[PATCH v2] i2c: at91: Fix driver for SAMA5D2 SoC

[PATCH v2] i2c: at91: Fix driver for SAMA5D2 SoC

[Alexander Shiyan]

Most bits of the AT91 SR status register are cleared after reading,
so we may lose the status at the end of the wait loop.
Let's use the value already readed to check for "completed" status.

Signed-off-by: Alexander Shiyan <eagle.alexander923@gmail.com>
---
 drivers/i2c/busses/i2c-at91.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/i2c/busses/i2c-at91.c b/drivers/i2c/busses/i2c-at91.c
index dbc5aa4ddf..8929dbaede 100644
--- a/drivers/i2c/busses/i2c-at91.c
+++ b/drivers/i2c/busses/i2c-at91.c
@@ -276,7 +276,7 @@ static int at91_twi_wait_completion(struct at91_twi_dev *dev)
 
 		dev->transfer_status |= status;
 
-	} while (!(at91_twi_read(dev, AT91_TWI_SR) & AT91_TWI_TXCOMP));
+	} while (!(status & AT91_TWI_TXCOMP));
 
 	at91_disable_twi_interrupts(dev);
 
-- 
2.39.1

Re: [PATCH v2] i2c: at91: Fix driver for SAMA5D2 SoC

[Ahmad Fatoum]

On 13.01.25 08:02, Alexander Shiyan wrote:
> Most bits of the AT91 SR status register are cleared after reading,
> so we may lose the status at the end of the wait loop.
> Let's use the value already readed to check for "completed" status.
> 
> Signed-off-by: Alexander Shiyan <eagle.alexander923@gmail.com>
> ---
>  drivers/i2c/busses/i2c-at91.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/i2c/busses/i2c-at91.c b/drivers/i2c/busses/i2c-at91.c
> index dbc5aa4ddf..8929dbaede 100644
> --- a/drivers/i2c/busses/i2c-at91.c
> +++ b/drivers/i2c/busses/i2c-at91.c
> @@ -276,7 +276,7 @@ static int at91_twi_wait_completion(struct at91_twi_dev *dev)
>  
>  		dev->transfer_status |= status;
>  
> -	} while (!(at91_twi_read(dev, AT91_TWI_SR) & AT91_TWI_TXCOMP));
> +	} while (!(status & AT91_TWI_TXCOMP));

This makes the function harder to reason about, because the status read
is done at the start, but only acted upon at the very end.

Would it make sense to change this to a while loop or move it into the
loop directly after the other status read and early exit if it's true?

>  
>  	at91_disable_twi_interrupts(dev);
>  


-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Re: [PATCH v2] i2c: at91: Fix driver for SAMA5D2 SoC

[Alexander Shiyan]

Hello.

As far as I understand, AT91_TWI_TXCOMP is set at the same time as
one of the AT91_TWI_RXRDY or AT91_TWI_TXRDY signals, so we need
to check these bits first to perform the read or write operation of
the next byte.
If it was the last byte (when the STOP bit is transmitted) we exit.

Thanks!

пн, 13 янв. 2025 г. в 16:17, Ahmad Fatoum <a.fatoum@pengutronix.de>:
>
> On 13.01.25 08:02, Alexander Shiyan wrote:
> > Most bits of the AT91 SR status register are cleared after reading,
> > so we may lose the status at the end of the wait loop.
> > Let's use the value already readed to check for "completed" status.
> >
> > Signed-off-by: Alexander Shiyan <eagle.alexander923@gmail.com>
> > ---
> >  drivers/i2c/busses/i2c-at91.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/i2c/busses/i2c-at91.c b/drivers/i2c/busses/i2c-at91.c
> > index dbc5aa4ddf..8929dbaede 100644
> > --- a/drivers/i2c/busses/i2c-at91.c
> > +++ b/drivers/i2c/busses/i2c-at91.c
> > @@ -276,7 +276,7 @@ static int at91_twi_wait_completion(struct at91_twi_dev *dev)
> >
> >               dev->transfer_status |= status;
> >
> > -     } while (!(at91_twi_read(dev, AT91_TWI_SR) & AT91_TWI_TXCOMP));
> > +     } while (!(status & AT91_TWI_TXCOMP));
>
> This makes the function harder to reason about, because the status read
> is done at the start, but only acted upon at the very end.
>
> Would it make sense to change this to a while loop or move it into the
> loop directly after the other status read and early exit if it's true?
>
> >
> >       at91_disable_twi_interrupts(dev);
> >
>
>
> --
> Pengutronix e.K.                           |                             |
> Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
> 31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
> Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Re: [PATCH v2] i2c: at91: Fix driver for SAMA5D2 SoC

[Ahmad Fatoum]

On 13.01.25 14:37, Alexander Shiyan wrote:
> Hello.
> 
> As far as I understand, AT91_TWI_TXCOMP is set at the same time as
> one of the AT91_TWI_RXRDY or AT91_TWI_TXRDY signals, so we need
> to check these bits first to perform the read or write operation of
> the next byte.
> If it was the last byte (when the STOP bit is transmitted) we exit.

If the case AT91_TWI_TXCOMP on its own can't happen, then this is fine.

Acked-by: Ahmad Fatoum <a.fatoum@pengutronix.de>

Thanks,
Ahmad

> 
> Thanks!
> 
> пн, 13 янв. 2025 г. в 16:17, Ahmad Fatoum <a.fatoum@pengutronix.de>:
>>
>> On 13.01.25 08:02, Alexander Shiyan wrote:
>>> Most bits of the AT91 SR status register are cleared after reading,
>>> so we may lose the status at the end of the wait loop.
>>> Let's use the value already readed to check for "completed" status.
>>>
>>> Signed-off-by: Alexander Shiyan <eagle.alexander923@gmail.com>
>>> ---
>>>  drivers/i2c/busses/i2c-at91.c | 2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/i2c/busses/i2c-at91.c b/drivers/i2c/busses/i2c-at91.c
>>> index dbc5aa4ddf..8929dbaede 100644
>>> --- a/drivers/i2c/busses/i2c-at91.c
>>> +++ b/drivers/i2c/busses/i2c-at91.c
>>> @@ -276,7 +276,7 @@ static int at91_twi_wait_completion(struct at91_twi_dev *dev)
>>>
>>>               dev->transfer_status |= status;
>>>
>>> -     } while (!(at91_twi_read(dev, AT91_TWI_SR) & AT91_TWI_TXCOMP));
>>> +     } while (!(status & AT91_TWI_TXCOMP));
>>
>> This makes the function harder to reason about, because the status read
>> is done at the start, but only acted upon at the very end.
>>
>> Would it make sense to change this to a while loop or move it into the
>> loop directly after the other status read and early exit if it's true?
>>
>>>
>>>       at91_disable_twi_interrupts(dev);
>>>
>>
>>
>> --
>> Pengutronix e.K.                           |                             |
>> Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
>> 31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
>> Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
> 


-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Re: [PATCH v2] i2c: at91: Fix driver for SAMA5D2 SoC

[Sascha Hauer]

On Mon, 13 Jan 2025 10:02:16 +0300, Alexander Shiyan wrote:
> Most bits of the AT91 SR status register are cleared after reading,
> so we may lose the status at the end of the wait loop.
> Let's use the value already readed to check for "completed" status.
> 
> 

Applied, thanks!

[1/1] i2c: at91: Fix driver for SAMA5D2 SoC
      https://git.pengutronix.de/cgit/barebox/commit/?id=da9d4089f0d8 (link may not be stable)

Best regards,
-- 
Sascha Hauer <s.hauer@pengutronix.de>