Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1622971 > unrolled thread

[PATCH RFC 0/5] *** SPI Slave mode support ***

Started by<jiada_wang@mentor.com>
First post2017-04-13 14:20 +0200
Last post2017-04-27 08:50 +0200
Articles 13 — 5 participants

Back to article view | Back to linux.kernel

Contents

  [PATCH RFC 0/5] *** SPI Slave mode support *** <jiada_wang@mentor.com> - 2017-04-13 14:20 +0200
    [PATCH RFC 2/5] spi: spidev: use different name for SPI controller slave mode device <jiada_wang@mentor.com> - 2017-04-13 14:20 +0200
    Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Mark Brown <broonie@kernel.org> - 2017-04-13 15:00 +0200
      Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Geert Uytterhoeven <geert@linux-m68k.org> - 2017-04-13 21:50 +0200
        Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Jiada Wang <jiada_wang@mentor.com> - 2017-04-14 07:40 +0200
          Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Geert Uytterhoeven <geert@linux-m68k.org> - 2017-04-24 13:00 +0200
            Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Jiada Wang <jiada_wang@mentor.com> - 2017-04-24 14:50 +0200
              Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Geert Uytterhoeven <geert@linux-m68k.org> - 2017-04-24 15:20 +0200
                Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Jiada Wang <jiada_wang@mentor.com> - 2017-04-25 10:00 +0200
                  Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Uwe Kleine-König          <u.kleine-koenig@pengutronix.de> - 2017-04-25 10:20 +0200
                  Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Geert Uytterhoeven <geert@linux-m68k.org> - 2017-04-25 10:20 +0200
            Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Mark Brown <broonie@kernel.org> - 2017-04-25 12:40 +0200
              Re: [PATCH RFC 0/5] *** SPI Slave mode support *** Jiada Wang <jiada_wang@mentor.com> - 2017-04-27 08:50 +0200

#1622971 — [PATCH RFC 0/5] *** SPI Slave mode support ***

From<jiada_wang@mentor.com>
Date2017-04-13 14:20 +0200
Subject[PATCH RFC 0/5] *** SPI Slave mode support ***
Message-ID<tvLTc-2Sx-7@gated-at.bofh.it>
From: Jiada Wang <jiada_wang@mentor.com>

v1:
  add Slave mode support in SPI core
  spidev create slave device when SPI controller work in slave mode
  spi-imx support to work in slave mode

Jiada Wang (5):
  spi: core: add support to work in Slave mode
  spi: spidev: use different name for SPI controller slave mode device
  spi: imx: add selection for iMX53 and iMX6 controller
  ARM: dts: imx: change compatiblity for SPI controllers on imx53 later
    soc
  spi: imx: Add support for SPI Slave mode for imx53 and imx6 chips

 Documentation/devicetree/bindings/spi/spi-bus.txt |  27 ++-
 Documentation/spi/spi-summary                     |  19 +-
 arch/arm/boot/dts/imx53.dtsi                      |   4 +-
 arch/arm/boot/dts/imx6q.dtsi                      |   2 +-
 arch/arm/boot/dts/imx6qdl.dtsi                    |   8 +-
 arch/arm/boot/dts/imx6sl.dtsi                     |   8 +-
 arch/arm/boot/dts/imx6sx.dtsi                     |   8 +-
 arch/arm/boot/dts/imx6ul.dtsi                     |   8 +-
 drivers/spi/Kconfig                               |  14 +-
 drivers/spi/spi-imx.c                             | 216 ++++++++++++++++++++--
 drivers/spi/spi.c                                 |  23 ++-
 drivers/spi/spidev.c                              |  15 +-
 include/linux/spi/spi.h                           |  15 ++
 13 files changed, 310 insertions(+), 57 deletions(-)

-- 
2.7.4

[toc] | [next] | [standalone]

#1622973 — [PATCH RFC 2/5] spi: spidev: use different name for SPI controller slave mode device

From<jiada_wang@mentor.com>
Date2017-04-13 14:20 +0200
Subject[PATCH RFC 2/5] spi: spidev: use different name for SPI controller slave mode device
Message-ID<tvLTd-2Sx-29@gated-at.bofh.it>
In reply to#1622971
From: Jiada Wang <jiada_wang@mentor.com>

