public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Frank Li <Frank.li@nxp.com>
To: Mukesh Kumar Savaliya <quic_msavaliy@quicinc.com>
Cc: Joshua Yeong <joshua.yeong@starfivetech.com>,
	Alexandre Belloni <alexandre.belloni@bootlin.com>,
	Miquel Raynal <miquel.raynal@bootlin.com>,
	Conor Culhane <conor.culhane@silvaco.com>,
	"linux-i3c@lists.infradead.org" <linux-i3c@lists.infradead.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"imx@lists.linux.dev" <imx@lists.linux.dev>
Subject: Re: [PATCH 1/2] i3c: Add basic HDR support
Date: Wed, 5 Feb 2025 10:44:10 -0500	[thread overview]
Message-ID: <Z6OHSsISZPClbRxw@lizhi-Precision-Tower-5810> (raw)
In-Reply-To: <0435e96d-6786-4adc-83eb-a6aa3766c2d5@quicinc.com>

On Wed, Feb 05, 2025 at 07:49:20PM +0530, Mukesh Kumar Savaliya wrote:
> Thanks Frank ! please review below.
>
> On 2/4/2025 9:13 PM, Frank Li wrote:
> > On Tue, Feb 04, 2025 at 11:24:03AM +0530, Mukesh Kumar Savaliya wrote:
> > >
> > >
> > > On 2/3/2025 9:22 PM, Frank Li wrote:
> > > > On Mon, Feb 03, 2025 at 02:32:51AM +0000, Joshua Yeong wrote:
> > > > > Hi Mukesh and Frank,
> > > > >
> > > > > On 02-Feb-2025 1:54 AM, Mukesh Kumar Savaliya wrote:
> > > > > > Hi Frank,
> > > > > >
> > > > > > On 1/30/2025 1:35 AM, Frank Li wrote:
> > > > > > > I3C HDR requires enter/exit patterns during each I3C transfer. Add a new
> > > > > > > API i3c_device_do_priv_xfers_mode(). The existing
> > > > > > > i3c_device_do_priv_xfers() now calls i3c_device_do_priv_xfers_mode(I3C_SDR)
> > > > > > > to maintain backward compatibility.
> > > > > > >
> > > > > > > Introduce a 'cmd' field in 'struct i3c_priv_xfer', using an anonymous union
> > > > > > > with 'rnw' since HDR mode relies on read/write commands instead of the
> > > > > > > 'rnw' bit in the address as in SDR mode.
> > > > > > >
> > > > > > > Add a priv_xfers_mode callback for I3C master drivers. If priv_xfers_mode
> > > > > > > is not implemented, fallback to SDR mode using the existing priv_xfers
> > > > > > > callback.
> > > > > > >
> > > > > > > Signed-off-by: Frank Li <Frank.Li@nxp.com>
> > > > > > > ---
> > > > > > > Why not add hdr mode in struct i3c_priv_xfer because mode can't be mixed in
> > > > > > > one i3c transfer. for example, can't send a HDR follow one SDR between
> > > > > > > START and STOP.
> > > > > > >
> > > > > > Is it wise to add two function inside main funcion and separate SDR vs HDR ? so we don't need to change current function arguments.
> > > >
> > > > I am not sure if understand your means. HDR have difference mode, anyway
> > > > need add new argument.
> > > >
> > > let me clarify, We can have main entry/hookup function as it is.
> > >  From priv_xfer , we can call sdr_priv_xfer and hdr_priv_xfer ? based on the
> > > structure information if its enum SDR or enum HDR.
> > >
> > > > >
> > > > > I think the 'private transfers' are limited to SDR communications,
> > > > > would it be better if we have a different interface/naming for hdr transfers?
> > > >
> > > > The key is what's difference between hdr and private transfers. So far only
> > > > command vs rnw, any other differences I missed?
> > > >
> > > yes, but can we make it priv_xfer() = sdr_priv_xfer() + hdr_priv_xfer() ?
> >
> > struct i3c_priv_xfer is array, each i3c_priv_xfer item is indepent. Can you
> > write a simple code to demostrate your idea.
> >
> > Frank
>
> ======
> Please see below. My only intention is to add hdr/sdr decision inside
> i3c_priv_xfer and rest of the functions should remain as it is.
>
> If you still need, may be we can add local function but no need to change
> prototype of function being explosed to many vendors.
>
> I hope i am not missing any major thing.
>
> struct i3c_priv_xfer {
> 	u8 rnw;
> 	u16 len;
> 	union {
> 		void *in;
> 		const void *out;
> 	} data;
>  	enum i3c_error_code err;
> // Just Add below two members and rest can be passed till end point/function
> and can process as required.
> 	enum i3c_hdr_mode mode;
> 	union {
> 		u8 rnw;
> 		u8 cmd;
> 	};
>  };
>
>
> device.c :
> i3c_device_do_priv_xfers => i3c_dev_do_priv_xfers_locked
>
> int i3c_device_do_priv_xfers(struct i3c_device *dev, struct i3c_priv_xfer
> *xfers, int nxfers) {
>
> ret = i3c_dev_do_priv_xfers_locked(dev->desc, xfers, xfers);
>
> }
>
>
> master.c :
> int i3c_dev_do_priv_xfers_locked(struct i3c_dev_desc *dev, struct
> i3c_priv_xfer *xfers, int nxfers) {
>
> 	if (master->ops->priv_xfers_mode)
> 		return master->ops->priv_xfers_mode(dev, xfers, nxfers);
>
> 	if (xfers->mode != I3C_SDR)
> 		return -EINVAL;
> }
>

