All of lore.kernel.org
 help / color / mirror / Atom feed
From: David Laight <David.Laight@ACULAB.COM>
To: 'Stephen Kitt' <steve@sk2.org>, "hare@suse.com" <hare@suse.com>,
	"axboe@kernel.dk" <axboe@kernel.dk>,
	"jejb@linux.vnet.ibm.com" <jejb@linux.vnet.ibm.com>,
	"martin.petersen@oracle.com" <martin.petersen@oracle.com>
Cc: "linux-scsi@vger.kernel.org" <linux-scsi@vger.kernel.org>,
	"kernel-hardening@lists.openwall.com"
	<kernel-hardening@lists.openwall.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-block@vger.kernel.org" <linux-block@vger.kernel.org>
Subject: RE: [PATCH] scsi: resolve COMMAND_SIZE at compile time
Date: Tue, 13 Mar 2018 11:34:31 +0000	[thread overview]
Message-ID: <a66f8abb84444cf79b818011667bc825@AcuMS.aculab.com> (raw)
In-Reply-To: <20180309223355.21222-1-steve@sk2.org>

From: Stephen Kitt
> Sent: 09 March 2018 22:34
>=20
> COMMAND_SIZE currently uses an array of values in block/scsi_ioctl.c.
> A number of device_handler functions use this to initialise arrays,
> and this is flagged by -Wvla.
>=20
> This patch replaces COMMAND_SIZE with a variant using a formula which
> can be resolved at compile time in cases where the opcode is fixed,
> resolving the array size and avoiding the VLA. The downside is that
> the code is less maintainable and that some call sites end up having
> more complex generated code.
>=20
> Since scsi_command_size_tbl is dropped, we can remove the dependency
> on BLK_SCSI_REQUEST from drivers/target/Kconfig.
>=20
> This was prompted by https://lkml.org/lkml/2018/3/7/621
>=20
> Signed-off-by: Stephen Kitt <steve@sk2.org>
> ---
>  block/scsi_ioctl.c         |  8 --------
>  drivers/target/Kconfig     |  1 -
>  include/scsi/scsi_common.h | 13 +++++++++++--
>  3 files changed, 11 insertions(+), 11 deletions(-)
>=20
> diff --git a/block/scsi_ioctl.c b/block/scsi_ioctl.c
> index 60b471f8621b..b9e176e537be 100644
> --- a/block/scsi_ioctl.c
> +++ b/block/scsi_ioctl.c
> @@ -41,14 +41,6 @@ struct blk_cmd_filter {
>=20
>  static struct blk_cmd_filter blk_default_cmd_filter;
>=20
> -/* Command group 3 is reserved and should never be used.  */
> -const unsigned char scsi_command_size_tbl[8] =3D
> -{
> -	6, 10, 10, 12,
> -	16, 12, 10, 10
> -};
> -EXPORT_SYMBOL(scsi_command_size_tbl);
> -
>  #include <scsi/sg.h>
>=20
>  static int sg_get_version(int __user *p)
> diff --git a/drivers/target/Kconfig b/drivers/target/Kconfig
> index 4c44d7bed01a..b5f05b60cf06 100644
> --- a/drivers/target/Kconfig
> +++ b/drivers/target/Kconfig
> @@ -4,7 +4,6 @@ menuconfig TARGET_CORE
>  	depends on SCSI && BLOCK
>  	select CONFIGFS_FS
>  	select CRC_T10DIF
> -	select BLK_SCSI_REQUEST # only for scsi_command_size_tbl..
>  	select SGL_ALLOC
>  	default n
>  	help
> diff --git a/include/scsi/scsi_common.h b/include/scsi/scsi_common.h
> index 731ac09ed231..48d950666376 100644
> --- a/include/scsi/scsi_common.h
> +++ b/include/scsi/scsi_common.h
> @@ -15,8 +15,17 @@ scsi_varlen_cdb_length(const void *hdr)
>  	return ((struct scsi_varlen_cdb_hdr *)hdr)->additional_cdb_length + 8;
>  }
>=20
> -extern const unsigned char scsi_command_size_tbl[8];
> -#define COMMAND_SIZE(opcode) scsi_command_size_tbl[((opcode) >> 5) & 7]
> +/*
> + * SCSI command sizes are as follows, in bytes, for fixed size commands,=
 per
> + * group: 6, 10, 10, 12, 16, 12, 10, 10. The top three bits of an opcode
> + * determine its group.
> + * The size table is encoded into a 32-bit value by subtracting each val=
ue
> + * from 16, resulting in a value of 1715488362
> + * (6 << 28 + 6 << 24 + 4 << 20 + 0 << 16 + 4 << 12 + 6 << 8 + 6 << 4 + =
10).
> + * Command group 3 is reserved and should never be used.
> + */
> +#define COMMAND_SIZE(opcode) \
> +	(16 - (15 & (1715488362 >> (4 * (((opcode) >> 5) & 7)))))