SPI bus controller has started to support to work in slave mode,
for device SPI controller itself works in slave mode, use name
'spidev[bus]-slv' as its name to differentiate from other
SPI devices

Signed-off-by: Jiada Wang <jiada_wang@mentor.com>
---
 drivers/spi/spidev.c | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/drivers/spi/spidev.c b/drivers/spi/spidev.c
index 9e2e099..e2996fb 100644
--- a/drivers/spi/spidev.c
+++ b/drivers/spi/spidev.c
@@ -781,9 +781,18 @@ static int spidev_probe(struct spi_device *spi)
 		struct device *dev;
 
 		spidev->devt = MKDEV(SPIDEV_MAJOR, minor);
-		dev = device_create(spidev_class, &spi->dev, spidev->devt,
-				    spidev, "spidev%d.%d",
-				    spi->master->bus_num, spi->chip_select);
+		if (spi->slave_mode)
+			dev = device_create(spidev_class, &spi->dev,
+					    spidev->devt, spidev,
+					    "spidev%d-slv",
+					    spi->master->bus_num);
+		else
+			dev = device_create(spidev_class, &spi->dev,
+					    spidev->devt, spidev,
+					    "spidev%d.%d",
+					    spi->master->bus_num,
+					    spi->chip_select);
+
 		status = PTR_ERR_OR_ZERO(dev);
 	} else {
 		dev_dbg(&spi->dev, "no minor number available!\n");
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1622998

FromMark Brown <broonie@kernel.org>
Date2017-04-13 15:00 +0200
Message-ID<tvMvT-388-1@gated-at.bofh.it>
In reply to#1622971

[Multipart message — attachments visible in raw view] — view raw

On Thu, Apr 13, 2017 at 05:13:59AM -0700, jiada_wang@mentor.com wrote:
> From: Jiada Wang <jiada_wang@mentor.com>
> 
> v1:
>   add Slave mode support in SPI core
>   spidev create slave device when SPI controller work in slave mode
>   spi-imx support to work in slave mode

Adding Geert who also had a series doing this in progress that was
getting very near to being merged.

> 
> Jiada Wang (5):
>   spi: core: add support to work in Slave mode
>   spi: spidev: use different name for SPI controller slave mode device
>   spi: imx: add selection for iMX53 and iMX6 controller
>   ARM: dts: imx: change compatiblity for SPI controllers on imx53 later
>     soc
>   spi: imx: Add support for SPI Slave mode for imx53 and imx6 chips
> 
>  Documentation/devicetree/bindings/spi/spi-bus.txt |  27 ++-
>  Documentation/spi/spi-summary                     |  19 +-
>  arch/arm/boot/dts/imx53.dtsi                      |   4 +-
>  arch/arm/boot/dts/imx6q.dtsi                      |   2 +-
>  arch/arm/boot/dts/imx6qdl.dtsi                    |   8 +-
>  arch/arm/boot/dts/imx6sl.dtsi                     |   8 +-
>  arch/arm/boot/dts/imx6sx.dtsi                     |   8 +-
>  arch/arm/boot/dts/imx6ul.dtsi                     |   8 +-
>  drivers/spi/Kconfig                               |  14 +-
>  drivers/spi/spi-imx.c                             | 216 ++++++++++++++++++++--
>  drivers/spi/spi.c                                 |  23 ++-
>  drivers/spi/spidev.c                              |  15 +-
>  include/linux/spi/spi.h                           |  15 ++
>  13 files changed, 310 insertions(+), 57 deletions(-)
> 
> -- 
> 2.7.4
> 
>

[toc] | [prev] | [next] | [standalone]

#1623304

FromGeert Uytterhoeven <geert@linux-m68k.org>
Date2017-04-13 21:50 +0200
Message-ID<tvSUG-7Bg-15@gated-at.bofh.it>
In reply to#1622998
On Thu, Apr 13, 2017 at 2:59 PM, Mark Brown <broonie@kernel.org> wrote:
> On Thu, Apr 13, 2017 at 05:13:59AM -0700, jiada_wang@mentor.com wrote:
>> From: Jiada Wang <jiada_wang@mentor.com>
>>
>> v1:
>>   add Slave mode support in SPI core
>>   spidev create slave device when SPI controller work in slave mode
>>   spi-imx support to work in slave mode
>
> Adding Geert who also had a series doing this in progress that was
> getting very near to being merged.