The problem is

struct i3c_priv_xfer xfer[3] = {
	{ .mode = SDR ...},
	{ .mode = HDR ...},
	{ .mode = SDR ...},
}

i3c_device_do_priv_xfers(dev, xfer, 3);

How to handle this case?

I think i3c_device_do_priv_xfers() means one whole i3c transcation:
START, xfer[0], xfer[1], xfer[2], STOP.

Frank

>
> svc-i3c-master.c
> //process everything using xfers.mode  and xfer.cmd.
>
> I hope you can add all rest of the changes inside this function itself.
> static int svc_i3c_master_priv_xfers(struct i3c_dev_desc *dev, struct
> i3c_priv_xfer *xfers, int nxfers)
> ==========
> >
> > > > >
> > > > > > > i3c_priv_xfer should be treat as whole i3c transactions. If user want send
> > > > > > > HDR follow SDR, should be call i3c_device_do_priv_xfers_mode() twice,
> > > > > > > instead put into a big i3c_priv_xfer[n].
> > > > > > > ---
> > > > > > >     drivers/i3c/device.c       | 19 +++++++++++++------
> > > > > > >     drivers/i3c/internals.h    |  2 +-
> > > > > > >     drivers/i3c/master.c       |  8 +++++++-
> > > > > > >     include/linux/i3c/device.h | 12 +++++++++++-
> > > > > > >     include/linux/i3c/master.h |  3 +++
> > > > > > >     5 files changed, 35 insertions(+), 9 deletions(-)
> > > > > > >
> > > > > > > diff --git a/drivers/i3c/device.c b/drivers/i3c/device.c
> > > > > > > index e80e487569146..e3db3a6a9e4f6 100644
> > > > > > > --- a/drivers/i3c/device.c
> > > > > > > +++ b/drivers/i3c/device.c
> > > > > > > @@ -15,12 +15,13 @@
> > > > > > >     #include "internals.h"
> > > > > > >       /**
> > > > > > > - * i3c_device_do_priv_xfers() - do I3C SDR private transfers directed to a
> > > > > > > - *                specific device
> > > > > > > + * i3c_device_do_priv_xfers_mode() - do I3C SDR private transfers directed to a
> > > > > > > + *                     specific device
> > > > > > >      *
> > > > > > >      * @dev: device with which the transfers should be done
> > > > > > >      * @xfers: array of transfers
> > > > > > >      * @nxfers: number of transfers
> > > > > > > + * @mode: transfer mode
> > > > > > >      *
> > > > > > >      * Initiate one or several private SDR transfers with @dev.
> > > > > > >      *
> > > > > > > @@ -32,9 +33,9 @@
> > > > > > >      *            driver needs to resend the 'xfers' some time later.
> > > > > > >      *            See I3C spec ver 1.1.1 09-Jun-2021. Section: 5.1.2.2.3.
> > > > > > >      */
> > > > > > > -int i3c_device_do_priv_xfers(struct i3c_device *dev,
> > > > > > > -                 struct i3c_priv_xfer *xfers,
> > > > > > > -                 int nxfers)
> > > > > > > +int i3c_device_do_priv_xfers_mode(struct i3c_device *dev,
> > > > > > > +                  struct i3c_priv_xfer *xfers,
> > > > > > > +                  int nxfers, enum i3c_hdr_mode mode)
> > > > > > why can't we pass mode as another structure member inside struct i3c_priv_xfer ? Adding function agrument parameter impacts existing drivers.
> > > >
> > > > If that, it will be possible, xfers[0] is sdr, xfers[1] is hdr. Actually,
> > > > I3C can't do that.  we assume finish whole xfers between one whole traction
> > > > (between START and STOP).
> > > >
> > > > Frank
> > > > > > >     {
> > > > > > >         int ret, i;
> > > > > > >     @@ -47,11 +48,17 @@ int i3c_device_do_priv_xfers(struct i3c_device *dev,
> > > > > > >         }
> > > > > > >           i3c_bus_normaluse_lock(dev->bus);
> > > > > > > -    ret = i3c_dev_do_priv_xfers_locked(dev->desc, xfers, nxfers);
> > > > > > > +    ret = i3c_dev_do_priv_xfers_locked(dev->desc, xfers, nxfers, mode);
> > > > > > >         i3c_bus_normaluse_unlock(dev->bus);
> > > > > > >           return ret;
> > > > > > >     }
> > > > > > > +EXPORT_SYMBOL_GPL(i3c_device_do_priv_xfers_mode);
> > > > > > > +
> > > > > > > +int i3c_device_do_priv_xfers(struct i3c_device *dev, struct i3c_priv_xfer *xfers, int nxfers)
> > > > > > > +{
> > > > > > > +    return i3c_device_do_priv_xfers_mode(dev, xfers, nxfers, I3C_SDR);
> > > > > > > +}
> > > > > > >     EXPORT_SYMBOL_GPL(i3c_device_do_priv_xfers);
> > > > > > >       /**
> > > > > > > diff --git a/drivers/i3c/internals.h b/drivers/i3c/internals.h
> > > > > > > index 433f6088b7cec..553edc9846ac0 100644
> > > > > > > --- a/drivers/i3c/internals.h
> > > > > > > +++ b/drivers/i3c/internals.h
> > > > > > > @@ -16,7 +16,7 @@ void i3c_bus_normaluse_unlock(struct i3c_bus *bus);
> > > > > > >     int i3c_dev_setdasa_locked(struct i3c_dev_desc *dev);
> > > > > > >     int i3c_dev_do_priv_xfers_locked(struct i3c_dev_desc *dev,
> > > > > > >                      struct i3c_priv_xfer *xfers,
> > > > > > > -                 int nxfers);
> > > > > > > +                 int nxfers, enum i3c_hdr_mode mode);
