linux-kernel - Re: [PATCH] seq_file: Optimize seq

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: <20240417010430.GB2118490@ZenIV>
Date: Wed, 17 Apr 2024 02:04:30 +0100
From: Al Viro <viro@...iv.linux.org.uk>
To: David Laight <David.Laight@...lab.com>
Cc: Christophe JAILLET <christophe.jaillet@...adoo.fr>,
	Christian Brauner <brauner@...nel.org>, Jan Kara <jack@...e.cz>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"kernel-janitors@...r.kernel.org" <kernel-janitors@...r.kernel.org>,
	"linux-fsdevel@...r.kernel.org" <linux-fsdevel@...r.kernel.org>
Subject: Re: [PATCH] seq_file: Optimize seq_puts()

On Tue, Apr 16, 2024 at 08:56:51PM +0000, David Laight wrote:

> > static inline void seq_puts(struct seq_file *m, const char *s)
> 
> That probably needs to be 'always_inline'.

What for?  If compiler fails to inline it (and I'd be very surprised
if that happened - if s is not a constant string, we get a straight call
of __seq_puts() and for constant strings it boils down to call of
seq_putc(m, constant) or seq_write(m, s, constant)), nothing bad
would happen; we'd still get correct behaviour.

> > {
> >	if (!__builtin_constant_p(*s))
> > 		__seq_puts(m, s);
> > 	else if (s[0] && !s[1])
> > 		seq_putc(m, s[0]);
> > 	else
> > 		seq_write(m, s, __builtin_strlen(s));
> > }
> 
> You missed seq_puts(m, "");

Where have you seen one?  And if it gets less than optimal, who cares?

> Could you do:
> 	size_t len = __builtin_strlen(s);
> 	if (!__builtin_constant_p(len))
> 		__seq_puts(m, s);
> 	else switch (len){
> 	case 0: break;
> 	case 1: seq_putc(m, s[0]);
> 	default: seq_write(m, s, len);
> 	}

Umm...  That's probably OK, but I wonder how useful would that
be...