Thank you!

Actually my plan is to fix the last remaining issues and resubmit for v4.13.

References:
  - v2: https://lkml.org/lkml/2016/9/12/1065
  - v1: https://lkml.org/lkml/2016/6/22/423

BTW Jiada, what's your use case? Just spidev?

Thx!

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

[toc] | [prev] | [next] | [standalone]

#1623523

FromJiada Wang <jiada_wang@mentor.com>
Date2017-04-14 07:40 +0200
Message-ID<tw27E-5Bu-11@gated-at.bofh.it>
In reply to#1623304
Hello Geert

On 04/13/2017 12:47 PM, Geert Uytterhoeven wrote:
> On Thu, Apr 13, 2017 at 2:59 PM, Mark Brown<broonie@kernel.org>  wrote:
>> On Thu, Apr 13, 2017 at 05:13:59AM -0700, jiada_wang@mentor.com wrote:
>>> From: Jiada Wang<jiada_wang@mentor.com>
>>>
>>> v1:
>>>    add Slave mode support in SPI core
>>>    spidev create slave device when SPI controller work in slave mode
>>>    spi-imx support to work in slave mode
>> Adding Geert who also had a series doing this in progress that was
>> getting very near to being merged.
> Thank you!
>
> Actually my plan is to fix the last remaining issues and resubmit for v4.13.
I noticed your patch set for SPI slave support,
(I am sure you can find out some of the change
in this patch set is based on your work).
we have similar requirement to add slave mode support to ecspi IP on 
imx6 Soc.

Our use case is to use spidev as an interface to communicate with 
external SPI master devices.
meanwhile the SPI bus controller can also act as master device to send 
data to other
SPI slave devices on the board.

I found in your implementation, SPI bus controller is limited to either 
work in master mode or
slave mode, is there any reasoning to not configure SPI mode based on 
SPI devices use case?


Thanks,
Jiada

> References:
>    - v2: https://lkml.org/lkml/2016/9/12/1065
>    - v1: https://lkml.org/lkml/2016/6/22/423
>
> BTW Jiada, what's your use case? Just spidev?
>
> Thx!
>
> Gr{oetje,eeting}s,
>
>                          Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
>
> 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

[toc] | [prev] | [next] | [standalone]

#1629427

FromGeert Uytterhoeven <geert@linux-m68k.org>
Date2017-04-24 13:00 +0200
Message-ID<tzJSO-5YI-27@gated-at.bofh.it>
In reply to#1623523
Hi Jiada,

On Fri, Apr 14, 2017 at 7:39 AM, Jiada Wang <jiada_wang@mentor.com> wrote:
> On 04/13/2017 12:47 PM, Geert Uytterhoeven wrote:
>> On Thu, Apr 13, 2017 at 2:59 PM, Mark Brown<broonie@kernel.org>  wrote:
>>> On Thu, Apr 13, 2017 at 05:13:59AM -0700, jiada_wang@mentor.com wrote:
>>>> From: Jiada Wang<jiada_wang@mentor.com>
>>>>
>>>> v1:
>>>>    add Slave mode support in SPI core
>>>>    spidev create slave device when SPI controller work in slave mode
>>>>    spi-imx support to work in slave mode
>>>
>>> Adding Geert who also had a series doing this in progress that was
>>> getting very near to being merged.
>>
>> Thank you!
>>
>> Actually my plan is to fix the last remaining issues and resubmit for
>> v4.13.
>
> I noticed your patch set for SPI slave support,
> (I am sure you can find out some of the change
> in this patch set is based on your work).
> we have similar requirement to add slave mode support to ecspi IP on imx6
> Soc.
>
> Our use case is to use spidev as an interface to communicate with external
> SPI master devices.
> meanwhile the SPI bus controller can also act as master device to send data
> to other
> SPI slave devices on the board.