> > > > > > >     int i3c_dev_disable_ibi_locked(struct i3c_dev_desc *dev);
> > > > > > >     int i3c_dev_enable_ibi_locked(struct i3c_dev_desc *dev);
> > > > > > >     int i3c_dev_request_ibi_locked(struct i3c_dev_desc *dev,
> > > > > > > diff --git a/drivers/i3c/master.c b/drivers/i3c/master.c
> > > > > > > index d5dc4180afbcf..67aaba0a38db2 100644
> > > > > > > --- a/drivers/i3c/master.c
> > > > > > > +++ b/drivers/i3c/master.c
> > > > > > > @@ -2945,7 +2945,7 @@ int i3c_dev_setdasa_locked(struct i3c_dev_desc *dev)
> > > > > > >       int i3c_dev_do_priv_xfers_locked(struct i3c_dev_desc *dev,
> > > > > > >                      struct i3c_priv_xfer *xfers,
> > > > > > > -                 int nxfers)
> > > > > > > +                 int nxfers, enum i3c_hdr_mode mode)
> > > > > > >     {
> > > > > > >         struct i3c_master_controller *master;
> > > > > > >     @@ -2956,9 +2956,15 @@ int i3c_dev_do_priv_xfers_locked(struct i3c_dev_desc *dev,
> > > > > > >         if (!master || !xfers)
> > > > > > >             return -EINVAL;
> > > > > > >     +    if (master->ops->priv_xfers_mode)
> > > > > > > +        return master->ops->priv_xfers_mode(dev, xfers, nxfers, mode);
> > > > > > > +
> > > > > > >         if (!master->ops->priv_xfers)
> > > > > > >             return -ENOTSUPP;
> > > > > > >     +    if (mode != I3C_SDR)
> > > > > > > +        return -EINVAL;
> > > > > > > +
> > > > > > >         return master->ops->priv_xfers(dev, xfers, nxfers);
> > > > > > >     }
> > > > > > >     diff --git a/include/linux/i3c/device.h b/include/linux/i3c/device.h
> > > > > > > index b674f64d0822e..7ce70d0967e27 100644
> > > > > > > --- a/include/linux/i3c/device.h
> > > > > > > +++ b/include/linux/i3c/device.h
> > > > > > > @@ -40,11 +40,13 @@ enum i3c_error_code {
> > > > > > >       /**
> > > > > > >      * enum i3c_hdr_mode - HDR mode ids
> > > > > > > + * @I3C_SDR: SDR mode (NOT HDR mode)
> > > > > > >      * @I3C_HDR_DDR: DDR mode
> > > > > > >      * @I3C_HDR_TSP: TSP mode
> > > > > > >      * @I3C_HDR_TSL: TSL mode
> > > > > > >      */
> > > > > > >     enum i3c_hdr_mode {
> > > > > > > +    I3C_SDR,
> > > > > > >         I3C_HDR_DDR,
> > > > > > >         I3C_HDR_TSP,
> > > > > > >         I3C_HDR_TSL,
> > > > > > > @@ -53,6 +55,7 @@ enum i3c_hdr_mode {
> > > > > > >     /**
> > > > > > >      * struct i3c_priv_xfer - I3C SDR private transfer
> > > > > > >      * @rnw: encodes the transfer direction. true for a read, false for a write
> > > > > > > + * @cmd: Read/Write command in HDR mode, read: 0x80 - 0xff, write: 0x00 - 0x7f
> > > > > > >      * @len: transfer length in bytes of the transfer
> > > > > > >      * @actual_len: actual length in bytes are transferred by the controller
> > > > > > >      * @data: input/output buffer
> > > > > > > @@ -61,7 +64,10 @@ enum i3c_hdr_mode {
> > > > > > >      * @err: I3C error code
> > > > > > >      */
> > > > > > >     struct i3c_priv_xfer {
> > > > > > > -    u8 rnw;
> > > > > > > +    union {
> > > > > > > +        u8 rnw;
> > > > > > > +        u8 cmd;
> > > > > > > +    };
> > > > > > >         u16 len;
> > > > > > >         u16 actual_len;
> > > > > > >         union {
> > > > > > > @@ -301,6 +307,10 @@ int i3c_device_do_priv_xfers(struct i3c_device *dev,
> > > > > > >                      struct i3c_priv_xfer *xfers,
> > > > > > >                      int nxfers);
> > > > > > >     +int i3c_device_do_priv_xfers_mode(struct i3c_device *dev,
> > > > > > > +                  struct i3c_priv_xfer *xfers,
> > > > > > > +                  int nxfers, enum i3c_hdr_mode mode);
> > > > > > > +
> > > > > > >     int i3c_device_do_setdasa(struct i3c_device *dev);
> > > > > > >       void i3c_device_get_info(const struct i3c_device *dev, struct i3c_device_info *info);
> > > > > > > diff --git a/include/linux/i3c/master.h b/include/linux/i3c/master.h
> > > > > > > index 12d532b012c5a..352bd41139569 100644
> > > > > > > --- a/include/linux/i3c/master.h
> > > > > > > +++ b/include/linux/i3c/master.h
> > > > > > > @@ -472,6 +472,9 @@ struct i3c_master_controller_ops {
> > > > > > >         int (*priv_xfers)(struct i3c_dev_desc *dev,
> > > > > > >                   struct i3c_priv_xfer *xfers,
> > > > > > >                   int nxfers);
> > > > > > > +    int (*priv_xfers_mode)(struct i3c_dev_desc *dev,
> > > > > > > +                   struct i3c_priv_xfer *xfers,
> > > > > > > +                   int nxfers, enum i3c_hdr_mode mode);
> > > > > > >         int (*attach_i2c_dev)(struct i2c_dev_desc *dev);
> > > > > > >         void (*detach_i2c_dev)(struct i2c_dev_desc *dev);
> > > > > > >         int (*i2c_xfers)(struct i2c_dev_desc *dev,
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > >
>

