX		= 5,
>>> +	CGROUP_NS_INDEX		= 6,
>>> +
>>> +	NAMESPACES_MAX,		/* maximum available namespaces */
>>> +};
>>> +
>>>   enum perf_event_type {
>>>     	/*
>>> @@ -862,6 +875,18 @@ enum perf_event_type {
>>>   	 */
>>>   	PERF_RECORD_SWITCH_CPU_WIDE		= 15,
>>>   +	/*
>>> +	 * struct {
>>> +	 *	struct perf_event_header	header;
>>> +	 *
>>> +	 *	u32				pid, tid;
>>> +	 *	u64				dev_num;
>>> +	 *	u64				inode_num[NAMESPACES_MAX];
>> There needs to be one device number per inode.  While it is true that
>> today the device number is always the same.  That is not necessarily so.
>> I reserve the right to have the device number vary per namespace
>> so that I don't need to implement a namespace of namespaces.
>>
>> These are st_dev and st_ino of the inode for the namespace.
>
> Do you mean..
>
>     st_dev = encode_dev(inode->i_sb->s_dev); ?
>     st_ino = inode->i_ino; ?

Yes.  I believe that is how those values make it to user space
during a stat system call.

Eric