linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Boris Brezillon <boris.brezillon@free-electrons.com>
To: Daniel Walter <dwalter@sigma-star.at>
Cc: linux-mtd@lists.infradead.org,
	Richard Weinberger <richard@nod.at>,
	computersforpeace@gmail.com, dwmw2@infradead.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 19/46] mtd: nandsim: UAPI v1
Date: Sun, 20 Nov 2016 11:13:35 +0100	[thread overview]
Message-ID: <20161120111335.125f4a4b@bbrezillon> (raw)
In-Reply-To: <d7d9e926ff122be5801b6788b96adb8d24289ceb.1474450296.git.dwalter@sigma-star.at>

Daniel, Richard,

On Wed, 21 Sep 2016 11:50:45 +0200
Daniel Walter <dwalter@sigma-star.at> wrote:

> From: Richard Weinberger <richard@nod.at>
> 
> Expose nandsim creation and delete functions
> to user-space

Did you consider using configfs for nandsim dev creation/removal. The
main benefit I see in using configfs is that you can extend the
parameters without breaking the ABI.

Here is an example of what I have in mind:

mtdsim subsystem:

 <configfs-mountpoint>/mtdsim

nand backend:

 <configfs-mountpoint>/mtdsim/nand

create and configure a new nandsim device:

 mkdir <configfs-mountpoint>/mtdsim/nand/<dev-name>
 echo xxx >  <configfs-mountpoint>/mtdsim/nand/<dev-name>/pagesize
 echo xxx >  <configfs-mountpoint>/mtdsim/nand/<dev-name>/oobsize
 ...

create and configure a partitition:

 mkdir <configfs-mountpoint>/mtdsim/nand/<dev-name>/<part-name>
 echo xxx > <configfs-mountpoint>/mtdsim/nand/<dev-name>/<part-name>/offset
 echo xxx > <configfs-mountpoint>/mtdsim/nand/<dev-name>/<part-name>/size

Once your done with the configuration of your nandsim device, you
can register it by doing:

 echo 1 >  <configfs-mountpoint>/mtdsim/nand/<dev-name>/enable

and unregister it with:

 echo 0 >  <configfs-mountpoint>/mtdsim/nand/<dev-name>/enable

This way, each time you need to add a new feature (bitflips emulation,
paired pages emulation, ...) or a new property, you can just add an
extra attribute to the configfs interface without breaking the previous
ABI.

I'd really prefer this solution over the nandsim_ctrl dev exposed in
/dev/.

What do you think.

> 
> Signed-off-by: Richard Weinberger <richard@nod.at>
> ---
>  include/uapi/mtd/nandsim-user.h | 102 ++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 102 insertions(+)
>  create mode 100644 include/uapi/mtd/nandsim-user.h
> 
> diff --git a/include/uapi/mtd/nandsim-user.h b/include/uapi/mtd/nandsim-user.h
> new file mode 100644
> index 0000000..b52d620
> --- /dev/null
> +++ b/include/uapi/mtd/nandsim-user.h
> @@ -0,0 +1,102 @@
> +#ifndef __NANDSIM_USER_H__
> +#define __NANDSIM_USER_H__
> +
> +#include <linux/types.h>
> +
> +#define NANDSIM_IOC_MAGIC 'n'
> +
> +#define NANDSIM_IOC_NEW_INSTANCE _IOW(NANDSIM_IOC_MAGIC, 0, struct ns_new_instance_req)
> +#define NANDSIM_IOC_DESTROY_INSTANCE _IOW(NANDSIM_IOC_MAGIC, 1, struct ns_destroy_instance_req)
> +
> +#define NANDSIM_MAX_DEVICES 32
> +#define NANDSIM_MAX_PARTS 32
> +
> +enum ns_backend_type {
> +	NANDSIM_BACKEND_RAM = 0,
> +	NANDSIM_BACKEND_CACHEFILE = 1,
> +	NANDSIM_BACKEND_FILE = 2,
> +	NANDSIM_BACKEND_MAX,
> +};
> +
> +/**
> + * struct ns_new_instance_req - Create a new nandsim instance.
> + *
> + * @id_bytes: NAND ID of the simulated NAND chip
> + * @bus_width: bus width to emulate, either 8 or 16
> + * @bbt_mode: bad block table mode, 0 OOB, 1 BBT with marker in OOB,
> + *            2 BBT with marker in data area
> + * @no_oob: backing file contains no OOB data
> + * @bch_strength: instead of hamming ECC use BCH with given strength
> + * @parts_num: number of MTD partitions to create
> + * @parts: partition sizes in physical erase blocks, used then @parts_num > 0
> + * @backend: backend type, see @ns_backend_type
> + * @file_fd: file describtor of backend, only for @NANDSIM_BACKEND_CACHEFILE
> + *           and @NANDSIM_BACKEND_FILE.
> + * @bitflips: maximum number of random bit flips per page
> + * @overridesize: specifies the NAND size overriding the ID bytes
> + * @access_delay: initial page access delay (microseconds)
> + * @program_delay: page program delay (microseconds)
> + * @erase_delay: sector erase delay (milliseconds)
> + * @output_cycle: word output, from flash, time (nanoseconds)
> + * @input_cycle: word input, to flash, time (nanoseconds)
> + * @simelem_num: number of simulation elements appened to this
> + *               data structure. see @ns_simelement_prop
> + *
> + * This struct is used with the @NANDSIM_IOC_NEW_INSTANCE ioctl command.
> + * It creates a new nandsim instance from the given parameter.
> + * The ioctl command returns in case of success the nandsim id of the new
> + * instance, in case of error a negative value.
> + *
> + * Not all fields in the struct have to be filled, if nandsim should
> + * use a default ignore the value, fill with 0.
> + * The only mandatory fields are @id_bytes and @bus_width.
> + * When @no_oob is non-zero @bch_strength cannot be used since
> + * @no_oob implies that no ECC is used.
> + */
> +struct ns_new_instance_req {
> +	__s8 id_bytes[8];
> +
> +	__s8 bus_width;
> +	__s8 bbt_mode;
> +	__s8 no_oob;
> +	__s32 bch_strength;
> +
> +	__s8 parts_num;
> +	__s32 parts[NANDSIM_MAX_PARTS];
> +
> +	__s8 backend;
> +	__s32 file_fd;
> +
> +	__s32 bitflips;
> +	__s32 overridesize;
> +	__s32 access_delay;
> +	__s32 program_delay;
> +	__s32 erase_delay;
> +	__s32 output_cycle;
> +	__s32 input_cycle;
> +
> +	__s32 padding[4];
> +
> +	__s32 simelem_num;
> +} __packed;
> +
> +enum {
> +	NANDSIM_SIMELEM_BADBLOCK = 0,
> +	NANDSIM_SIMELEM_WEAKBLOCK,
> +	NANDSIM_SIMELEM_WEAKPAGE,
> +	NANDSIM_SIMELEM_GRAVEPAGE,
> +};
> +
> +struct ns_simelement_prop {
> +	__s8 elem_type;
> +	__s32 elem_id;
> +	__s32 elem_attr;
> +	__s8 padding[7];
> +} __packed;
> +
> +struct ns_destroy_instance_req {
> +	__s8 id;
> +	__s8 padding[7];
> +} __packed;
> +
> +#endif /* __NANDSIM_USER_H__ */

