Messages in this thread | | | Date | Tue, 02 Dec 2014 16:12:48 +0100 | From | Lars-Peter Clausen <> | Subject | Re: [PATCH 3/4] devicetree: bindings: Add defeature-repeated-start property for Cadence I2C |
| |
On 12/02/2014 03:15 PM, Wolfram Sang wrote: >>> What do you do when disable repeated start? Sending STOP and START? If >>> so, this is really something different than repeated start. By using >>> I2C_FUNC_I2C a user expects repeated start, so if the HW does not >>> support it, we should say so and don't try to emulate it with something >>> different. >> >> Yes, we send stop. > > As said before, this is wrong. Another master could interfere between > the messages when using stop+start. This is no replacement for repeated > start.
More importantly a lot of I2C slaves also reset their internal state machine on a stop. So e.g. if reading a register is implemented by doing start,write,repeated start,read,stop and you replace that with start,write,stop,start,read,stop you'll always read register zero instead of the register you wanted to read.
- Lars
| |