linux-kernel - Re: [RFC] timers, pointers to functions and type safety

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:	Sun, 3 Dec 2006 00:06:40 +0100 (CET)
From:	Roman Zippel <zippel@...ux-m68k.org>
To:	Al Viro <viro@....linux.org.uk>
cc:	Thomas Gleixner <tglx@...utronix.de>,
	Matthew Wilcox <matthew@....cx>,
	Linus Torvalds <torvalds@...l.org>, linux-arch@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [RFC] timers, pointers to functions and type safety

Hi,

On Sat, 2 Dec 2006, Al Viro wrote:

> > > > You need some more magic macros to access/modify the data field.
> > > 
> > > Which is done bloody rarely.  grep and you'll see...  BTW, there are
> > > other reasons why passing struct timer_list * is wrong:
> > > 	* direct calls of the timer callback
> > 
> > Why should that be wrong?
> 
> Need to arrange a struct timer_list?

How is that different from arranging some struct foo? Either you need a 
reference object anyway or a NULL pointer works in both cases.

> > > 	* see a timer callback, decide it looks better as a tasklet.
> > > What, need a different glue now?
> > 
> > What's wrong with changing the prototype? If you don't do it, the compiler 
> > will complain about it anyway.
> 
> How about "not having to change it at all"?

If it were a common operation, this might have some value...

> > > Look, it's a delayed call.  The less glue we need, the better - the
> > > rules are much simpler that way, so that alone means that we'll get
> > > fewer fsckups.
> > 
> > You have the glue in a different place, so what?
> 
> Where?

In SETUP_TIMER.

> > The other alternative has real _practical_ value in almost every case, 
> > which I very much prefer. What's wrong with that?
> 
> Lack of any type safety whatsoever, magic boilerplates in callback instances,
> rules more complex than "your callback should take a pointer, don't cast
> anything, it's just a way to arrange for a delayed call, nothing magical
> needed"?

I admit the compile check in SETUP_TIMER() is clever, but this way all the
magic is in this place and is it really worth it? You're only adding _one_ 
extra typecheck for mostly simple cases anyway.
The container_of() method is not new and is already well used in other 
places, it has real practical advantages, which would make the change 
actually worthwile.

bye, Roman
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/