mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Marek Vasut <marek.vasut@gmail.com>
To: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Cc: computersforpeace@gmail.com, boris.brezillon@free-electrons.com,
	richard@nod.at, linux-mtd@lists.infradead.org,
	linux-kernel@vger.kernel.org, linux-spi@vger.kernel.org,
	broonie@kernel.org
Subject: Re: [PATCH 2/2] mtd: spi-nor: add a stateless method to support memory size above 128Mib
Date: Wed, 7 Dec 2016 17:32:45 +0100	[thread overview]
Message-ID: <6443cfb4-2d6d-c929-0d8a-053082bda506@gmail.com> (raw)
In-Reply-To: <6623dc41-f7a8-9788-fe01-7fe3d5b2062f@atmel.com>

On 12/07/2016 05:29 PM, Cyrille Pitchen wrote:
> Le 07/12/2016 à 17:20, Marek Vasut a écrit :
>> On 12/06/2016 05:52 PM, Cyrille Pitchen wrote:
>>> This patch provides an alternative mean to support memory above 16MiB
>>> (128Mib) by replacing 3byte address op codes by their associated 4byte
>>> address versions.
>>>
>>> Using the dedicated 4byte address op codes doesn't change the internal
>>> state of the SPI NOR memory as opposed to using other means such as
>>> updating a Base Address Register (BAR) and sending command to enter/leave
>>> the 4byte mode.
>>>
>>> Hence when a CPU reset occurs, early bootloaders don't need to be aware
>>> of BAR value or 4byte mode being enabled: they can still access the first
>>> 16MiB of the SPI NOR memory using the regular 3byte address op codes.
>>>
>>> Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
>>> Tested-by: Vignesh R <vigneshr@ti.com>
>>> ---
>>>  drivers/mtd/spi-nor/spi-nor.c | 114 ++++++++++++++++++++++++++++++++++--------
>>>  1 file changed, 93 insertions(+), 21 deletions(-)
>>>
>>> diff --git a/drivers/mtd/spi-nor/spi-nor.c b/drivers/mtd/spi-nor/spi-nor.c
>>> index 8abe134e174a..606c030c566d 100644
>>> --- a/drivers/mtd/spi-nor/spi-nor.c
>>> +++ b/drivers/mtd/spi-nor/spi-nor.c
>>> @@ -75,6 +75,10 @@ struct flash_info {
>>>  					 * bit. Must be used with
>>>  					 * SPI_NOR_HAS_LOCK.
>>>  					 */
>>> +#define SPI_NOR_4B_OPCODES	BIT(10)	/*
>>> +					 * Use dedicated 4byte address op codes
>>> +					 * to support memory size above 128Mib.
>>> +					 */
>>>  };
>>>  
>>>  #define JEDEC_MFR(info)	((info)->id[0])
>>> @@ -188,6 +192,91 @@ static inline struct spi_nor *mtd_to_spi_nor(struct mtd_info *mtd)
>>>  	return mtd->priv;
>>>  }
>>>  
>>> +
>>> +struct spi_nor_address_entry {
>>> +	u8	src_opcode;
>>> +	u8	dst_opcode;
>>> +};
>>> +
>>> +static u8 spi_nor_convert_opcode(u8 opcode,
>>> +				 const struct spi_nor_address_entry *entries,
>>> +				 size_t num_entries)
>>> +{
>>> +	int min, max;
>>> +
>>> +	/*
>>> +	 * This function implements a dichotomic search in the entries[]
>>> +	 * array indexed by src_opcode. Hence we assume that the entries[]
>>> +	 * array is sorted by src_opcode.
>>> +	 * The dichotomic search has a logarithmic complexity as opposed
>>> +	 * to a simple loop on all entires, which has a linear complexity:
>>> +	 * it means that when n is the number of entries in the input array,
>>> +	 * the dichotomic search performs O(log2(n)) comparisons whereas
>>> +	 * a simple loop performs O(n) comparisons.
>>> +	 */
>>> +	min = 0;
>>> +	max = num_entries - 1;
>>> +	while (min <= max) {
>>> +		int mid = (min + max) >> 1;
>>> +		const struct spi_nor_address_entry *entry = &entries[mid];
>>> +
>>> +		if (opcode == entry->src_opcode)
>>> +			return entry->dst_opcode;
>>> +
>>> +		if (opcode < entry->src_opcode)
>>> +			max = mid - 1;
>>> +		else
>>> +			min = mid + 1;
>>> +	}
>>
>> You have like 16 entries in that table, just do a linear search, this is
>> only complex for no benefit.
> 
> Well ok, I agree with you, it's too much overkill for what it does.
> For readiness purpose, what about a simple and straight forward switch()
> statement? Let's forget about cache-line and other memory/time optimizations :)

If you do a switch, you cannot walk a table which you pass in, see below.

