Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1383763 > unrolled thread

Re: [PATCH v6 08/19] i2c: octeon: Enable High-Level Controller

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v6 08/19] i2c: octeon: Enable High-Level Controller Wolfram Sang <wsa@the-dreams.de> - 2016-04-20 23:50 +0200
    Re: [PATCH v6 08/19] i2c: octeon: Enable High-Level Controller David Daney <ddaney@caviumnetworks.com> - 2016-04-21 00:00 +0200
      Re: [PATCH v6 08/19] i2c: octeon: Enable High-Level Controller Wolfram Sang <wsa@the-dreams.de> - 2016-04-21 16:00 +0200
    Re: [PATCH v6 08/19] i2c: octeon: Enable High-Level Controller Jan Glauber <jan.glauber@caviumnetworks.com> - 2016-04-21 16:20 +0200

#1383763 — Re: [PATCH v6 08/19] i2c: octeon: Enable High-Level Controller

[Multipart message — attachments visible in raw view] — view raw

On Mon, Apr 11, 2016 at 05:28:39PM +0200, Jan Glauber wrote:
> From: David Daney <ddaney@caviumnetworks.com>
> 
> Use High-Level Controller (HLC) when possible. The HLC can read/write
> up to 8 bytes and is completely optional. The most important difference
> of the HLC is that it only requires one interrupt for a transfer
> (up to 8 bytes) where the low-level read/write requires 2 interrupts
> plus one interrupt per transferred byte. Since the interrupts are costly
> using the HLC improves the performance. Also, the HLC provides improved error
> handling.

Much better description, thanks!

> +	while (1) {
> +		val = octeon_i2c_ctl_read(i2c);
> +		if (!(val & (TWSI_CTL_STA | TWSI_CTL_STP)));
> +			break;
> +
> +		/* clear IFLG event */
> +		if (val & TWSI_CTL_IFLG)
> +			octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
> +
> +		if (try++ > 100) {
> +			pr_err("%s: giving up\n", __func__);
> +			break;
> +		}
> +
> +		/* spin until any start/stop has finished */
> +		udelay(10);
> +	}

Maybe you can use one of the readx_poll_timeout() functions?

> +/**
> + * octeon_i2c_hlc_wait - wait for an HLC operation to complete
> + * @i2c: The struct octeon_i2c
> + *
> + * Returns 0 on success, otherwise a negative errno.
> + */
> +static int octeon_i2c_hlc_wait(struct octeon_i2c *i2c)
> +{
> +	int time_left;
> +
> +	octeon_i2c_hlc_int_enable(i2c);
> +	time_left = wait_event_interruptible_timeout(i2c->queue,
> +					octeon_i2c_hlc_test_ready(i2c),
> +					i2c->adap.timeout);

Have you tested signal handling thoroughly? Most driver dropped the
_interruptible after a while. Mostly they found out that the state
machine of the interrupt handler couldn't gracefully deal with it and
nobody really needed the interruptible. Just saying.

> +	octeon_i2c_int_disable(i2c);
> +	if (!time_left) {
> +		octeon_i2c_hlc_int_clear(i2c);
> +		dev_dbg(i2c->dev, "%s: timeout\n", __func__);
> +		return -ETIMEDOUT;
> +	}
> +
> +	if (time_left < 0) {
> +		dev_dbg(i2c->dev, "%s: wait interrupted\n", __func__);
> +		return time_left;
> +	}
> +	return 0;
> +}

Drop the debug messages?

I can't say much about the HW details, of course. Didn't spot anything
suspicious there.

[toc] | [next] | [standalone]

#1383766

On 04/20/2016 02:43 PM, Wolfram Sang wrote:
> On Mon, Apr 11, 2016 at 05:28:39PM +0200, Jan Glauber wrote:
[...]
>> + */
>> +static int octeon_i2c_hlc_wait(struct octeon_i2c *i2c)
>> +{
>> +	int time_left;
>> +
>> +	octeon_i2c_hlc_int_enable(i2c);
>> +	time_left = wait_event_interruptible_timeout(i2c->queue,
>> +					octeon_i2c_hlc_test_ready(i2c),
>> +					i2c->adap.timeout);
>
> Have you tested signal handling thoroughly? Most driver dropped the
> _interruptible after a while. Mostly they found out that the state
> machine of the interrupt handler couldn't gracefully deal with it and
> nobody really needed the interruptible. Just saying.

Good point.  We know that exiting with a signal leaves us in an 
undefined state.

We will have to think on this point.


>
>> +	octeon_i2c_int_disable(i2c);
>> +	if (!time_left) {
>> +		octeon_i2c_hlc_int_clear(i2c);
>> +		dev_dbg(i2c->dev, "%s: timeout\n", __func__);
>> +		return -ETIMEDOUT;
>> +	}
>> +
>> +	if (time_left < 0) {
>> +		dev_dbg(i2c->dev, "%s: wait interrupted\n", __func__);
>> +		return time_left;
>> +	}
>> +	return 0;
>> +}
>
> Drop the debug messages?
>
> I can't say much about the HW details, of course. Didn't spot anything
> suspicious there.
>

[toc] | [prev] | [next] | [standalone]

#1384260

[Multipart message — attachments visible in raw view] — view raw

> I think we should just drop the _interruptible_ and use
> wait_event_timeout. The same is already used in the octeon_i2c_wait().
> The 2ms timeout should not hurt anyone.

This is what most people have good experience with :)

[toc] | [prev] | [next] | [standalone]

#1384271

On Wed, Apr 20, 2016 at 11:43:54PM +0200, Wolfram Sang wrote:
> On Mon, Apr 11, 2016 at 05:28:39PM +0200, Jan Glauber wrote:
> > From: David Daney <ddaney@caviumnetworks.com>
> > 
> > Use High-Level Controller (HLC) when possible. The HLC can read/write
> > up to 8 bytes and is completely optional. The most important difference
> > of the HLC is that it only requires one interrupt for a transfer
> > (up to 8 bytes) where the low-level read/write requires 2 interrupts
> > plus one interrupt per transferred byte. Since the interrupts are costly
> > using the HLC improves the performance. Also, the HLC provides improved error
> > handling.
> 
> Much better description, thanks!
> 
> > +	while (1) {
> > +		val = octeon_i2c_ctl_read(i2c);
> > +		if (!(val & (TWSI_CTL_STA | TWSI_CTL_STP)));
> > +			break;
> > +
> > +		/* clear IFLG event */
> > +		if (val & TWSI_CTL_IFLG)
> > +			octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
> > +
> > +		if (try++ > 100) {
> > +			pr_err("%s: giving up\n", __func__);
> > +			break;
> > +		}
> > +
> > +		/* spin until any start/stop has finished */
> > +		udelay(10);
> > +	}
> 
> Maybe you can use one of the readx_poll_timeout() functions?

Nice, but I don't think we can use readx_poll_timeout() here because
of the octeon_i2c_ctl_write inside the loop...

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web