linux-mtd.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
From: Miquel Raynal <miquel.raynal@bootlin.com>
To: Boris Brezillon <boris.brezillon@collabora.com>
Cc: Rickard Andersson <rickaran@axis.com>,
	Richard Weinberger <richard@nod.at>,
	linux-mtd@lists.infradead.org,
	Vignesh Raghavendra <vigneshr@ti.com>,
	Tudor Ambarus <Tudor.Ambarus@microchip.com>
Subject: Re: [PATCH v5 28/28] mtd: rawnand: Allocate the best data interface structure dynamically
Date: Wed, 27 May 2020 09:57:32 +0200	[thread overview]
Message-ID: <20200527095732.467db722@xps13> (raw)
In-Reply-To: <20200527002844.7e54aa22@collabora.com>


Boris Brezillon <boris.brezillon@collabora.com> wrote on Wed, 27 May
2020 00:28:44 +0200:

> On Tue, 26 May 2020 21:17:25 +0200
> Miquel Raynal <miquel.raynal@bootlin.com> wrote:
> 
> > Instead of manipulating the statically allocated structure and copy
> > timings around, allocate one at identification time and save it in the
> > nand_chip structure once it has been initialized.
> > 
> > This way, either there is a "best data interface" and the requested
> > timings will be these, or there is none, forcing the core to use the
> > default set, statically defined in the core, shared across all NAND
> > chips.
> > 
> > Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
> > ---
> >  drivers/mtd/nand/raw/nand_base.c | 42 ++++++++++++++++++++++----------
> >  include/linux/mtd/rawnand.h      | 13 +++++++---
> >  2 files changed, 39 insertions(+), 16 deletions(-)
> > 
> > diff --git a/drivers/mtd/nand/raw/nand_base.c b/drivers/mtd/nand/raw/nand_base.c
> > index 2547136a9cd7..247bf5faaf55 100644
> > --- a/drivers/mtd/nand/raw/nand_base.c
> > +++ b/drivers/mtd/nand/raw/nand_base.c
> > @@ -947,7 +947,7 @@ static int nand_reset_data_interface(struct nand_chip *chip, int chipnr)
> >   */
> >  static int nand_setup_data_interface(struct nand_chip *chip, int chipnr)
> >  {
> > -	u8 mode = chip->data_interface.timings.mode;
> > +	u8 mode = chip->best_iface->timings.mode;
> >  	u8 tmode_param[ONFI_SUBFEATURE_PARAM_LEN] = { mode, };
> >  	int ret;
> >  
> > @@ -966,7 +966,7 @@ static int nand_setup_data_interface(struct nand_chip *chip, int chipnr)
> >  
> >  	/* Change the mode on the controller side */
> >  	ret = chip->controller->ops->setup_data_interface(chip, chipnr,
> > -							&chip->data_interface);
> > +							  chip->best_iface);
> >  	if (ret)
> >  		return ret;
> >  
> > @@ -1030,8 +1030,10 @@ int nand_choose_best_sdr_timings(struct nand_chip *chip,
> >  		ret = ops->setup_data_interface(chip,
> >  						NAND_DATA_IFACE_CHECK_ONLY,
> >  						iface);
> > -		if (!ret)
> > +		if (!ret) {
> > +			chip->best_iface = iface;
> >  			return ret;
> > +		}
> >  
> >  		/* Fallback to slower modes */
> >  		best_mode = iface->timings.mode;
> > @@ -1052,6 +1054,8 @@ int nand_choose_best_sdr_timings(struct nand_chip *chip,
> >  			break;
> >  	}
> >  
> > +	chip->best_iface = iface;
> > +
> >  	return 0;
> >  }
> >  
> > @@ -1070,14 +1074,25 @@ int nand_choose_best_sdr_timings(struct nand_chip *chip,
> >   */
> >  static int nand_choose_data_interface(struct nand_chip *chip)
> >  {
> > +	struct nand_data_interface *iface;
> > +	int ret;
> > +
> >  	if (!nand_controller_has_setup_data_iface(chip))
> >  		return 0;
> >  
> > +	iface = kzalloc(sizeof(*iface), GFP_KERNEL);
> > +	if (!iface)
> > +		return -ENOMEM;
> > +
> >  	if (chip->ops.choose_data_interface)
> > -		return chip->ops.choose_data_interface(chip,
> > -						       &chip->data_interface);
> > +		ret = chip->ops.choose_data_interface(chip, iface);
> > +	else
> > +		ret = nand_choose_best_sdr_timings(chip, iface, NULL);
> >  
> > -	return nand_choose_best_sdr_timings(chip, &chip->data_interface, NULL);
> > +	if (ret)
> > +		kfree(iface);
> > +
> > +	return ret;
> >  }
> >  
> >  /**
> > @@ -2514,8 +2529,7 @@ int nand_reset(struct nand_chip *chip, int chipnr)
> >  	 * nand_setup_data_interface() uses ->set/get_features() which would
> >  	 * fail anyway as the parameter page is not available yet.
> >  	 */
> > -	if (!memcmp(&chip->data_interface, nand_reset_data_iface,
> > -		    sizeof(*nand_reset_data_iface)))
> > +	if (!chip->best_iface)
> >  		return 0;
> >    
> 
> If you assign ->best_iface for mode 0 (as done above) and keep the
> !chip->best_iface test that means you apply timing mode 0 twice. Not
> a big deal but I wanted to point it out.