That sounds a bit hackish to me. SPI was never meant to be a multi-master bus.
While it can be done, you will need external synchronization (signals) to
avoid conflicts between the SPI masters.

> I found in your implementation, SPI bus controller is limited to either work
> in master mode or
> slave mode, is there any reasoning to not configure SPI mode based on SPI
> devices use case?

If you really need both master and slave support, you can use 2 subnodes
in DT, the first representing the master, the second the slave.

Mark, what's your opinion about this?

Thanks!

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

[toc] | [prev] | [next] | [standalone]

#1629505

FromJiada Wang <jiada_wang@mentor.com>
Date2017-04-24 14:50 +0200
Message-ID<tzLBh-72B-35@gated-at.bofh.it>
In reply to#1629427
Hello Geert

On 04/24/2017 03:55 AM, Geert Uytterhoeven wrote:
> Hi Jiada,
>
> On Fri, Apr 14, 2017 at 7:39 AM, Jiada Wang<jiada_wang@mentor.com>  wrote:
>> On 04/13/2017 12:47 PM, Geert Uytterhoeven wrote:
>>> On Thu, Apr 13, 2017 at 2:59 PM, Mark Brown<broonie@kernel.org>   wrote:
>>>> On Thu, Apr 13, 2017 at 05:13:59AM -0700, jiada_wang@mentor.com wrote:
>>>>> From: Jiada Wang<jiada_wang@mentor.com>
>>>>>
>>>>> v1:
>>>>>     add Slave mode support in SPI core
>>>>>     spidev create slave device when SPI controller work in slave mode
>>>>>     spi-imx support to work in slave mode
>>>> Adding Geert who also had a series doing this in progress that was
>>>> getting very near to being merged.
>>> Thank you!
>>>
>>> Actually my plan is to fix the last remaining issues and resubmit for
>>> v4.13.
>> I noticed your patch set for SPI slave support,
>> (I am sure you can find out some of the change
>> in this patch set is based on your work).
>> we have similar requirement to add slave mode support to ecspi IP on imx6
>> Soc.
>>
>> Our use case is to use spidev as an interface to communicate with external
>> SPI master devices.
>> meanwhile the SPI bus controller can also act as master device to send data
>> to other
>> SPI slave devices on the board.
> That sounds a bit hackish to me. SPI was never meant to be a multi-master bus.
> While it can be done, you will need external synchronization (signals) to
> avoid conflicts between the SPI masters.
It doesn't need to be a multi-master bus,
for example A is master device for slave device B.
while B has its own slave device C
for each SPI connection A <=> B, and B <=> C, there is only one master 
device.

and I think from use case point of view, it's very normal,
one CPU upon receives command from external SPI master device,
it writes data to its own slave device (EEPROM) connected to it.

Thanks,
Jiada
>> I found in your implementation, SPI bus controller is limited to either work
>> in master mode or
>> slave mode, is there any reasoning to not configure SPI mode based on SPI
>> devices use case?
> If you really need both master and slave support, you can use 2 subnodes
> in DT, the first representing the master, the second the slave.
>
> Mark, what's your opinion about this?
>
> Thanks!
>
> Gr{oetje,eeting}s,
>
>                          Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
>
> 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

[toc] | [prev] | [next] | [standalone]

#1629533

FromGeert Uytterhoeven <geert@linux-m68k.org>
Date2017-04-24 15:20 +0200
Message-ID<tzM4h-7tv-15@gated-at.bofh.it>
In reply to#1629505
Hi Jiada,

