Re: [PATCH 1/3] spi: Add addr_width field to struct spi_message

[Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>]

Hi Zhiqiang,

On Tue, May 12, 2015 at 10:27 AM, Hou Zhiqiang <B48286@xxxxxxxxxxxxx> wrote:
>> -----Original Message-----
>> From: geert.uytterhoeven@xxxxxxxxx [mailto:geert.uytterhoeven@xxxxxxxxx]
>> On Behalf Of Geert Uytterhoeven
>> Sent: 2015年5月12日 16:12
>> To: Hou Zhiqiang-B48286
>> Cc: linux-spi; Mark Brown; Hu Mingkai-B21284
>> Subject: Re: [PATCH 1/3] spi: Add addr_width field to struct spi_message
>>
>> On Tue, May 12, 2015 at 10:06 AM, Hou Zhiqiang <B48286@xxxxxxxxxxxxx>
>> wrote:
>> >> From: geert.uytterhoeven@xxxxxxxxx
>> >> [mailto:geert.uytterhoeven@xxxxxxxxx]
>> >> On Behalf Of Geert Uytterhoeven
>> >> Sent: 2015年5月11日 15:43
>> >> To: Hou Zhiqiang-B48286
>> >> Cc: linux-spi; Mark Brown; Hu Mingkai-B21284
>> >> Subject: Re: [PATCH 1/3] spi: Add addr_width field to struct
>> >> spi_message
>> >>
>> >> On Sun, May 10, 2015 at 4:47 PM, Zhiqiang Hou <B48286@xxxxxxxxxxxxx>
>> >> wrote:
>> >> > From: Hou Zhiqiang <B48286@xxxxxxxxxxxxx>
>> >> >
>> >> > Make SPI controller can get address width of the SPI device.
>> >>
>> >> What does "address wdith" mean? From your other patches, I infer it's
>> >> either
>> >> 24 or 32 bit addressing as used by nor-jedec ("m25p80").
>> >> Why does the generic spi_message needs to have nor-jedec-specific
>> fields?
>> >
>> > Yes, the address width is the address width of spi flash.
>> > The reason to add this nor-jedec-specific field to spi_message is the
>> > eSPI controller of Freescale limits the MAX transaction
>> > length(TRANLEN) to 0x10000 each transaction, so if the data size
>> > transferred exceed the TRANLEN, the eSPI controller can only transfer
>> data for the first TRANLEN, and need to issue new write/read commands to
>> complete the remainders.
>> > So, eSPI controller need getting the address width of spi flash, and
>> > then calculate the new address to initialize the new write/read
>> commands.
>>
>> Shouldn't this be handled by the m25p80 driver instead?
>>
>> I see spi_master has a max_dma_len field.
>
> No, only the eSPI controller need to do this, it caused by the eSPI controller's
> feature and other SPI controller have not this TRANLEN limit.
> And it has no relation with DMA.

IC.

Still, your doing m25p80-specific handling in your spi master driver.
Worse, you cannot connect any other type of spi device to your spi controller,
right?

So it looks like we need something like i2c_adapter_quirks, where spi masters
can advertise their limitations?

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe linux-spi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html