Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Mon, 14 Sep 2020 12:36:03 -0700
From:   Dmitry Torokhov <dmitry.torokhov@gmail.com>
To:     Dmitry Osipenko <digetx@gmail.com>
Cc:     Jiada Wang <jiada_wang@mentor.com>, nick@shmanahar.org,
        linux-input@vger.kernel.org, linux-kernel@vger.kernel.org,
        andy.shevchenko@gmail.com, erosca@de.adit-jv.com,
        Andrew_Gabbasov@mentor.com
Subject: Re: [PATCH v4 1/1] Input: atmel_mxt_ts - implement I2C retries
Message-ID: <20200914193603.GW1665100@dtor-ws>
References: <20200912005521.26319-1-jiada_wang@mentor.com>
 <20200913165644.GF1665100@dtor-ws>
 <65d1b9f2-a8e3-6398-712d-41d8067d06a1@gmail.com>
 <20200914193340.GV1665100@dtor-ws>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20200914193340.GV1665100@dtor-ws>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Sep 14, 2020 at 12:33:40PM -0700, Dmitry Torokhov wrote:
> On Mon, Sep 14, 2020 at 08:29:44PM +0300, Dmitry Osipenko wrote:
> > 13.09.2020 19:56, Dmitry Torokhov пишет:
> > > Hi Jiada,
> > > 
> > > On Sat, Sep 12, 2020 at 09:55:21AM +0900, Jiada Wang wrote:
> > >> From: Nick Dyer <nick.dyer@itdev.co.uk>
> > >>
> > >> Some maXTouch chips (eg mXT1386) will not respond on the first I2C request
> > >> when they are in a sleep state. It must be retried after a delay for the
> > >> chip to wake up.
> > > 
> > > Do we know when the chip is in sleep state? Can we do a quick I2C
> > > transaction in this case instead of adding retry logic to everything? Or
> > > there is another benefit for having such retry logic?
> > 
> > Hello!
> > 
> > Please take a look at page 29 of:
> > 
> > https://ww1.microchip.com/downloads/en/DeviceDoc/mXT1386_1vx_Datasheet_LX.pdf
> > 
> > It says that the retry is needed after waking up from a deep-sleep mode.
> > 
> > There are at least two examples when it's needed:
> > 
> > 1. Driver probe. Controller could be in a deep-sleep mode at the probe
> > time, and then first __mxt_read_reg() returns I2C NACK on reading out TS
> > hardware info.
> > 
> > 2. Touchscreen input device is opened. The touchscreen is in a
> > deep-sleep mode at the time when input device is opened, hence first
> > __mxt_write_reg() invoked from mxt_start() returns I2C NACK.
> > 
> > I think placing the retries within __mxt_read() / write_reg() should be
> > the most universal option.
> > 
> > Perhaps it should be possible to add mxt_wake() that will read out some
> > generic register
> 
> I do not think we need to read a particular register, just doing a quick
> read:
> 
> 	i2c_smbus_xfer(client->adapter, client->addr,
> 			0, I2C_SMBUS_READ, 0, I2C_SMBUS_BYTE, &dummy)
> 
> should suffice.
> 
> > and then this helper should be invoked after HW
> > resetting (before mxt_read_info_block()) and from mxt_start() (before
> > mxt_set_t7_power_cfg()). But this approach feels a bit fragile to me.
> >
> 
> Actually, reading the spec, it all depends on how the WAKE pin is wired
> up on a given board. In certain setups retrying transaction is the right
> approach, while in others explicit control is needed. So indeed, we need
> a "wake" helper that we should call in probe and resume paths.

By the way, I would like to avoid the unnecessary retries in probe paths
if possible. I.e. on Chrome OS we really keep an eye on boot times and
in case of multi-sourced touchscreens we may legitimately not have
device at given address.

Thanks.

-- 
Dmitry