Il giorno lun, 07/07/2008 alle 20.02 +0200, Wolfgang Denk ha scritto:
In message 1215447465.12657.15.camel@localhost you wrote:
I just extended the fsl_i2c.c driver that already uses a precomputed
Yes, I know. I never understood what such a complicated driver for sich a simple protocol was good for, especially as we don't do multimegabytepersecond transfers over such a bus.
I agree with you. But I prefer to have a (too) complex working driver instead an buggy driver. In other hands if m547x cpus have to use the fsl_driver... this should be fixed in order to work fine. Otherwise the HW i2c driver should be dropped from m57x/m5445x boards in order to avoid to confuse the u-boot users/developers.
Today my HW-man ask me why the i2c bus ran at 1Mhz... The cause was that I used the driver with confidence... too confidence...
table to setup the fsl i2c. I haven't invented these values but just copied from the 547x and 5445x manuals. Furthermore, the h/w i2c is more accurate that the simple bitbanging.
Accurate? There is no requirement to be "accurate" anywhere in the I2C protocol. It is a simple, brain-dead protokol where timing is highly uncritical. And given that we need it to read a few bytes from EEPROM or RTC or similar I really see no benefit in using such a complicated driver.
I know :) but I like see 100kHz +/-0.01% when I configure 100kHz... I'm kidding, of course. eheheh... In the last time I preferred to disable i2c driver on all my applications... I ask me why all embedded controller on the SoC that I'm using are bugged... are the masks too expensive to work fine?
Ciao ciao
luigi
Best regards,
Wolfgang Denk
Ing. Luigi Mantellini Industrie Dial Face S.p.A. Via Canzo, 4 20068 Peschiera Borromeo (MI) Tel.: +39 02 5167 2813 Fax: +39 02 5167 2459 E-mail: luigi.mantellini@idf-hit.com