[PATCH 4/6] mtd: m25p80: change the read function to read page by page

[Grant Likely]

On Mon, Jul 26, 2010 at 1:33 AM, Hu Mingkai-B21284 <B21284 at freescale.com> wrote:
>
>
>> -----Original Message-----
>> From: Grant Likely [mailto:glikely at secretlab.ca] On Behalf Of
>> Grant Likely
>> Sent: Monday, July 26, 2010 8:30 AM
>> To: Hu Mingkai-B21284
>> Cc: linuxppc-dev at ozlabs.org; galak at kernel.crashing.org; Zang
>> Roy-R61911
>> Subject: Re: [PATCH 4/6] mtd: m25p80: change the read
>> function to read page by page
>>
>> On Tue, Jul 20, 2010 at 10:08:23AM +0800, Mingkai Hu wrote:
>> > For Freescale's eSPI controller, the max transaction length
>> one time
>> > is limitted by the SPCOM[TRANSLEN] field which is 0x10000. When used
>> > mkfs.ext2 command to create ext2 filesystem on the flash, the read
>> > length will exceed the max value of the SPCOM[TRANSLEN] field, so
>> > change the read function to read page by page.
>> >
>> > For other SPI flash driver, also needed to change the read
>> function if
>> > used the eSPI controller.
>> >
>> > Signed-off-by: Mingkai Hu <Mingkai.hu at freescale.com>
>> > ---
>> >  drivers/mtd/devices/m25p80.c |   18 ++++++++++++++----
>> >  1 files changed, 14 insertions(+), 4 deletions(-)
>> >
>> > diff --git a/drivers/mtd/devices/m25p80.c
>> > b/drivers/mtd/devices/m25p80.c index 81e49a9..6cbe6b1 100644
>> > --- a/drivers/mtd/devices/m25p80.c
>> > +++ b/drivers/mtd/devices/m25p80.c
>> > @@ -317,6 +317,7 @@ static int m25p80_read(struct mtd_info
>> *mtd, loff_t from, size_t len,
>> >     struct m25p *flash = mtd_to_m25p(mtd);
>> >     struct spi_transfer t[2];
>> >     struct spi_message m;
>> > +   u32 i, page_size = 0;
>> >
>> >     DEBUG(MTD_DEBUG_LEVEL2, "%s: %s %s 0x%08x, len %zd\n",
>> >                     dev_name(&flash->spi->dev), __func__,
>> "from", @@ -341,7 +342,6 @@
>> > static int m25p80_read(struct mtd_info *mtd, loff_t from,
>> size_t len,
>> >     spi_message_add_tail(&t[0], &m);
>> >
>> >     t[1].rx_buf = buf;
>> > -   t[1].len = len;
>> >     spi_message_add_tail(&t[1], &m);
>> >
>> >     /* Byte count starts at zero. */
>> > @@ -364,11 +364,21 @@ static int m25p80_read(struct mtd_info *mtd,
>> > loff_t from, size_t len,
>> >
>> >     /* Set up the write data buffer. */
>> >     flash->command[0] = OPCODE_READ;
>> > -   m25p_addr2cmd(flash, from, flash->command);
>> >
>> > -   spi_sync(flash->spi, &m);
>> > +   for (i = page_size; i < len; i += page_size) {
>> > +           page_size = len - i;
>> > +           if (page_size > flash->page_size)
>> > +                   page_size = flash->page_size;
>> >
>> > -   *retlen = m.actual_length - m25p_cmdsz(flash) -
>> FAST_READ_DUMMY_BYTE;
>> > +           m25p_addr2cmd(flash, from + i, flash->command);
>> > +           t[1].len = page_size;
>> > +           t[1].rx_buf = buf + i;
>> > +
>> > +           spi_sync(flash->spi, &m);
>> > +
>> > +           *retlen += m.actual_length - m25p_cmdsz(flash)
>> > +                   - FAST_READ_DUMMY_BYTE;
>> > +   }
>>
>> This patch seems to cripple all users just because eSPI
>> cannot handle it.  Am I reading it wrong?
>>
>
> You're right, this will cripple all users.
>
> At first, I want to contain this specific code with CONFIG_SPI_FSL_ESPI
> option,
> or register a specific read function like:
>
> #ifdef CONFIG_SPI_FSL_ESPI
>        flash->mtd.read = m25p80_read_espi;
> #endif
>
> but this will make the code looks arguly,

And it is multiplatform-unfriendly.

> are there other ways?

You need to select the correct transfer behaviour at driver probe
time.  You'll need a way to indicate that the spi_master isn't able to
handle your transfer request.  I'd need to think about it more to come
up with a concrete suggestion.

g.