>>> +	/* No conversion found */
>>> +	return opcode;
>>> +}
>>> +
>>> +static u8 spi_nor_3to4_opcode(u8 opcode)
>>> +{
>>> +	/* MUST be sorted by 3byte opcode (cf spi_nor_convert_opcode). */
>>> +#define ENTRY_3TO4(_opcode)	{ _opcode, _opcode##_4B }
>>> +	static const struct spi_nor_address_entry spi_nor_3to4_table[] = {
>>
>> You can make this static const struct const for extra constness :-)
>>
>>> +		ENTRY_3TO4(SPINOR_OP_PP),		/* 0x02 */
>>> +		ENTRY_3TO4(SPINOR_OP_READ),		/* 0x03 */
>>> +		ENTRY_3TO4(SPINOR_OP_READ_FAST),	/* 0x0b */
>>> +		ENTRY_3TO4(SPINOR_OP_BE_4K),		/* 0x20 */
>>> +		ENTRY_3TO4(SPINOR_OP_PP_1_1_4),		/* 0x32 */
>>> +		ENTRY_3TO4(SPINOR_OP_PP_1_4_4),		/* 0x38 */
>>> +		ENTRY_3TO4(SPINOR_OP_READ_1_1_2),	/* 0x3b */
>>> +		ENTRY_3TO4(SPINOR_OP_BE_32K),		/* 0x52 */
>>> +		ENTRY_3TO4(SPINOR_OP_READ_1_1_4),	/* 0x6b */
>>> +		ENTRY_3TO4(SPINOR_OP_READ_1_2_2),	/* 0xbb */
>>> +		ENTRY_3TO4(SPINOR_OP_SE),		/* 0xd8 */
>>> +		ENTRY_3TO4(SPINOR_OP_READ_1_4_4),	/* 0xeb */
>>
>> I'd probably break this into three smaller tables, read/program/erase
>> and then call something like:
>>
>> spi_nor_3to4_opcode(nor->read_opcode, read_opcode_table,
>>                     ARRAY_SIZE(read_opcode_table));
>>
>> This would further reduce the table size (heck, it'd probably fit into a
>> cacheline), so linear search would be more than enough.
>>
>>> +	};
>>> +#undef ENTRY_3TO4
>>> +
>>> +	return spi_nor_convert_opcode(opcode, spi_nor_3to4_table,
>>> +				      ARRAY_SIZE(spi_nor_3to4_table));
>>> +}
>>> +
>>> +static void spi_nor_set_4byte_opcodes(struct spi_nor *nor,
>>> +				      const struct flash_info *info)
>>> +{
>>> +	/* Do some manufacturer fixups first */
>>> +	switch (JEDEC_MFR(info)) {
>>> +	case SNOR_MFR_SPANSION:
>>> +		/* No small sector erase for 4-byte command set */
>>> +		nor->erase_opcode = SPINOR_OP_SE;
>>> +		nor->mtd.erasesize = info->sector_size;
>>> +		break;
>>> +
>>> +	default:
>>> +		break;
>>> +	}
>>> +
>>> +	nor->read_opcode	= spi_nor_3to4_opcode(nor->read_opcode);
>>> +	nor->program_opcode	= spi_nor_3to4_opcode(nor->program_opcode);
>>> +	nor->erase_opcode	= spi_nor_3to4_opcode(nor->erase_opcode);
>>> +}
>>> +
>>>  /* Enable/disable 4-byte addressing mode. */
>>>  static inline int set_4byte(struct spi_nor *nor, const struct flash_info *info,
>>>  			    int enable)
>>> @@ -1486,27 +1575,10 @@ int spi_nor_scan(struct spi_nor *nor, const char *name, enum read_mode mode)
>>>  	else if (mtd->size > 0x1000000) {
>>>  		/* enable 4-byte addressing if the device exceeds 16MiB */
>>>  		nor->addr_width = 4;
>>> -		if (JEDEC_MFR(info) == SNOR_MFR_SPANSION) {
>>> -			/* Dedicated 4-byte command set */
>>> -			switch (nor->flash_read) {
>>> -			case SPI_NOR_QUAD:
>>> -				nor->read_opcode = SPINOR_OP_READ_1_1_4_4B;
>>> -				break;
>>> -			case SPI_NOR_DUAL:
>>> -				nor->read_opcode = SPINOR_OP_READ_1_1_2_4B;
>>> -				break;
>>> -			case SPI_NOR_FAST:
>>> -				nor->read_opcode = SPINOR_OP_READ_FAST_4B;
>>> -				break;
>>> -			case SPI_NOR_NORMAL:
>>> -				nor->read_opcode = SPINOR_OP_READ_4B;
>>> -				break;
>>> -			}
>>> -			nor->program_opcode = SPINOR_OP_PP_4B;
>>> -			/* No small sector erase for 4-byte command set */
>>> -			nor->erase_opcode = SPINOR_OP_SE_4B;
>>> -			mtd->erasesize = info->sector_size;
>>> -		} else
>>> +		if (JEDEC_MFR(info) == SNOR_MFR_SPANSION ||
>>> +		    info->flags & SPI_NOR_4B_OPCODES)
>>> +			spi_nor_set_4byte_opcodes(nor, info);
>>> +		else
>>>  			set_4byte(nor, info, 1);
>>>  	} else {
>>>  		nor->addr_width = 3;
>>>
>>
>>
> 


-- 
Best regards,
Marek Vasut

  reply	other threads:[~2016-12-07 16:33 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-12-06 16:52 [PATCH 0/2] " Cyrille Pitchen
2016-12-06 16:52 ` [PATCH 1/2] mtd: spi-nor: rename SPINOR_OP_* macros of the 4-byte address op codes Cyrille Pitchen
2016-12-07  9:51   ` Mark Brown
2016-12-07 16:08     ` Marek Vasut
2016-12-06 16:52 ` [PATCH 2/2] mtd: spi-nor: add a stateless method to support memory size above 128Mib Cyrille Pitchen
2016-12-07 16:20   ` Marek Vasut
2016-12-07 16:29     ` Cyrille Pitchen
2016-12-07 16:32       ` Marek Vasut [this message]
2016-12-07 16:59         ` Cyrille Pitchen
2016-12-08  3:50           ` Marek Vasut

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=6443cfb4-2d6d-c929-0d8a-053082bda506@gmail.com \
    --to=marek.vasut@gmail.com \
    --cc=boris.brezillon@free-electrons.com \
    --cc=broonie@kernel.org \
    --cc=computersforpeace@gmail.com \
    --cc=cyrille.pitchen@atmel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mtd@lists.infradead.org \
    --cc=linux-spi@vger.kernel.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

Powered by JetHome