On Wed, 2016-03-16 at 16:32 +0200, Grygorii Strashko wrote: > On 03/16/2016 02:10 PM, Ben Hutchings wrote: > > The binding definition for the PCF857x GPIO expanders doesn't mention > > a "ti,pcf8575" compatible string. This is apparently because TI is > > only a second source - there is no functional difference between > > PCF8575 chips manufactured by TI and NXP, and the same board might be > > populated with either depending on availability. > > > > As a practical matter, the gpio-pcf857x driver doesn't match this > > string, so these device trees currently don't work. > > Sry, but could you explain more about "these device trees currently don't work"? > On my boards pcf_gpio_21: gpio@21 is enumerated, because the > i2c core uses modalias (of_modalias_node()). The main problem I had was that the gpio-pcf857x driver wasn't enabled in my kernel config, so probing of dependent devices would return -EPROBE_DEFER. When looking for the missing driver I initially searched for the compatible string and found nothing, which led me to change the compatible string. I have to admit I *didn't* test with just the driver enabled. Having done that, I agree that this isn't a problem in practice. I think the point about using an undefined compatible string stands, so I'll re-send with a different commit message. Ben. > root@dra7xx-evm:/# cat /proc/device-tree/ocp/i2c\@48070000/gpio\@21/compatible > ti,pcf8575 > > root@dra7xx-evm:/# cat /sys/kernel/debug/gpio > GPIOs 480-495, i2c/0-0021, pcf8575, can sleep: > gpio-481 (id ) in hi IRQ > gpio-482 (id ) in hi IRQ > gpio-485 (evm_3v3_sd ) out hi > gpio-494 (enable ) out lo > > just worry if I missed smth. > > -- Ben Hutchings Software Developer, Codethink Ltd. -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
