linux-kernel - Re: [PATCH] vsprintf: Drop unused assignment of fmt.state

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: <Z6n6iWwnpA6wuUMD@pathway.suse.cz>
Date: Mon, 10 Feb 2025 14:09:53 +0100
From: Petr Mladek <pmladek@...e.com>
To: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
Cc: I Hsin Cheng <richard120310@...il.com>, rostedt@...dmis.org,
	linux@...musvillemoes.dk, senozhatsky@...omium.org,
	akpm@...ux-foundation.org, linux-kernel@...r.kernel.org,
	jserv@...s.ncku.edu.tw, shuah@...nel.org
Subject: Re: [PATCH] vsprintf: Drop unused assignment of fmt.state

On Thu 2025-02-06 17:32:32, Andy Shevchenko wrote:
> On Thu, Feb 06, 2025 at 01:25:07AM +0800, I Hsin Cheng wrote:
> > Remove unused assignment of "fmt.state", in both cases the value of
> > "fmt.state" will be overwritten by either "FORMAT_STATE_PRECISION" or
> > "FORMAT_STATE_NUM", the value "FORMAT_STATE_NONE" isn't going to be used
> > after the assignment.
> 
> ...
> 
> > struct fmt format_decode(struct fmt fmt, struct printf_spec *spec)
> 
> >  			spec->field_width = -spec->field_width;
> >  			spec->flags |= LEFT;
> >  		}
> > -		fmt.state = FORMAT_STATE_NONE;
> > +
> >  		goto precision;
> >  	}
> >  
> 
> While both are kinda redundant, this is not obvious what's stated in the commit
> message. Yes, `goto qualifier;` is straightforward, but not `goto precision;`.
> Which makes me think that these assignments can make code robust against
> potential future changes to allow to catch up the wrong code paths.

I fully agree with Andy here.

That said, I see the following right below the two conditions modified
in this patch:

	/* By default */
	fmt.state = FORMAT_STATE_NONE;

A good solution would be to move it up. It will be then obvious
that we could remove these two initializations. I mean
to do the following:

--- a/lib/vsprintf.c
+++ b/lib/vsprintf.c
@@ -2563,13 +2563,15 @@ struct fmt format_decode(struct fmt fmt, struct printf_spec *spec)
 	const char *start = fmt.str;
 	char flag;
 
+	/* By default */
+	fmt.state = FORMAT_STATE_NONE;
+
 	/* we finished early by reading the field width */
 	if (unlikely(fmt.state == FORMAT_STATE_WIDTH)) {
 		if (spec->field_width < 0) {
 			spec->field_width = -spec->field_width;
 			spec->flags |= LEFT;
 		}
-		fmt.state = FORMAT_STATE_NONE;
 		goto precision;
 	}
 
@@ -2578,13 +2580,9 @@ struct fmt format_decode(struct fmt fmt, struct printf_spec *spec)
 		if (spec->precision < 0)
 			spec->precision = 0;
 
-		fmt.state = FORMAT_STATE_NONE;
 		goto qualifier;
 	}
 
-	/* By default */
-	fmt.state = FORMAT_STATE_NONE;
-
 	for (; *fmt.str ; fmt.str++) {
 		if (*fmt.str == '%')
 			break;


Best Regards,
Petr