[lm-sensors] i2cdetect running slow but not "bus inactive" slow, not sure why?
guenter.roeck at ericsson.com
Tue Dec 20 17:18:53 CET 2011
On Tue, Dec 20, 2011 at 11:00:35AM -0500, Wayne Tams wrote:
> Hi Guenter,
> Turns out it was an I2C voltage translator had popped, thankfully all
> my I2C devices are ok. Did not realise i2cdetect could behave in such
> a peculiar way.
please don't top-post.
Nothing to do with i2cdetect, but with the i2c bus driver. i2cdetect tries to read,
and the bus driver ends up getting a transfer timeout for each read. Depending on
the I2C bus chip and the driver, timeouts can be _very_ long, up to 500ms per read
or even more.
> On Fri, Dec 16, 2011 at 4:29 PM, Guenter Roeck
> <guenter.roeck at ericsson.com> wrote:
> > On Fri, 2011-12-16 at 10:33 -0500, Wayne Tams wrote:
> >> Hi,
> >> i2cdetect is behaving in a fashion that I have not seen before and I
> >> just would like to know if anyone has a suggestion on what it is
> >> telling me.
> >> Normally i2cdetect returns in <1s, if the bus were inactive it would
> >> take closer to 20s. However I'm finding that the scanning is taken a
> >> round 5s and the addresses returned appear random, except for one.
> >> Obviously there must be a fault in my hardware, but I just wanted to
> >> make sure that was/wasn't a common explanation for this behaviour
> >> before I go debugging up the wrong tree.
> > I typically see this kind of problem if there is something wrong on the
> > i2c bus, for example if a signal is pulled high or low. I have also seen
> > it after I managed to physically "destroy" an I2C controller by applying
> > a reversed voltage to the i2c signal pins.
> > If you have a logic analyzer or a scope, you might want to connect it to
> > the clock and data lines to see what is happening on the bus.
> > Guenter
More information about the lm-sensors