That's partially true, because what is your way to discriminate between
a virgin mode 0 and a mode 1 but not totally 1 so it is reported as 0?
We should do a memcmp(). So IMHO it's worth the extra call to
nand_setup_data_interface() to have a clear and quick condition here.

> 
> >  	ret = nand_setup_data_interface(chip, chipnr);
> > @@ -5252,9 +5266,6 @@ static int nand_scan_ident(struct nand_chip *chip, unsigned int maxchips,
> >  
> >  	mutex_init(&chip->lock);
> >  
> > -	/* Enforce the right timings for reset/detection */
> > -	onfi_fill_data_interface(chip, &chip->data_interface, NAND_SDR_IFACE, 0);
> > -
> >  	ret = nand_dt_init(chip);
> >  	if (ret)
> >  		return ret;
> > @@ -6059,7 +6070,7 @@ static int nand_scan_tail(struct nand_chip *chip)
> >  	for (i = 0; i < nanddev_ntargets(&chip->base); i++) {
> >  		ret = nand_setup_data_interface(chip, i);
> >  		if (ret)
> > -			goto err_nanddev_cleanup;
> > +			goto err_free_data_iface;
> >  	}
> >  
> >  	/* Check, if we should skip the bad block table scan */
> > @@ -6069,10 +6080,12 @@ static int nand_scan_tail(struct nand_chip *chip)
> >  	/* Build bad block table */
> >  	ret = nand_create_bbt(chip);
> >  	if (ret)
> > -		goto err_nanddev_cleanup;
> > +		goto err_free_data_iface;
> >  
> >  	return 0;
> >  
> > +err_free_data_iface:
> > +	kfree(chip->best_iface);
> >  
> >  err_nanddev_cleanup:
> >  	nanddev_cleanup(&chip->base);
> > @@ -6166,6 +6179,9 @@ void nand_cleanup(struct nand_chip *chip)
> >  			& NAND_BBT_DYNAMICSTRUCT)
> >  		kfree(chip->badblock_pattern);
> >  
> > +	/* Free the data interface */
> > +	kfree(chip->best_iface);
> > +
> >  	/* Free manufacturer priv data. */
> >  	nand_manufacturer_cleanup(chip);
> >  
> > diff --git a/include/linux/mtd/rawnand.h b/include/linux/mtd/rawnand.h
> > index 00557e553827..0f215faa3072 100644
> > --- a/include/linux/mtd/rawnand.h
> > +++ b/include/linux/mtd/rawnand.h
> > @@ -1070,7 +1070,9 @@ struct nand_manufacturer {
> >   * @options: Various chip options. They can partly be set to inform nand_scan
> >   *           about special functionality. See the defines for further
> >   *           explanation.
> > - * @data_interface: NAND interface timing information
> > + * @best_iface: The best NAND data interface which fits both the NAND chip and  
> 
> 						^ configuration
> 
> > + *              NAND controller constraints. If unset, the default reset data
> > + *              interface must be used.
> >   * @bbt_erase_shift: Number of address bits in a bbt entry
> >   * @bbt_options: Bad block table specific options. All options used here must
> >   *               come from bbm.h. By default, these options will be copied to
> > @@ -1117,7 +1119,7 @@ struct nand_chip {
> >  	unsigned int options;
> >  
> >  	/* Data interface */
> > -	struct nand_data_interface data_interface;
> > +	struct nand_data_interface *best_iface;  
> 
> Not sure why you rename this field, but if we go for a name update, I'd
> vote for best_iface_cfg (or something that clarifies the fact that this
> is the data interface configuration).

Maybe I misunderstood your request, you were saying that allocating a
"best data interface object" would be good, so I interpreted it as:
rename it, and allocated it dynamically. I'm fine keeping
data_interface and just declaring it as a pointer.

Anyway, I like talking about the "interface" rather than the "interface
configuration" which is implied in my mind, I saw you were asking to
add "configuration" sometimes, do you have something in mind that I
don't?

