2017-12-05 8:44 GMT+01:00 Sakari Ailus <sakari.ailus@xxxxxxxxxxxxxxx>: > On Mon, Dec 04, 2017 at 05:24:33PM -0500, Sven Van Asbroeck wrote: >> > If this is truly specific to at24, then vendor prefix would be appropriate, >> > plus it'd go to an at24 specific binding file. However if it isn't I'd just >> > remove the above sentence. I guess the latter? >> >> Yes, no-read-rollover is truly specific to at24.c, because it applies only >> to i2c multi-address chips. The at25 is spi based so cannot have multiple >> addresses. >> >> So yes, "at24,no-read-rollover" would perhaps be a better name. >> >> Regarding an at24 specific binding file. You're saying I should create >> Documentation/devicetree/bindings/eeprom/at24.txt ? Should I indicate >> that at24.txt "inherits from" eeprom.txt? Note that at25.txt does not >> currently do this. > > Hmm. I actually missed we didn't have one to begin with. at25.txt exists > and it documents a number of properties specific to at25, so if at24 will > have an at24-specific property, then I think it should go to a separate > file. The eeprom.txt file in the bindings directory actually describes the bindings for at24. There's a patch[1] from Wolfram waiting for Rob's ack that renames it to at24.txt. I hope that clears any confusion. @Sven: please split the patch into two: one for bindings and one for code. As for the name: I would change it to at24,no-read-rollover and remove the fragment saying it's only supported in at24 - as I said: this file only concerns at24 and will be renamed. > > Aren't there really other chips which need this? It'd be (a little bit) > easier to just remove the sentence. :-) > > -- > Regards, > > Sakari Ailus > sakari.ailus@xxxxxxxxxxxxxxx Thanks, Bartosz [1] http://patchwork.ozlabs.org/patch/842500/
