linux-kernel - Re: DEFINE_FREE/CLASS && code readability (Was: [PATCH v2 2/6] Provide __free(argv) for argv

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]

Message-Id: <20250106211946.2a1b1351421298150ca8c6bb@kernel.org>
Date: Mon, 6 Jan 2025 21:19:46 +0900
From: Masami Hiramatsu (Google) <mhiramat@...nel.org>
To: Oleg Nesterov <oleg@...hat.com>
Cc: Peter Zijlstra <peterz@...radead.org>, Steven Rostedt
 <rostedt@...dmis.org>, Anil S Keshavamurthy
 <anil.s.keshavamurthy@...el.com>, "David S . Miller" <davem@...emloft.net>,
 Mathieu Desnoyers <mathieu.desnoyers@...icios.com>, Tzvetomir Stoyanov
 <tz.stoyanov@...il.com>, Naveen N Rao <naveen@...nel.org>, Josh Poimboeuf
 <jpoimboe@...nel.org>, Jason Baron <jbaron@...mai.com>, Ard Biesheuvel
 <ardb@...nel.org>, linux-kernel@...r.kernel.org,
 linux-trace-kernel@...r.kernel.org
Subject: Re: DEFINE_FREE/CLASS && code readability (Was: [PATCH v2 2/6]
 Provide __free(argv) for argv_split() users)

On Sun, 5 Jan 2025 15:14:22 +0100
Oleg Nesterov <oleg@...hat.com> wrote:

> Masami,
> 
> Sorry for abusing this thread. Your patches look fine to me, it is not
> that I suggest to change them. I will use your patch as an example for
> off-topic discussion.
> 
> On 01/05, Masami Hiramatsu (Google) wrote:
> >
> > +DEFINE_FREE(argv, char **, if (!IS_ERR_OR_NULL(_T)) argv_free(_T))
> 
> (IS_ERR looks unneeded but this is cosmetic).
> 
> OK, so it can be used as
> 
> 	void func(void)
> 	{
> 		char **argv __free(argv) = argv_split(...);
> 		do_something(argv);
> 		return;
> 	}
> 
> And I cry every time when I read the code like this ;)
> 
> Because, to understand this code, I need to do the "nontrivial" grep to find
> "DEFINE_FREE(argv,".
> 
> Perhaps we can establish a simple rule that every DEFINE_FREE() or DEFINE_CLASS()
> should add another #define? I mean something like
> 
> 
> 	DEFINE_FREE(argv, char **, if (!IS_ERR_OR_NULL(_T)) argv_free(_T))
> 	#define __FREE_ARGV	__free(argv)
> 	
> 	void func(void)
> 	{
> 		char **argv __FREE_ARGV = argv_split(...);
> 		do_something(argv);
> 		return;
> 	}
> 
> This way I can press Ctrl-] and see what the cleanup code actually does.
> Can save a second or two. Important when you try to read the code you are
> not familiar with.

That sounds lile a problem of your tool. Do you really need to find the
DEFINE_FREE() or do you think "__free(argv)" is too generic name?
If it is latter, we can make it "__free(argv_free) so that it is more
obvious to call argv_free()?

> 
> Same for DEFINE_CLASS. For example,
> 
> 	int ksys_fchown(unsigned int fd, uid_t user, gid_t group)
> 	{
> 		CLASS(fd, f)(fd);
> 
> 		if (fd_empty(f))
> 			return -EBADF;
> 
> 		return vfs_fchown(fd_file(f), user, group);
> 	}
> 
> If you are not familiar with this code, it looks mysterious until you find
> DEFINE_CLASS(fd, ...) in include/linux/file.h.

DEFINE_CLASS() is somewhat mysterious to me too :) But if I understand
correctly, it is for intermediate macro for implementing guard().

Whether we like it or not, cleanup.h has been introduced, and it will be
more popular. What we need is a document about cleanup.h which includes
better naming conventions for its label.

BTW, I agree that 'argv' was too simple. Basically the label name of
DEFINE_FREE() is better to be a function name for free.
Let me fix that.

Thank you,

-- 
Masami Hiramatsu (Google) <mhiramat@...nel.org>