  reply	other threads:[~2016-11-20 10:13 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-09-21  9:43 [PATCH v2 00/46] Nandsim facelift (part I of II) Daniel Walter
2016-09-21  9:43 ` [PATCH v2 01/46] mtdpart: Propagate _get/put_device() Daniel Walter
2016-09-21 10:15   ` Boris Brezillon
2016-09-28 20:16     ` Brian Norris
2016-12-14 19:24       ` Karl Beldan
2016-12-14 21:09         ` Brian Norris
2016-12-14 21:12           ` Richard Weinberger
2016-12-14 23:40             ` Brian Norris
2016-12-15  7:09           ` Karl Beldan
2016-12-15  7:51             ` Richard Weinberger
2016-12-28 18:53               ` Karl Beldan
2016-09-21  9:44 ` [PATCH v2 02/46] mtd: nand: Provide nand_cleanup() function to free NAND related resources Daniel Walter
2016-09-21  9:58   ` Boris Brezillon
2016-09-21 12:43   ` kbuild test robot
2016-09-21 14:25   ` Boris Brezillon
2016-09-21 14:38     ` Daniel Walter
2016-09-21 14:42       ` Boris Brezillon
2016-09-21  9:45 ` [PATCH v2 03/46] mtd: Don't unconditionally unregister reboot notifier Daniel Walter
2016-09-21 14:31   ` Boris Brezillon
2016-09-21 14:33     ` Daniel Walter
2016-10-09  5:20   ` Brian Norris
2016-09-21  9:45 ` [PATCH v2 04/46] mtd: Don't unconditionally execute remove notifiers Daniel Walter
2016-09-21  9:46 ` [PATCH v2 05/46] mtd: Don't print a scary message when trying to remove a busy MTD Daniel Walter
2016-09-21  9:46 ` [PATCH v2 06/46] mtd: nandsim: Add basic control file support Daniel Walter
2016-09-21  9:47 ` [PATCH v2 07/46] mtd: nandsim: Begin with removal of global state Daniel Walter
2016-09-21  9:47 ` [PATCH v2 08/46] mtd: nandsim: Kill global nsmtd Daniel Walter
2016-09-21  9:47 ` [PATCH v2 09/46] mtd: nandsim: Don't directly use module parameters Daniel Walter
2016-09-21  9:48 ` [PATCH v2 10/46] mtd: nandsim: Add helper functions for pointer magic Daniel Walter
2016-09-21  9:48 ` [PATCH v2 11/46] mtd: nandsim: Factor out nandsim parameters Daniel Walter
2016-09-21  9:48 ` [PATCH v2 12/46] mtd: nandsim: Make debugfs logic multi instance capable Daniel Walter
2016-09-21  9:49 ` [PATCH v2 13/46] mtd: nandsim: Add final logic for multiple instances Daniel Walter
2016-09-21  9:49 ` [PATCH v2 14/46] mtd: nandsim: Add simulator id to MTD parition name Daniel Walter
2016-09-21  9:49 ` [PATCH v2 15/46] mtd: nandsim: Introduce backend operations Daniel Walter
2016-09-21  9:49 ` [PATCH v2 16/46] mtd: nandsim: Print error when backend init failed Daniel Walter
2016-09-21  9:50 ` [PATCH v2 17/46] mtd: nandsim: Allow external backends Daniel Walter
2016-09-21  9:50 ` [PATCH v2 18/46] mtd: nandsim: Add basic support for a file backend Daniel Walter
2016-09-21  9:50 ` [PATCH v2 19/46] mtd: nandsim: UAPI v1 Daniel Walter
2016-11-20 10:13   ` Boris Brezillon [this message]
2016-09-21  9:51 ` [PATCH v2 20/46] mtd: nandsim: Implement preliminary constructor function Daniel Walter
2016-09-21  9:51 ` [PATCH v2 21/46] mtd: nandsim: Implement preliminary destructor function Daniel Walter
