Patchwork i2c-cpm: Fix to takeback i2c bus master-ship after a collision

login
register
mail settings
Submitter sachin surendran
Date Oct. 31, 2012, 3 a.m.
Message ID <CABCb4NUJ0uJA5Z9vdgLndvpLrnK=m5xaeiHZxAwHKFU2DroaFw@mail.gmail.com>
Download mbox | patch
Permalink /patch/195767/
State Superseded
Delegated to: Kumar Gala
Headers show

Comments

sachin surendran - Oct. 31, 2012, 3 a.m.
In case of collision on i2c bus the controller which lost bus mastership
stays as a slave
for all subsequent transfers. This results in the i2c controller never
writing to the bus
for future transactions, resulting in i2c transfer timeouts.
  This fix checks for a collision on last I2C transaction and sets the
I2COM_MASTER
bit for the new transaction.

Signed-off-by: Sachin Surendran <sachin.surendran@alliedtelesis.co.nz>
---
 drivers/i2c/busses/i2c-cpm.c |    8 ++++++++
 1 file changed, 8 insertions(+)

                dev_dbg(&adap->dev, "R: %d T: %d\n", rptr, tptr);
Jochen Friedrich - Oct. 31, 2012, 9:52 a.m.
Hi Sachin,

looks good to me.

Acked-By: Jochen Friedrich <jochen@scram.de>

> Signed-off-by: Sachin Surendran <sachin.surendran@alliedtelesis.co.nz
> <mailto:sachin.surendran@alliedtelesis.co.nz>>
> ---
>   drivers/i2c/busses/i2c-cpm.c |    8 ++++++++
>   1 file changed, 8 insertions(+)
>
> diff --git a/drivers/i2c/busses/i2c-cpm.c b/drivers/i2c/busses/i2c-cpm.c
> index c1e1096..bb94c6d 100644
> --- a/drivers/i2c/busses/i2c-cpm.c
> +++ b/drivers/i2c/busses/i2c-cpm.c
> @@ -338,6 +338,14 @@ static int cpm_i2c_xfer(struct i2c_adapter *adap,
> struct i2c_msg *msgs, int num)
>          tptr = 0;
>          rptr = 0;
> +       /*
> +        * If there was a collision in the last i2c transaction,
> +        * Set I2COM_MASTER as it was cleared during collision.
> +        */
> +       if (in_be16(&tbdf->cbd_sc) & BD_SC_CL) {
> +               out_8(&cpm->i2c_reg->i2com, I2COM_MASTER);
> +       }
> +
>          while (tptr < num) {
>                  pmsg = &msgs[tptr];
>                  dev_dbg(&adap->dev, "R: %d T: %d\n", rptr, tptr);
>

Thanks, Jochen
Kumar Gala - Nov. 25, 2012, 1:21 p.m.
On Oct 30, 2012, at 10:00 PM, sachin surendran wrote:

> In case of collision on i2c bus the controller which lost bus mastership stays as a slave
> for all subsequent transfers. This results in the i2c controller never writing to the bus 
> for future transactions, resulting in i2c transfer timeouts.
>   This fix checks for a collision on last I2C transaction and sets the I2COM_MASTER 
> bit for the new transaction.
> 
> Signed-off-by: Sachin Surendran <sachin.surendran@alliedtelesis.co.nz>
> ---
>  drivers/i2c/busses/i2c-cpm.c |    8 ++++++++
>  1 file changed, 8 insertions(+)

patch doesn't apply, please resend properly.

- k

> 
> diff --git a/drivers/i2c/busses/i2c-cpm.c b/drivers/i2c/busses/i2c-cpm.c
> index c1e1096..bb94c6d 100644
> --- a/drivers/i2c/busses/i2c-cpm.c
> +++ b/drivers/i2c/busses/i2c-cpm.c
> @@ -338,6 +338,14 @@ static int cpm_i2c_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs, int num)
>         tptr = 0;
>         rptr = 0;
>  
> +       /*
> +        * If there was a collision in the last i2c transaction,
> +        * Set I2COM_MASTER as it was cleared during collision.
> +        */
> +       if (in_be16(&tbdf->cbd_sc) & BD_SC_CL) {
> +               out_8(&cpm->i2c_reg->i2com, I2COM_MASTER);
> +       }
> +
>         while (tptr < num) {
>                 pmsg = &msgs[tptr];
>                 dev_dbg(&adap->dev, "R: %d T: %d\n", rptr, tptr);
> 
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev
sachin surendran - Nov. 25, 2012, 10:57 p.m.
Hi Kumar,
             I have resent the updated patch.

Regards
Sachin

On Mon, Nov 26, 2012 at 2:21 AM, Kumar Gala <galak@kernel.crashing.org> wrote:
>
> On Oct 30, 2012, at 10:00 PM, sachin surendran wrote:
>
>> In case of collision on i2c bus the controller which lost bus mastership stays as a slave
>> for all subsequent transfers. This results in the i2c controller never writing to the bus
>> for future transactions, resulting in i2c transfer timeouts.
>>   This fix checks for a collision on last I2C transaction and sets the I2COM_MASTER
>> bit for the new transaction.
>>
>> Signed-off-by: Sachin Surendran <sachin.surendran@alliedtelesis.co.nz>
>> ---
>>  drivers/i2c/busses/i2c-cpm.c |    8 ++++++++
>>  1 file changed, 8 insertions(+)
>
> patch doesn't apply, please resend properly.
>
> - k
>
>>
>> diff --git a/drivers/i2c/busses/i2c-cpm.c b/drivers/i2c/busses/i2c-cpm.c
>> index c1e1096..bb94c6d 100644
>> --- a/drivers/i2c/busses/i2c-cpm.c
>> +++ b/drivers/i2c/busses/i2c-cpm.c
>> @@ -338,6 +338,14 @@ static int cpm_i2c_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs, int num)
>>         tptr = 0;
>>         rptr = 0;
>>
>> +       /*
>> +        * If there was a collision in the last i2c transaction,
>> +        * Set I2COM_MASTER as it was cleared during collision.
>> +        */
>> +       if (in_be16(&tbdf->cbd_sc) & BD_SC_CL) {
>> +               out_8(&cpm->i2c_reg->i2com, I2COM_MASTER);
>> +       }
>> +
>>         while (tptr < num) {
>>                 pmsg = &msgs[tptr];
>>                 dev_dbg(&adap->dev, "R: %d T: %d\n", rptr, tptr);
>>
>> _______________________________________________
>> Linuxppc-dev mailing list
>> Linuxppc-dev@lists.ozlabs.org
>> https://lists.ozlabs.org/listinfo/linuxppc-dev
>

Patch

diff --git a/drivers/i2c/busses/i2c-cpm.c b/drivers/i2c/busses/i2c-cpm.c
index c1e1096..bb94c6d 100644
--- a/drivers/i2c/busses/i2c-cpm.c
+++ b/drivers/i2c/busses/i2c-cpm.c
@@ -338,6 +338,14 @@  static int cpm_i2c_xfer(struct i2c_adapter *adap,
struct i2c_msg *msgs, int num)
        tptr = 0;
        rptr = 0;

+       /*
+        * If there was a collision in the last i2c transaction,
+        * Set I2COM_MASTER as it was cleared during collision.
+        */
+       if (in_be16(&tbdf->cbd_sc) & BD_SC_CL) {
+               out_8(&cpm->i2c_reg->i2com, I2COM_MASTER);
+       }
+
        while (tptr < num) {
                pmsg = &msgs[tptr];