libi2c: Claiming driver slot failed

Chris Johns chrisj at
Mon Aug 12 08:08:24 UTC 2013

Sebastian Huber wrote:
> On 2013-08-12 05:10, Chris Johns wrote:
>> Nick Withers wrote:
>>> It seems to me like there's an inversion of the App -> BSP -> RTEMS
>>> abstraction going on here, where the app effectively needs to know
>>> "black-box" details of the BSP, making maintenence and portability more
>>> difficult and error prone than needs be...?
>> I would like to see the drivers and the device table become unlimited
>> just like
>> the other unlimited resources in the kernel. The value becomes an
>> allocation unit.
> We should not start adding new tables, instead we should get rid of
> them. Propagating this major/minor number concept down to the lowest
> implementation levels is a major design problem from my point of view.
> For drivers I would use IMFS generic nodes (e.g. a RTEMS device file
> system).
> It is possible to strip down IMFS to get a directory + generic nodes
> subset with very low overhead. It is also possible to use static only
> memory (no malloc()) for this.

This is a much better idea. You have my support for this.


More information about the users mailing list