[PATCH] i2c: cadence: Increase timeout per message if necessary

[tanureal at opensource.cirrus.com]

On 3/22/22 5:18 PM, Michal Simek <michal.simek at xilinx.com> wrote:
> 
> 
> On 3/22/22 16:34, tanureal at opensource.cirrus.com wrote:
> > On 3/21/22 3:57 PM, Michal Simek <michal.simek at xilinx.com> wrote:
> >> +Shubhrajyoti
> >>
> >> On 3/9/22 10:31, Lucas Tanure wrote:
> >> > Timeout as 1 second sets a upper limit on the length of
> >> > the transfer executed, but there is no maximum length of
> >> > a write or read message set in i2c_adapter_quirks for this
> >> > controller.
> >>
> >> nit: I would expect that you have run any test and you reached an issue.
> >> Would be good to describe what exactly you have tried on which 
> >> configuration to make it super clear.
> >>
> >> >
> >> > To remove that limitation calculate the minimal time
> >> > necessary, plus some wiggle room, for every message
> >> > and use it instead of the default one second, if
> >> > more than one second.
> >> >
> >> > Signed-off-by: Lucas Tanure <tanureal at opensource.cirrus.com>
> >> > ---
> >> >   drivers/i2c/busses/i2c-cadence.c | 12 ++++++++++--
> >> >   1 file changed, 10 insertions(+), 2 deletions(-)
> >> >
> >> > diff --git a/drivers/i2c/busses/i2c-cadence.c > 
> >> b/drivers/i2c/busses/i2c-cadence.c
> >> > index 805c77143a0f..b4c1ad19cdae 100644
> >> > --- a/drivers/i2c/busses/i2c-cadence.c
> >> > +++ b/drivers/i2c/busses/i2c-cadence.c
> >> > @@ -760,7 +760,7 @@ static void cdns_i2c_master_reset(struct > 
> >> i2c_adapter *adap)
> >> >   static int cdns_i2c_process_msg(struct cdns_i2c *id, struct 
> >> i2c_msg > *msg,
> >> >           struct i2c_adapter *adap)
> >> >   {
> >> > -    unsigned long time_left;
> >> > +    unsigned long time_left, msg_timeout;
> >> >       u32 reg;
> >> >       id->p_msg = msg;
> >> > @@ -785,8 +785,16 @@ static int cdns_i2c_process_msg(struct 
> >> cdns_i2c > *id, struct i2c_msg *msg,
> >> >       else
> >> >           cdns_i2c_msend(id);
> >> > +    /* Minimal time to execute this message */
> >> > +    msg_timeout = msecs_to_jiffies((1000 * msg->len * 
> >> BITS_PER_BYTE) > / id->i2c_clk);
> >> > +    /* Plus some wiggle room */
> >> > +    msg_timeout += msecs_to_jiffies(500);
> >> > +
> >> > +    if (msg_timeout < adap->timeout)
> >> > +        msg_timeout = adap->timeout;
> >> > +
> >> >       /* Wait for the signal of completion */
> >> > -    time_left = wait_for_completion_timeout(&id->xfer_done, > 
> >> adap->timeout);
> >> > +    time_left = wait_for_completion_timeout(&id->xfer_done, > 
> >> msg_timeout);
> >> >       if (time_left == 0) {
> >> >           cdns_i2c_master_reset(adap);
> >> >           dev_err(id->adap.dev.parent,
> >>
> >>
> >> If my assumption is right and there is any actual issue you had 
> >> please send v2 and feel free to add there my:
> >> Acked-by: Michal Simek <michal.simek at xilinx.com>
> >>
> >> Thanks,
> >> Michal
> >>
> >>
> >>
> > The issue happens for I2C devices that have firmware, which will send 
> > a big I2C message, but the I2C controller will timeout on it.
> > That happened for CS35L41 DSP firmware tests, so no particular 
> > configuration, just a driver sending firmware blob over I2C.
> 
> How big is it?
> 
> M
> 
The firmware has 33868 bytes, and it is split in a few writes. The first one to time out has 20240 bytes:

[   53.398444] cs35l41 0-0040: DSP1: Firmware version: 3
[   53.403522] cs35l41 0-0040: DSP1: cs35l41-dsp1-spk-prot.wmfw: Fri 04 Feb 2022 12:01:42 W. Europe Standard Time
[   55.331688] cdns-i2c e0004000.i2c: timeout waiting on completion
[   55.336721] cs35l41 0-0040: DSP1: cs35l41-dsp1-spk-prot.wmfw.5: Failed to write 20240 bytes at 0 in PM_PACKED: -110

20240 bytes at 100k clock should take 1.6192 seconds, which is more than the current timeout of one second.