Why not:
	(6 + (15 & (0x446a6440u ....

Specifying the constant in hex makes it more obvious.
It is a shame about the 16, but an offset is easier than the negate.

	David

WARNING: multiple messages have this Message-ID (diff)
From: David Laight <David.Laight@ACULAB.COM>
To: 'Stephen Kitt' <steve@sk2.org>, "hare@suse.com" <hare@suse.com>,
	"axboe@kernel.dk" <axboe@kernel.dk>,
	"jejb@linux.vnet.ibm.com" <jejb@linux.vnet.ibm.com>,
	"martin.petersen@oracle.com" <martin.petersen@oracle.com>
Cc: "linux-scsi@vger.kernel.org" <linux-scsi@vger.kernel.org>,
	"kernel-hardening@lists.openwall.com"
	<kernel-hardening@lists.openwall.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-block@vger.kernel.org" <linux-block@vger.kernel.org>
Subject: RE: [PATCH] scsi: resolve COMMAND_SIZE at compile time
Date: Tue, 13 Mar 2018 11:34:31 +0000	[thread overview]
Message-ID: <a66f8abb84444cf79b818011667bc825@AcuMS.aculab.com> (raw)
In-Reply-To: <20180309223355.21222-1-steve@sk2.org>

From: Stephen Kitt
> Sent: 09 March 2018 22:34
> 
> COMMAND_SIZE currently uses an array of values in block/scsi_ioctl.c.
> A number of device_handler functions use this to initialise arrays,
> and this is flagged by -Wvla.
> 
> This patch replaces COMMAND_SIZE with a variant using a formula which
> can be resolved at compile time in cases where the opcode is fixed,
> resolving the array size and avoiding the VLA. The downside is that
> the code is less maintainable and that some call sites end up having
> more complex generated code.
> 
> Since scsi_command_size_tbl is dropped, we can remove the dependency
> on BLK_SCSI_REQUEST from drivers/target/Kconfig.
> 
> This was prompted by https://lkml.org/lkml/2018/3/7/621
> 
> Signed-off-by: Stephen Kitt <steve@sk2.org>
> ---
>  block/scsi_ioctl.c         |  8 --------
>  drivers/target/Kconfig     |  1 -
>  include/scsi/scsi_common.h | 13 +++++++++++--
>  3 files changed, 11 insertions(+), 11 deletions(-)
> 
> diff --git a/block/scsi_ioctl.c b/block/scsi_ioctl.c
> index 60b471f8621b..b9e176e537be 100644
> --- a/block/scsi_ioctl.c
> +++ b/block/scsi_ioctl.c
> @@ -41,14 +41,6 @@ struct blk_cmd_filter {
> 
>  static struct blk_cmd_filter blk_default_cmd_filter;
> 
> -/* Command group 3 is reserved and should never be used.  */
> -const unsigned char scsi_command_size_tbl[8] =
> -{
> -	6, 10, 10, 12,
> -	16, 12, 10, 10
> -};
> -EXPORT_SYMBOL(scsi_command_size_tbl);
> -
>  #include <scsi/sg.h>
> 
>  static int sg_get_version(int __user *p)
> diff --git a/drivers/target/Kconfig b/drivers/target/Kconfig
> index 4c44d7bed01a..b5f05b60cf06 100644
> --- a/drivers/target/Kconfig
> +++ b/drivers/target/Kconfig
> @@ -4,7 +4,6 @@ menuconfig TARGET_CORE
>  	depends on SCSI && BLOCK
>  	select CONFIGFS_FS
>  	select CRC_T10DIF
