linux-kernel - Re: [PATCH V3] block: uapi: Use unsigned int type for IOPRIO_PRIO

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: <dafe01f5-5f08-4298-b020-7e3c80e4f15d@kernel.org>
Date: Wed, 31 Jul 2024 18:28:22 +0900
From: Damien Le Moal <dlemoal@...nel.org>
To: Zhiguo Niu <zhiguo.niu@...soc.com>, axboe@...nel.dk, hch@....de
Cc: bvanassche@....org, linux-block@...r.kernel.org,
 linux-kernel@...r.kernel.org, niuzhiguo84@...il.com, ke.wang@...soc.com,
 Hao_hao.Wang@...soc.com
Subject: Re: [PATCH V3] block: uapi: Use unsigned int type for
 IOPRIO_PRIO_MASK

On 7/31/24 16:01, Zhiguo Niu wrote:
> Generally, the input of IOPRIO_PRIO_DATA has 16 bits, but the output of
> IOPRIO_PRIO_DATA will be expanded to "UL" from IOPRIO_PRIO_MASK.
>  #define IOPRIO_PRIO_MASK	((1UL << IOPRIO_CLASS_SHIFT) - 1)
> This is not reasonable and meaningless, unsigned int is more suitable for it.
> 
> So if use format "%d" to print IOPRIO_PRIO_DATA directly, there will be a
> build warning or error showned as the following, which is from the
> local test when I modify f2fs codes.
> 
> fs/f2fs/sysfs.c:348:31: warning: format ‘%d’ expects argument of type ‘int’,
> but argument 4 has type ‘long unsigned int’ [-Wformat=]
>    return sysfs_emit(buf, "%s,%d\n",
>                               ~^
>                               %ld
> 
> When modules use IOPRIO_PRIO_CLASS & IOPRIO_PRIO_LEVEL get ioprio's class and
> level, their outputs are both unsigned int.
>  IOPRIO_CLASS_MASK is:
>  #define IOPRIO_CLASS_SHIFT	13
>  #define IOPRIO_NR_CLASSES	8
>  #define IOPRIO_CLASS_MASK	(IOPRIO_NR_CLASSES - 1)
>  IOPRIO_LEVEL_MASK is:
>  #define IOPRIO_LEVEL_NR_BITS	3
>  #define IOPRIO_NR_LEVELS	(1 << IOPRIO_LEVEL_NR_BITS)
>  #define IOPRIO_LEVEL_MASK	(IOPRIO_NR_LEVELS - 1)
> 
> Ioprio is passed along as an int internally, so we should not be using an
> unsigned long for IOPRIO_PRIO_MASK to not end up with IOPRIO_PRIO_DATA
> returning an unsigned long as well.

I would write this commit message like this:


An ioprio is passed internally as an int value. When IOPRIO_PRIO_CLASS() and
IOPRIO_PRIO_LEVEL() are used to extract from it the priority class and level,
the values obtained are thus also int.
However, the IOPRIO_PRIO_MASK() macro used to define the IOPRIO_PRIO_DATA()
macro is defined as:

#define IOPRIO_PRIO_MASK	((1UL << IOPRIO_CLASS_SHIFT) - 1)

that is, the macro gives an unsigned long value, which leads to
IOPRIO_PRIO_DATA() also returning an unsigned long.

Make things consistent between class, level and data and use int everywhere by
removing forced unsigned long from IOPRIO_PRIO_MASK.



-- 
Damien Le Moal
Western Digital Research