Re: i2c issue on Panda with DT boot, v3.10-rc4

[Tomi Valkeinen <tomi.valkeinen@iki.fi>]

[-- Attachment #1: Type: text/plain, Size: 1373 bytes --]

On 07/06/13 21:39, Felipe Balbi wrote:

> sounds like there's something left in FIFO which is not getting read
> out, then we end up timing out.
> 
> Can you try the patch below ? It's patch of a bigger patchset which I
> still need to clean a few things up, but they should be very close to
> being ready. IIRC, one of the patches creates a problem for N900 (only)
> which gets fixed later, I just need to combine those two patches into
> one to avoid the regression.
> 
> diff --git a/drivers/i2c/busses/i2c-omap.c b/drivers/i2c/busses/i2c-omap.c
> index aa3b91e..471b434 100644
> --- a/drivers/i2c/busses/i2c-omap.c
> +++ b/drivers/i2c/busses/i2c-omap.c
> @@ -1022,9 +1022,8 @@ omap_i2c_isr_thread(int this_irq, void *dev_id)
>  		}
>  	} while (stat);
>  
> -	omap_i2c_complete_cmd(dev, err);
> -
>  out:
> +	omap_i2c_complete_cmd(dev, err);
>  	spin_unlock_irqrestore(&dev->lock, flags);
>  
>  	return IRQ_HANDLED;
> 

With this change the boot becomes unreliable:

[    3.024322] V2V1: 2100 mV
[    4.049530] omap_i2c 48070000.i2c: timeout waiting for bus ready
[    5.059417] omap_i2c 48070000.i2c: timeout waiting for bus ready
[    5.059448] twl: Write failed (mod 9, reg 0xe5 count 1)
and this continues.

I did manage to boot once, and running i2cdump printed each byte very
slowly, and with 0xff as the data.

 Tomi



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 901 bytes --]