  reply	other threads:[~2025-02-05 15:44 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-01-29 20:05 [PATCH 0/2] i3c: Add basic HDR mode support Frank Li
2025-01-29 20:05 ` [PATCH 1/2] i3c: Add basic HDR support Frank Li
2025-02-01 17:54   ` Mukesh Kumar Savaliya
2025-02-03  2:32     ` Joshua Yeong
2025-02-03  5:22       ` Mukesh Kumar Savaliya
2025-02-03 15:52       ` Frank Li
2025-02-04  3:17         ` Joshua Yeong
2025-02-04 15:38           ` Frank Li
2025-02-05  6:30             ` Joshua Yeong
2025-02-05 15:58               ` Frank Li
2025-02-06  7:05                 ` Joshua Yeong
2025-02-06 15:55                   ` Frank Li
2025-02-09 22:26                     ` Joshua Yeong
2025-02-04  5:54         ` Mukesh Kumar Savaliya
2025-02-04 15:43           ` Frank Li
2025-02-05 14:19             ` Mukesh Kumar Savaliya
2025-02-05 15:44               ` Frank Li [this message]
2025-02-06  9:44                 ` Mukesh Kumar Savaliya
2025-02-06 15:32                   ` Frank Li
2025-02-07  9:08                     ` Mukesh Kumar Savaliya
2025-01-29 20:05 ` [PATCH 2/2] i3c: master: svc: Add basic HDR mode support Frank Li

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=Z6OHSsISZPClbRxw@lizhi-Precision-Tower-5810 \
    --to=frank.li@nxp.com \
    --cc=alexandre.belloni@bootlin.com \
    --cc=conor.culhane@silvaco.com \
    --cc=imx@lists.linux.dev \
    --cc=joshua.yeong@starfivetech.com \
    --cc=linux-i3c@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=miquel.raynal@bootlin.com \
    --cc=quic_msavaliy@quicinc.com \
    /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