On Mon, Apr 24, 2017 at 2:48 PM, Jiada Wang <jiada_wang@mentor.com> wrote:
> On 04/24/2017 03:55 AM, Geert Uytterhoeven wrote:
>> On Fri, Apr 14, 2017 at 7:39 AM, Jiada Wang<jiada_wang@mentor.com>  wrote:
>>> On 04/13/2017 12:47 PM, Geert Uytterhoeven wrote:
>>>> On Thu, Apr 13, 2017 at 2:59 PM, Mark Brown<broonie@kernel.org>   wrote:
>>>>> On Thu, Apr 13, 2017 at 05:13:59AM -0700, jiada_wang@mentor.com wrote:
>>>>>> From: Jiada Wang<jiada_wang@mentor.com>
>>>>>>
>>>>>> v1:
>>>>>>     add Slave mode support in SPI core
>>>>>>     spidev create slave device when SPI controller work in slave mode
>>>>>>     spi-imx support to work in slave mode
>>>>>
>>>>> Adding Geert who also had a series doing this in progress that was
>>>>> getting very near to being merged.
>>>>
>>>> Thank you!
>>>>
>>>> Actually my plan is to fix the last remaining issues and resubmit for
>>>> v4.13.
>>>
>>> I noticed your patch set for SPI slave support,
>>> (I am sure you can find out some of the change
>>> in this patch set is based on your work).
>>> we have similar requirement to add slave mode support to ecspi IP on imx6
>>> Soc.
>>>
>>> Our use case is to use spidev as an interface to communicate with
>>> external
>>> SPI master devices.
>>> meanwhile the SPI bus controller can also act as master device to send
>>> data
>>> to other
>>> SPI slave devices on the board.
>>
>> That sounds a bit hackish to me. SPI was never meant to be a multi-master
>> bus.
>> While it can be done, you will need external synchronization (signals) to
>> avoid conflicts between the SPI masters.
>
> It doesn't need to be a multi-master bus,
> for example A is master device for slave device B.
> while B has its own slave device C
> for each SPI connection A <=> B, and B <=> C, there is only one master
> device.
>
> and I think from use case point of view, it's very normal,
> one CPU upon receives command from external SPI master device,
> it writes data to its own slave device (EEPROM) connected to it.

So "A <=> B" and "B <=> C" are two distinct SPI buses?
Or do they share some signals?

Your comment seems to suggest otherwise:

> > > I found in your implementation, SPI bus controller is limited to either work in master mode or
> > > slave mode, is there any reasoning to not configure SPI mode based on SPI devices use case?

If they are distinct, it should work. Then B has two SPI controllers: one slave
controller controlled by A, and one master controller to control C.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

[toc] | [prev] | [next] | [standalone]

#1630264

FromJiada Wang <jiada_wang@mentor.com>
Date2017-04-25 10:00 +0200
Message-ID<tA3y9-1M8-3@gated-at.bofh.it>
In reply to#1629533
Hi Geert

On 04/24/2017 06:10 AM, Geert Uytterhoeven wrote:
> Hi Jiada,
>
> On Mon, Apr 24, 2017 at 2:48 PM, Jiada Wang<jiada_wang@mentor.com>  wrote:
>> On 04/24/2017 03:55 AM, Geert Uytterhoeven wrote:
>>> On Fri, Apr 14, 2017 at 7:39 AM, Jiada Wang<jiada_wang@mentor.com>   wrote:
>>>> On 04/13/2017 12:47 PM, Geert Uytterhoeven wrote:
>>>>> On Thu, Apr 13, 2017 at 2:59 PM, Mark Brown<broonie@kernel.org>    wrote:
>>>>>> On Thu, Apr 13, 2017 at 05:13:59AM -0700, jiada_wang@mentor.com wrote:
>>>>>>> From: Jiada Wang<jiada_wang@mentor.com>
>>>>>>>
>>>>>>> v1:
>>>>>>>      add Slave mode support in SPI core
>>>>>>>      spidev create slave device when SPI controller work in slave mode
>>>>>>>      spi-imx support to work in slave mode
>>>>>> Adding Geert who also had a series doing this in progress that was
>>>>>> getting very near to being merged.
>>>>> Thank you!
>>>>>
>>>>> Actually my plan is to fix the last remaining issues and resubmit for
>>>>> v4.13.
>>>> I noticed your patch set for SPI slave support,
>>>> (I am sure you can find out some of the change
>>>> in this patch set is based on your work).
>>>> we have similar requirement to add slave mode support to ecspi IP on imx6
>>>> Soc.
>>>>
>>>> Our use case is to use spidev as an interface to communicate with
>>>> external
>>>> SPI master devices.
>>>> meanwhile the SPI bus controller can also act as master device to send
>>>> data
>>>> to other
>>>> SPI slave devices on the board.
>>> That sounds a bit hackish to me. SPI was never meant to be a multi-master
>>> bus.
>>> While it can be done, you will need external synchronization (signals) to
>>> avoid conflicts between the SPI masters.
>> It doesn't need to be a multi-master bus,
>> for example A is master device for slave device B.
>> while B has its own slave device C
>> for each SPI connection A<=>  B, and B<=>  C, there is only one master
>> device.
>>
>> and I think from use case point of view, it's very normal,
>> one CPU upon receives command from external SPI master device,
>> it writes data to its own slave device (EEPROM) connected to it.
> So "A<=>  B" and "B<=>  C" are two distinct SPI buses?
> Or do they share some signals?
>
> Your comment seems to suggest otherwise:
the use case of
"A (master) <=> B (slave)", "B (master) <=> C(slave)", do share MISO and 
MOSI lines,
but there is no SS line between A and C. so for each SPI slave device, 
there is only one
master device.