2016-09-21 12:56   ` kbuild test robot
2016-09-21  9:51 ` [PATCH v2 22/46] mtd: nandsim: Cleanup destroy handlers Daniel Walter
2016-09-21  9:51 ` [PATCH v2 23/46] mtd: nandsim: Unify file backend init logic Daniel Walter
2016-09-21  9:51 ` [PATCH v2 24/46] mtd: nandsim: Wire up NANDSIM_MODE_CACHEFILE ioctl mode Daniel Walter
2016-09-21  9:52 ` [PATCH v2 25/46] mtd: nandsim: Print backend name Daniel Walter
2016-09-21  9:52 ` [PATCH v2 26/46] mtd: nandsim: use the existing output macros Daniel Walter
2016-09-21  9:52 ` [PATCH v2 27/46] mtd: nandsim: Add no_oob mode Daniel Walter
2016-09-21  9:52 ` [PATCH v2 28/46] mtd: nandsim: Refine exports Daniel Walter
2016-09-21  9:54 ` [PATCH v2 29/46] um: Add nandsim backend driver Daniel Walter
2016-09-21  9:54 ` [PATCH v2 30/46] mtd: nandsim: Use pr_ style logging Daniel Walter
2016-09-21  9:54 ` [PATCH v2 31/46] mtd: nandsim: Remove NS_RAW_OFFSET_OOB Daniel Walter
2016-09-21  9:54 ` [PATCH v2 32/46] mtd: nandsim: Remove NS_IS_INITIALIZED Daniel Walter
2016-09-21  9:55 ` [PATCH v2 33/46] mtd: nandsim: Relax page size restrictions Daniel Walter
2016-09-21  9:55 ` [PATCH v2 34/46] mtd: nandsim: Support bitflip and read error emulation in file backend Daniel Walter
2016-09-21  9:55 ` [PATCH v2 35/46] mtd: nandsim: Make NANDSIM_MAX_DEVICES part of uapi Daniel Walter
2016-09-21  9:55 ` [PATCH v2 36/46] mtd: nandsim: Cleanup constants Daniel Walter
2016-09-21  9:55 ` [PATCH v2 37/46] mtd: nandsim: Turn parts[] into a integer Daniel Walter
2016-09-21  9:56 ` [PATCH v2 38/46] mtd: nandsim: Expose partition creation logic to user space Daniel Walter
2016-09-21  9:56 ` [PATCH v2 39/46] mtd: nandsim: Rework init error paths Daniel Walter
2016-09-21  9:56 ` [PATCH v2 40/46] mtd: nandsim: Expose BBT, delays, etc.. to userspace Daniel Walter
2016-09-21  9:56 ` [PATCH v2 41/46] mtd: nandsim: Expose support for weakpages/blocks " Daniel Walter
2016-09-21  9:57 ` [PATCH v2 42/46] mtd: nandsim: Don't printk on ENOMEM Daniel Walter
2016-09-21  9:57 ` [PATCH v2 43/46] mtd: nandsim: Wire up NANDSIM_IOC_NEW_INSTANCE Daniel Walter
2016-09-21  9:57 ` [PATCH v2 44/46] mtd: nandsim: Wire up NANDSIM_IOC_DESTROY_INSTANCE Daniel Walter
2016-09-21  9:57 ` [PATCH v2 45/46] mtd: nandsim: Always answer all 8 bytes from NAND_CMD_READID Daniel Walter
2016-09-21  9:57 ` [PATCH v2 46/46] mtd/nandsim: Add ioctl for info Daniel Walter
2016-10-16 16:24 ` [PATCH v2 00/46] Nandsim facelift (part I of II) Boris Brezillon
2016-11-14 16:24   ` Richard Weinberger
2016-11-20 10:26     ` Boris Brezillon

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=20161120111335.125f4a4b@bbrezillon \
    --to=boris.brezillon@free-electrons.com \
    --cc=computersforpeace@gmail.com \
    --cc=dwalter@sigma-star.at \
    --cc=dwmw2@infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mtd@lists.infradead.org \
    --cc=richard@nod.at \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).