Re: [PATCH, newlib] Allow locking routine to be retargeted

[Thomas Preudhomme <thomas dot preudhomme at foss dot arm dot com>]

Hi Freddie,

Thanks for your feedback.

On 10/11/16 15:33, Freddie Chopin wrote:
> Hi Thomas!
>
> Why don't you do the same thing for recursive functions? At least the
> lock used by malloc() has to be recursive, so with your patch exactly
> all mutexes should be recursive too.

Yes, I did not want to expose all functions until there is a need to 
differentiate. it is easy to add them later but removing it if noone use it 
would be quite difficult.

> I have doubts about practical implementation of these functions for any
> RTOS, because of the __LOCK_INIT() macro used for initialization. In
> every retargeted function you'll have to start critical section (most
> likely by disabling interrupts) to initialize the object on heap, but
> then how would you use heap if malloc()'s lock is used via these
> functions too?

I do not understand. Why would __LOCK_INIT need to start a critical section?

Best regards,

Thomas