so I think the question becomes whether the above mentioned hardware 
setup is valid or not.

Thanks,
Jiada
>>>> I found in your implementation, SPI bus controller is limited to either work in master mode or
>>>> slave mode, is there any reasoning to not configure SPI mode based on SPI devices use case?
> If they are distinct, it should work. Then B has two SPI controllers: one slave
> controller controlled by A, and one master controller to control C.
>
> Gr{oetje,eeting}s,
>
>                          Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
>
> 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

[toc] | [prev] | [next] | [standalone]

#1630274

FromUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Date2017-04-25 10:20 +0200
Message-ID<tA3Rw-28G-3@gated-at.bofh.it>
In reply to#1630264
Hello Jiada,

On Tue, Apr 25, 2017 at 12:56:23AM -0700, Jiada Wang wrote:
> the use case of
> "A (master) <=> B (slave)", "B (master) <=> C(slave)", do share MISO and
> MOSI lines,
> but there is no SS line between A and C. so for each SPI slave device, there
> is only one
> master device.

So you need a mutex to make A not use the bus while B communicates to C.
Otherwise you have two drivers on MOSI (A and B) and MISO (B and C).

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |

[toc] | [prev] | [next] | [standalone]

#1630283

FromGeert Uytterhoeven <geert@linux-m68k.org>
Date2017-04-25 10:20 +0200
Message-ID<tA3Rx-28G-29@gated-at.bofh.it>
In reply to#1630264
Hi Jiada,

On Tue, Apr 25, 2017 at 9:56 AM, Jiada Wang <jiada_wang@mentor.com> wrote:
> On 04/24/2017 06:10 AM, Geert Uytterhoeven wrote:
>> On Mon, Apr 24, 2017 at 2:48 PM, Jiada Wang<jiada_wang@mentor.com>  wrote:
>>> On 04/24/2017 03:55 AM, Geert Uytterhoeven wrote:
>>>> On Fri, Apr 14, 2017 at 7:39 AM, Jiada Wang<jiada_wang@mentor.com>
>>>> wrote:
>>>>> Our use case is to use spidev as an interface to communicate with
>>>>> external
>>>>> SPI master devices.
>>>>> meanwhile the SPI bus controller can also act as master device to send
>>>>> data
>>>>> to other
>>>>> SPI slave devices on the board.
>>>>
>>>> That sounds a bit hackish to me. SPI was never meant to be a
>>>> multi-master
>>>> bus.
>>>> While it can be done, you will need external synchronization (signals)
>>>> to
>>>> avoid conflicts between the SPI masters.
>>>
>>> It doesn't need to be a multi-master bus,
>>> for example A is master device for slave device B.
>>> while B has its own slave device C
>>> for each SPI connection A<=>  B, and B<=>  C, there is only one master
>>> device.
>>>
>>> and I think from use case point of view, it's very normal,
>>> one CPU upon receives command from external SPI master device,
>>> it writes data to its own slave device (EEPROM) connected to it.
>>
>> So "A<=>  B" and "B<=>  C" are two distinct SPI buses?
>> Or do they share some signals?
>>
>> Your comment seems to suggest otherwise:
>
> the use case of
> "A (master) <=> B (slave)", "B (master) <=> C(slave)", do share MISO and
> MOSI lines,
> but there is no SS line between A and C. so for each SPI slave device, there
> is only one
> master device.

