Ask for help on SPI slave side driver support

Ask for help on SPI slave side driver support

[YouShenghui]

Hi Samsung expert,
I am working a project that need spi slave side driver support in smdk6410 in linux.
I am connecting two 6410 board with spi bus, one side as master, the other as slave. Data is passed from one side to the other.
I looked over spi related code in kernel and only find spi master support.
I see http://www.spinics.net/lists/arm-kernel/msg73403.html that said there was already spi slave support in Samsungs current code.
Could you give me a suggestion where I could find that code? I would like to port slave side support code to my code though it's not upstream yet.

thanks
ysh

Re: Ask for help on SPI slave side driver support

[Tomasz Figa]

Hi,

[Ccing Mark Brown, maintainer of Linux SPI subsystem]

On Monday 30 of September 2013 00:23:32 YouShenghui wrote:
> Hi Samsung expert,
> I am working a project that need spi slave side driver support in
> smdk6410 in linux. I am connecting two 6410 board with spi bus, one
> side as master, the other as slave. Data is passed from one side to the
> other. I looked over spi related code in kernel and only find spi
> master support. I see
> http://www.spinics.net/lists/arm-kernel/msg73403.html that said there
> was already spi slave support in Samsungs current code. Could you give
> me a suggestion where I could find that code? I would like to port
> slave side support code to my code though it's not upstream yet.

Mark will probably know better, but AFAIK there is no API for slave mode 
SPI available in Linux kernel. Please correct me if I'm wrong.

The code mentioned in the post from your link was some internal vendor 
code, which is unlikely to be available publicly. However support for 
slave mode from hardware side seems more or less similar to master mode, 
which means that assuming you can add some kernel API for it, the 
modification to the spi-s3c64xx wouldn't be too big.

Best regards,
Tomasz

Re:Re: Ask for help on SPI slave side driver support

[YouShenghui]

At 2013-09-30 07:20:42,"Tomasz Figa" <tomasz.figa@gmail.com> wrote: >Hi, > >[Ccing Mark Brown, maintainer of Linux SPI subsystem] > >On Monday 30 of September 2013 00:23:32 YouShenghui wrote: >> Hi Samsung expert, >> I am working a project that need spi slave side driver support in >> smdk6410 in linux. I am connecting two 6410 board with spi bus, one >> side as master, the other as slave. Data is passed from one side to the >> other. I looked over spi related code in kernel and only find spi >> master support. I see >> http://www.spinics.net/lists/arm-kernel/msg73403.html that said there >> was already spi slave support in Samsungs current code. Could you give >> me a suggestion where I could find that code? I would like to port >> slave side support code to my code though it's not upstre
 am yet. > >Mark will probably know better, but AFAIK there is no API for slave mode >SPI available in Linux kernel. Please correct me if I'm wrong. > >The code mentioned in the post from your link was some internal vendor >code, which is unlikely to be available publicly. However support for >slave mode from hardware side seems more or less similar to master mode, >which means that assuming you can add some kernel API for it, the >modification to the spi-s3c64xx wouldn't be too big. > >Best regards, >Tomasz >
Thanks for your answer.
Yes, It will not be a big patch. But I feel hard to add the support as a application developer so I want if they could send me a small patch to make it work.
ysh

Re: Ask for help on SPI slave side driver support

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 928 bytes --]

On Mon, Sep 30, 2013 at 01:20:42AM +0200, Tomasz Figa wrote:

> Mark will probably know better, but AFAIK there is no API for slave mode 
> SPI available in Linux kernel. Please correct me if I'm wrong.

That's correct.

> The code mentioned in the post from your link was some internal vendor 
> code, which is unlikely to be available publicly. However support for 
> slave mode from hardware side seems more or less similar to master mode, 
> which means that assuming you can add some kernel API for it, the 
> modification to the spi-s3c64xx wouldn't be too big.

