[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <C3B45B71-C7D1-45EB-B749-39514A49C521@gmail.com>
Date: Sun, 25 Aug 2024 22:37:06 +0800
From: zhang warden <zhangwarden@...il.com>
To: Christoph Hellwig <hch@...radead.org>
Cc: Josh Poimboeuf <jpoimboe@...nel.org>,
Miroslav Benes <mbenes@...e.cz>,
jikos@...nel.org,
pmladek@...e.com,
joe.lawrence@...hat.com,
live-patching@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3 1/2] Introduce klp_ops into klp_func structure
> On Aug 25, 2024, at 12:48, Christoph Hellwig <hch@...radead.org> wrote:
>
> On Thu, Aug 22, 2024 at 11:01:58AM +0800, Wardenjohn wrote:
>> 1. Move klp_ops into klp_func structure.
>> Rewrite the logic of klp_find_ops and
>> other logic to get klp_ops of a function.
>>
>> 2. Move definition of struct klp_ops into
>> include/linux/livepatch.h
>
> Why?
>
Hi, Christoph.
When introducing feature of "using", we should handle the klp_ops check.
In order to get klp_ops from transition, we may have more complex logic for
checking.
If we move klp_ops into klp_func. We can remove the global list head of klp_ops.
What's more, there are some code in livepatch should get function's ops.
With this feature, we don't need the original search to the one ops.
It can be simple and straightforward.
Regards,
Wardenjohn
Powered by blists - more mailing lists