Do you share CLK, too? Then you need a slave select from B to C.
If you use a separate clock, the slave select from B to C can be optional.

> so I think the question becomes whether the above mentioned hardware setup
> is valid or not.

It's a non-conventional SPI bus setup, but it can work, provided you have
some form of synchronization between A and B.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

[toc] | [prev] | [next] | [standalone]

#1630386

FromMark Brown <broonie@kernel.org>
Date2017-04-25 12:40 +0200
Message-ID<tA630-3ve-11@gated-at.bofh.it>
In reply to#1629427

[Multipart message — attachments visible in raw view] — view raw

On Mon, Apr 24, 2017 at 12:55:21PM +0200, Geert Uytterhoeven wrote:
> On Fri, Apr 14, 2017 at 7:39 AM, Jiada Wang <jiada_wang@mentor.com> wrote:

> > Our use case is to use spidev as an interface to communicate with external
> > SPI master devices.
> > meanwhile the SPI bus controller can also act as master device to send data
> > to other
> > SPI slave devices on the board.

> That sounds a bit hackish to me. SPI was never meant to be a multi-master bus.
> While it can be done, you will need external synchronization (signals) to
> avoid conflicts between the SPI masters.

> > I found in your implementation, SPI bus controller is limited to either work
> > in master mode or
> > slave mode, is there any reasoning to not configure SPI mode based on SPI
> > devices use case?

> If you really need both master and slave support, you can use 2 subnodes
> in DT, the first representing the master, the second the slave.

> Mark, what's your opinion about this?

That sounds like a mess...   we *could* put the slave flag on the device
rather than the controller I guess but there's also going to need to be
something representing whatever avoids collisions on the bus somewhere.

[toc] | [prev] | [next] | [standalone]

#1631901

FromJiada Wang <jiada_wang@mentor.com>
Date2017-04-27 08:50 +0200
Message-ID<tALpv-5dO-17@gated-at.bofh.it>
In reply to#1630386
Hello Geert and Mark

On 04/25/2017 03:31 AM, Mark Brown wrote:
> On Mon, Apr 24, 2017 at 12:55:21PM +0200, Geert Uytterhoeven wrote:
>> On Fri, Apr 14, 2017 at 7:39 AM, Jiada Wang<jiada_wang@mentor.com>  wrote:
>>> Our use case is to use spidev as an interface to communicate with external
>>> SPI master devices.
>>> meanwhile the SPI bus controller can also act as master device to send data
>>> to other
>>> SPI slave devices on the board.
>> That sounds a bit hackish to me. SPI was never meant to be a multi-master bus.
>> While it can be done, you will need external synchronization (signals) to
>> avoid conflicts between the SPI masters.
>>> I found in your implementation, SPI bus controller is limited to either work
>>> in master mode or
>>> slave mode, is there any reasoning to not configure SPI mode based on SPI
>>> devices use case?
>> If you really need both master and slave support, you can use 2 subnodes
>> in DT, the first representing the master, the second the slave.
>> Mark, what's your opinion about this?
> That sounds like a mess...   we *could* put the slave flag on the device
> rather than the controller I guess but there's also going to need to be
> something representing whatever avoids collisions on the bus somewhere.
The reason I gave the example use case is want to point out that
with Geert's patch set, a SPI device (with only one controller) can no 
longer
act as master and slave at the same time. because IMO as a SPI core 
function,
it needs to cover all the use cases, and to be as generic as possible.

BUT if you think the use case don't need to be supported from SPI core,
then I don't have objection either, I will only submit imx SPI slave 
support patch,
after your SPI slave support patch set been applied


Thanks,
Jiada

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web