linux-kernel - Re: [PATCH] linux/kernel.h: add container

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 28 Aug 2020 12:37:11 +0530
From:   Allen <allen.lkml@...il.com>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Kees Cook <keescook@...omium.org>,
        James Bottomley <James.Bottomley@...senpartnership.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jens Axboe <axboe@...nel.dk>
Subject: Re: [PATCH] linux/kernel.h: add container_from()

>
> I don't see that kind of redundancy being a _problem_, though. "So
> much redundancy" is just over-stating the issue completely.
>
> In fact, we often encourage people to split declaration from
> initialization exactly because it results in simpler expressions and
> more legible code, even if that name is now redundant. So it's a small
> extra typing of the type. Big deal.
>
> The above is also a common pattern that once you know how
> container_of() works, it's very legible.
>
> Sure, if you're new to the kernel, and haven't seen "container_of()"
> in other projects, it might initially be a bit of an odd pattern, but
> that's the advantage of having one single standardized model: it
> becomes a pattern, and you don't have to think about it.
>
> And particularly with that argument-type pattern, you really have to
> work at making over-long lines, since the indentation level will by
> definition be just one.
>
> Looking around, I do see a lot of people doing line-breaks, but that
> tends to be when they insist on putting the variable initialization in
> the declaration. And even then, it often seems pointless (eg
>
>         struct idp_led *led = container_of(cdev,
>                         struct idp_led, cdev);
>
> was split for no good reason I can see, but it seems to be a pattern
> in that file).
>
> You really have to pick some pretty excessive type names (or variable
> names) to get close to 80 characters. Again, to pick an example:
>
>         struct timer_group_priv *priv = container_of(handle,
>                         struct timer_group_priv, timer[handle->num]);
>
> ends up being long even if you were to split it, but that funky
> container_from() wouldn't have helped the real problem - the fact that
> the above is complex and nasty.
>

 An example with a really long member name is

+struct nokia_modem_device *modem = from_tasklet(modem, t,
+        nokia_modem_rst_ind_tasklet);

With container_of() one can imagine how long it would end up. And
am sure we have many more examples in the kernel.

I agree, It would have been simpler to use container_of() as it's been
widely used, but as mentioned by Kees, for 1-to-many conversions
it does not work well.

I guess container_from() is a NAK, if you would want us to just use
container_of() to keep the code clean and simple instead of using wrappers,
or any other method, we are open to suggestions.

Thanks,
- Allen