> 
> >  
> >  	/* Bad block information */
> >  	unsigned int bbt_erase_shift;
> > @@ -1212,7 +1214,12 @@ extern const struct nand_data_interface *nand_reset_data_iface;
> >  static inline const struct nand_sdr_timings *
> >  nand_get_sdr_timings(struct nand_chip *chip)
> >  {
> > -	return nand_extract_sdr_timings(&chip->data_interface);
> > +	const struct nand_data_interface *iface = chip->best_iface;
> > +
> > +	if (!iface)
> > +		iface = nand_reset_data_iface;
> > +
> > +	return nand_extract_sdr_timings(iface);
> >  }
> >  
> >  /*  
> 


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

  reply	other threads:[~2020-05-27  7:57 UTC|newest]

Thread overview: 62+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-05-26 19:16 [PATCH v5 00/28] Allow vendor drivers to propose their own timings Miquel Raynal
2020-05-26 19:16 ` [PATCH v5 01/28] mtd: rawnand: Use unsigned types for nand_chip unsigned values Miquel Raynal
2020-05-26 19:16 ` [PATCH v5 02/28] mtd: rawnand: Only use u8 instead of uint8_t in nand_chip structure Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 03/28] mtd: rawnand: Create a nand_chip operations structure Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 04/28] mtd: rawnand: Rename the manufacturer structure Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 05/28] mtd: rawnand: Declare the nand_manufacturer structure out of nand_chip Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 06/28] mtd: rawnand: Reorganize the nand_chip structure Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 07/28] mtd: rawnand: Compare the actual timing values Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 08/28] mtd: rawnand: Use the data interface mode entry when relevant Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 09/28] mtd: rawnand: Rename nand_has_setup_data_interface() Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 10/28] mtd: rawnand: Fix nand_setup_data_interface() description Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 11/28] mtd: rawnand: Rename nand_init_data_interface() Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 12/28] mtd: rawnand: timings: Update onfi_fill_data_interface() kernel doc Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 13/28] mtd: rawnand: timings: Provide onfi_fill_data_interface() with a data interface Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 14/28] mtd: rawnand: timings: Add a helper to find the closest ONFI mode Miquel Raynal
2020-05-26 21:42   ` Boris Brezillon
2020-05-27  7:42     ` Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 15/28] mtd: rawnand: timings: Avoid redefining tR_max and tCCS_min Miquel Raynal
2020-05-26 20:33   ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 16/28] mtd: rawnand: timings: Use default values for tPROG_max and tBERS_max Miquel Raynal
2020-05-26 20:32   ` Boris Brezillon
2020-05-26 21:25   ` Boris Brezillon
2020-05-27  7:17     ` Miquel Raynal
2020-05-27  7:22       ` Miquel Raynal
2020-05-27  7:28         ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 17/28] mtd: rawnand: Define a unique reset data interface Miquel Raynal
2020-05-26 21:23   ` Boris Brezillon
2020-05-27  7:20     ` Miquel Raynal
2020-05-27  8:10       ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 18/28] mtd: rawnand: marvell: Use a helper to access the timings Miquel Raynal
2020-05-26 21:26   ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 19/28] mtd: rawnand: legacy: " Miquel Raynal
2020-05-26 21:28   ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 20/28] mtd: rawnand: Hide the chip->data_interface indirection Miquel Raynal
2020-05-26 21:36   ` Boris Brezillon
2020-05-27  7:36     ` Miquel Raynal
2020-05-27  8:14       ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 21/28] mtd: rawnand: Introduce nand_choose_best_sdr_timings() Miquel Raynal
2020-05-26 21:50   ` Boris Brezillon
2020-05-27  7:45     ` Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 22/28] mtd: rawnand: Add the ->choose_data_interface() hook Miquel Raynal
2020-05-26 21:52   ` Boris Brezillon
2020-05-27  7:46     ` Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 23/28] mtd: rawnand: toshiba: Implement ->choose_data_interface() for TC58TEG5DCLTA00 Miquel Raynal
2020-05-26 21:54   ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 24/28] mtd: rawnand: toshiba: Implement ->choose_data_interface() for TC58NVG0S3E Miquel Raynal
2020-05-26 22:02   ` Boris Brezillon
2020-05-27  7:47     ` Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 25/28] mtd: rawnand: hynix: Implement ->choose_data_interface() for H27UCG8T2ATR-BC Miquel Raynal
2020-05-26 22:06   ` Boris Brezillon
2020-05-27  7:50     ` Miquel Raynal
2020-05-26 19:17 ` [PATCH v5 26/28] mtd: rawnand: toshiba: Choose the data interface for TH58NVG2S3HBAI4 Miquel Raynal
2020-05-26 22:10   ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 27/28] mtd: rawnand: Get rid of the default ONFI timing mode Miquel Raynal
2020-05-26 22:13   ` Boris Brezillon
2020-05-26 19:17 ` [PATCH v5 28/28] mtd: rawnand: Allocate the best data interface structure dynamically Miquel Raynal
2020-05-26 22:28   ` Boris Brezillon
2020-05-27  7:57     ` Miquel Raynal [this message]
2020-05-27  8:35       ` Boris Brezillon
2020-05-27  8:49         ` Miquel Raynal
2020-05-27  8:57           ` Boris Brezillon
2020-05-27  9:34             ` Miquel Raynal

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=20200527095732.467db722@xps13 \
    --to=miquel.raynal@bootlin.com \
    --cc=Tudor.Ambarus@microchip.com \
    --cc=boris.brezillon@collabora.com \
    --cc=linux-mtd@lists.infradead.org \
    --cc=richard@nod.at \
    --cc=rickaran@axis.com \
    --cc=vigneshr@ti.com \
    /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).