Re: [PATCH v3 net-next 01/23] lib: add reference counting tracking infrastructure

From: Andrzej Hajda <andrzej.hajda@intel.com>
To: Jakub Kicinski <kuba@kernel.org>
Cc: Eric Dumazet <eric.dumazet@gmail.com>,
	"David S . Miller" <davem@davemloft.net>,
	netdev <netdev@vger.kernel.org>,
	Eric Dumazet <edumazet@google.com>,
	Dmitry Vyukov <dvyukov@google.com>,
	Thomas Gleixner <tglx@linutronix.de>, Greg KH <greg@kroah.com>,
	Ingo Molnar <mingo@redhat.com>
Subject: Re: [PATCH v3 net-next 01/23] lib: add reference counting tracking infrastructure
Date: Wed, 8 Dec 2021 16:11:03 +0100	[thread overview]
Message-ID: <1c7a4f98-b02c-ac68-9873-ab6449f7301f@intel.com> (raw)
In-Reply-To: <20211208065943.61b6220b@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>


On 08.12.2021 15:59, Jakub Kicinski wrote:
> On Wed, 8 Dec 2021 15:09:17 +0100 Andrzej Hajda wrote:
>> On 05.12.2021 05:21, Eric Dumazet wrote:
>>> From: Eric Dumazet <edumazet@google.com>
>>>
>>> It can be hard to track where references are taken and released.
>>>
>>> In networking, we have annoying issues at device or netns dismantles,
>>> and we had various proposals to ease root causing them.
>>>
>>> This patch adds new infrastructure pairing refcount increases
>>> and decreases. This will self document code, because programmers
>>> will have to associate increments/decrements.
>>>
>>> This is controled by CONFIG_REF_TRACKER which can be selected
>>> by users of this feature.
>>>
>>> This adds both cpu and memory costs, and thus should probably be
>>> used with care.
>>>
>>> Signed-off-by: Eric Dumazet <edumazet@google.com>
>>> Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
>> Life is surprising, I was working on my own framework, solving the same
>> issue, with intention to publish it in few days :)
>>
>> My approach was little different:
>>
>> 1. Instead of creating separate framework I have extended debug_objects.
>>
>> 2. There were no additional fields in refcounted object and trackers -
>> infrastructure of debug_objects was reused - debug_objects tracked both
>> pointers of refcounted object and its users.
>>
>>
>> Have you considered using debug_object? it seems to be good place to put
>> it there, I am not sure about performance differences.
> Something along these lines?
>
> https://lore.kernel.org/all/20211117174723.2305681-1-kuba@kernel.org/
>
> ;)


Ups, I was not clear in my last sentence, I meant "extend debug_objects 
with tracking users of the object" :)


Regards

Andrzej


>
>> One more thing about design - as I understand CONFIG_REF_TRACKER turns
>> on all trackers in whole kernel, have you considered possibility/helpers
>> to enable/disable tracking per class of objects?