From: Willy Tarreau <willy@w.ods.org>
To: Robert Love <rml@novell.com>
Cc: Russell King <rmk+lkml@arm.linux.org.uk>,
Linux Kernel List <linux-kernel@vger.kernel.org>,
Linus Torvalds <torvalds@osdl.org>,
Al Viro <viro@ftp.linux.org.uk>
Subject: Re: p = kmalloc(sizeof(*p), )
Date: Sun, 18 Sep 2005 18:52:19 +0200 [thread overview]
Message-ID: <20050918165219.GA595@alpha.home.local> (raw)
In-Reply-To: <1127061146.6939.6.camel@phantasy>
On Sun, Sep 18, 2005 at 12:32:26PM -0400, Robert Love wrote:
> On Sun, 2005-09-18 at 11:06 +0100, Russell King wrote:
>
> > +The preferred form for passing a size of a struct is the following:
> > +
> > + p = kmalloc(sizeof(*p), ...);
> > +
> > +The alternative form where struct name is spelled out hurts readability and
> > +introduces an opportunity for a bug when the pointer variable type is changed
> > +but the corresponding sizeof that is passed to a memory allocator is not.
>
> Agreed.
>
> Also, after Alan's #4:
>
> 5. Contrary to the above statement, such coding style does not help,
> but in fact hurts, readability. How on Earth is sizeof(*p) more
> readable and information-rich than sizeof(struct foo)? It looks
> like the remains of a 5,000 year old wolverine's spleen and
> conveys no information about the type of the object that is being
> created.
>
> Robert Love
To be honnest, before reading this thread, I would have voted for the
sizeof(*p). However, I completely agree that there is a high risk of
messing up the initialization, and that structures don't change often.
The situations where I think that sizeof(*p) is better than
sizeof(struct foo) is more on functions such as memset() than {,k}malloc() :
forgetting to initialize a struct member is always a high risk, but if the
object is not a struct (eg, a scalar), then it could be tolerated. I don't
know anybody who does kmalloc(sizeof(int)) nor kmalloc(sizeof(char)), but
with memset, it's different. Doing memset(p, 0, sizeof(*p)) seems better
to me than memset(p, 0, sizeof(short)), and represents a smaller risk
when 'p' will silently evolve to a long int.
Last, there's little probability that a scalar will evolve into a struct
without code modifications, while it has happened often that a __u8 or
__u16 was changed to __u32. So perhaps we could accept use of sizeof(*p)
when (*p) is a scalar to protect against silent type changes, and reject
it when (*p) is a structure to avoid incomplete initialization ?
Alan, I like your proposal BTW ;-)
Regards,
Willy
next prev parent reply other threads:[~2005-09-18 16:52 UTC|newest]
Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-09-18 10:06 p = kmalloc(sizeof(*p), ) Russell King
2005-09-18 11:04 ` Alan Cox
2005-09-18 14:39 ` Al Viro
2005-09-18 16:25 ` Denis Vlasenko
2005-09-18 17:30 ` Al Viro
2005-09-18 18:00 ` Willy Tarreau
2005-09-18 17:47 ` Alan Cox
2005-09-18 16:32 ` Robert Love
2005-09-18 16:52 ` Willy Tarreau [this message]
2005-09-18 17:18 ` Al Viro
2005-09-18 17:31 ` Linus Torvalds
2005-09-18 17:45 ` Al Viro
2005-09-18 20:34 ` Roman Zippel
2005-09-18 21:12 ` Al Viro
2005-09-18 21:52 ` Al Viro
2005-09-18 22:25 ` Linus Torvalds
2005-09-18 23:07 ` Al Viro
2005-09-20 6:31 ` Richard Henderson
2005-09-19 21:20 ` Matthias Urlichs
2005-09-19 21:28 ` Matthias Urlichs
2005-09-18 19:07 ` Al Viro
2005-09-18 21:30 ` Alan Cox
2005-09-18 21:14 ` Al Viro
2005-09-19 6:09 ` Coywolf Qi Hunt
2005-09-21 2:18 ` Miles Bader
2005-09-18 17:32 ` Randy.Dunlap
2005-09-19 6:47 ` Coywolf Qi Hunt
2005-09-20 8:53 ` Pekka Enberg
2005-09-20 9:39 ` Al Viro
2005-09-20 9:47 ` Pekka J Enberg
2005-09-20 9:53 ` Al Viro
2005-09-20 10:07 ` Pekka J Enberg
2005-09-20 15:14 ` Randy.Dunlap
2005-09-20 11:18 ` Pekka Enberg
2005-09-20 11:40 ` Russell King
2005-09-20 11:56 ` Denis Vlasenko
2005-09-20 12:20 ` Pekka J Enberg
2005-09-20 12:31 ` Russell King
2005-09-20 12:35 ` Pekka J Enberg
2005-09-20 15:21 ` Randy.Dunlap
2005-09-20 12:53 ` Pekka J Enberg
2005-09-20 17:11 ` Andrew Morton
2005-09-20 17:17 ` Russell King
2005-09-20 18:02 ` Alan Cox
2005-09-20 17:59 ` Andrew Morton
2005-09-20 18:11 ` Russell King
2005-09-20 18:41 ` Jeff Garzik
2005-09-20 20:41 ` Alan Cox
2005-09-20 19:41 ` Horst von Brand
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20050918165219.GA595@alpha.home.local \
--to=willy@w.ods.org \
--cc=linux-kernel@vger.kernel.org \
--cc=rmk+lkml@arm.linux.org.uk \
--cc=rml@novell.com \
--cc=torvalds@osdl.org \
--cc=viro@ftp.linux.org.uk \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).