There were some people working on slave mode (or at least talking about
it) but I don't think they ever sent it.  It'd probably just be the same
as master mode but with transfers triggered from /CS instead of from
driver action.  The biggest trick would be making sure the transfer is
running before the master actually starts sending data.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Re: Ask for help on SPI slave side driver support

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 1450 bytes --]

On Sat, Oct 05, 2013 at 12:25:10AM +0800, YouShenghui wrote:
> At 2013-09-30 17:33:17,"Mark Brown" <broonie@kernel.org> wrote: >On Mon, Sep 30, 2013 at 01:20:42AM +0200, Tomasz Figa wrote: > >> Mark will probably know better, but AFAIK there is no API for slave mode  >> SPI available in Linux kernel. Please correct me if I'm wrong. > >That's correct. > >> The code mentioned in the post from your link was some internal vendor  >> code, which is unlikely to be available publicly. However support for  >> slave mode from hardware side seems more or less similar to master mode,  >> which means that assuming you can add some kernel API for it, the  >> modification to the spi-s3c64xx wouldn't be too big. > >There were some people working on slave mode (or at least talking about >it) but I don't think they ever sent it.  It'd probably just be the same >as master mode but with transfers triggered from /CS instead of from >driver action.  The biggest trick would be making sure the transfer is >running before the master actually starts sending data

The formatting of your mail is really bad, I can barely read it.  Please
check the configuration of your mail program, there are some suggstions
for common clients in Documentation/email-clients.txt.

> Thanks for Mark for your answer.
> But how could I trigger from /CS? Could I enable transfer in open syscall?

I'd expect this to be done in-kernel, at least in the first instance.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Re: Re: Ask for help on SPI slave side driver support

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 913 bytes --]

On Mon, Oct 14, 2013 at 03:05:58AM +0800, YouShenghui wrote:

Please fix your mailer to word wrap within paragraphs, your messages are
very hard to read.

> Ok, I have one question when reading drivers/spi/spi-s3c64xx.c
> In s3c64xx_spi_setup, cs->line(S3C64XX_GPC(3)) is set to GPIOF_OUT_INIT_HIGH, that means an general output pin.
> But the other three(clk, mosi, miso) was set to 2 by s3c64xx_spi0_cfg_gpio. This is just what is needed by SPI.
> My question is why wasn't cs->line also set to 2 to use SPI CS0?

In common with many drivers this driver controls the chp select as a
GPIO.  Generally the hardware is less flexible than Linux can be if it
implements automatic chip select handling at all.

> When I want to use spi as slave side, should I set S3C64XX_GPC(3)'s cfg of  to 0(input) or 2?

I don't think the controller has any slave support natively so you'd
probably need to put it into GPIO mode.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Ask for help on SPI slave side driver support

[ysh]

On 2013-10-14 19:25, Mark Brown wrote:
> On Mon, Oct 14, 2013 at 03:05:58AM +0800, YouShenghui wrote:
>
> Please fix your mailer to word wrap within paragraphs, your messages are
> very hard to read.
>
>> Ok, I have one question when reading drivers/spi/spi-s3c64xx.c
>> In s3c64xx_spi_setup, cs->line(S3C64XX_GPC(3)) is set to GPIOF_OUT_INIT_HIGH, that means an general output pin.
>> But the other three(clk, mosi, miso) was set to 2 by s3c64xx_spi0_cfg_gpio. This is just what is needed by SPI.
>> My question is why wasn't cs->line also set to 2 to use SPI CS0?
>
> In common with many drivers this driver controls the chp select as a
> GPIO.  Generally the hardware is less flexible than Linux can be if it
> implements automatic chip select handling at all.
>
>> When I want to use spi as slave side, should I set S3C64XX_GPC(3)'s cfg of  to 0(input) or 2?
>
> I don't think the controller has any slave support natively so you'd
> probably need to put it into GPIO mode.
Got it, then what about the setting of clk, mosi, miso pin in slave mode?
Do you mean I need to simulate spi slave side using four gpios?

thanks
zduan

Re: Ask for help on SPI slave side driver support

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 359 bytes --]

On Mon, Oct 14, 2013 at 10:51:06PM +0800, ysh wrote:

> Got it, then what about the setting of clk, mosi, miso pin in slave mode?
> Do you mean I need to simulate spi slave side using four gpios?

Hrm, now I think about it clock is going to be a problem...  you're
probably stuck with bitbanging unless there's some controller support
which won't be awesome.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]