> -	select BLK_SCSI_REQUEST # only for scsi_command_size_tbl..
>  	select SGL_ALLOC
>  	default n
>  	help
> diff --git a/include/scsi/scsi_common.h b/include/scsi/scsi_common.h
> index 731ac09ed231..48d950666376 100644
> --- a/include/scsi/scsi_common.h
> +++ b/include/scsi/scsi_common.h
> @@ -15,8 +15,17 @@ scsi_varlen_cdb_length(const void *hdr)
>  	return ((struct scsi_varlen_cdb_hdr *)hdr)->additional_cdb_length + 8;
>  }
> 
> -extern const unsigned char scsi_command_size_tbl[8];
> -#define COMMAND_SIZE(opcode) scsi_command_size_tbl[((opcode) >> 5) & 7]
> +/*
> + * SCSI command sizes are as follows, in bytes, for fixed size commands, per
> + * group: 6, 10, 10, 12, 16, 12, 10, 10. The top three bits of an opcode
> + * determine its group.
> + * The size table is encoded into a 32-bit value by subtracting each value
> + * from 16, resulting in a value of 1715488362
> + * (6 << 28 + 6 << 24 + 4 << 20 + 0 << 16 + 4 << 12 + 6 << 8 + 6 << 4 + 10).
> + * Command group 3 is reserved and should never be used.
> + */
> +#define COMMAND_SIZE(opcode) \
> +	(16 - (15 & (1715488362 >> (4 * (((opcode) >> 5) & 7)))))

Why not:
	(6 + (15 & (0x446a6440u ....

Specifying the constant in hex makes it more obvious.
It is a shame about the 16, but an offset is easier than the negate.

	David

  parent reply	other threads:[~2018-03-13 11:34 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-03-09 22:29 VLA removal, device_handler and COMMAND_SIZE Stephen Kitt
2018-03-09 22:32 ` [PATCH] device_handler: remove VLAs Stephen Kitt
2018-03-09 22:48   ` Bart Van Assche
2018-03-09 22:48     ` Bart Van Assche
2018-03-10 13:14     ` Stephen Kitt
2018-03-12 15:41       ` Bart Van Assche
2018-03-12 15:41         ` Bart Van Assche
2018-03-12 19:26         ` Stephen Kitt
2018-03-12  6:25   ` Hannes Reinecke
2018-03-13  2:37   ` Martin K. Petersen
2018-03-09 22:33 ` [PATCH] scsi: resolve COMMAND_SIZE at compile time Stephen Kitt
2018-03-09 22:47   ` Bart Van Assche
2018-03-09 22:47     ` Bart Van Assche
2018-03-10 13:29     ` Stephen Kitt
2018-03-10 20:49       ` James Bottomley
2018-03-10 21:16         ` Douglas Gilbert
2018-03-11 15:01         ` Stephen Kitt
2018-03-11 15:01           ` Stephen Kitt
2018-03-13 11:34   ` David Laight [this message]
2018-03-13 11:34     ` David Laight

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=a66f8abb84444cf79b818011667bc825@AcuMS.aculab.com \
    --to=david.laight@aculab.com \
    --cc=axboe@kernel.dk \
    --cc=hare@suse.com \
    --cc=jejb@linux.vnet.ibm.com \
    --cc=kernel-hardening@lists.openwall.com \
    --cc=linux-block@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-scsi@vger.kernel.org \
    --cc=martin.petersen@oracle.com \
    --cc=steve@sk2.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.