harald.kipp at egnite.de
Wed Sep 5 09:21:05 CEST 2012
On 04.09.2012 16:01, Uwe Bonnes wrote:
>>>>>> "Harald" == Harald Kipp <harald.kipp at egnite.de> writes:
> Harald> Hi Uwe, On 04.09.2012 11:43, Uwe Bonnes wrote:
> >> - For __CORTEX__ asp and ssi handling were not enabled
> Harald> Using the "marketing" macro __CORTEX__ instead of the build-in
> Harald> __ARM_ARCH_7M__ never made any sense to me. Your post shows,
> Harald> that it even creates confusion.
> Either I "stand on the conductor" (translated from a german saying) or you
> are too tense. Do you want to replace __CORTEX__ with __ARM_ARCH_7M__ ? But
> CORTEX isn't M7. The arm site
> lists Cortex_A?Cortex_R/Cortex_M/Arm11/Arm9 and Arm7. So for me
> __ARM_ARCH_7M__ is something different then Cortex_M.
The predefined macros are not referring to product family names, but to the underlying architecture. See
IMHO, this is the only interesting information for code separation. However, the majority of Nut/OS users seems to prefer the popular product names. See the thread that started about one year ago
I never liked this decision and Ole recently followed my view (his post from Oct. 23rd). There had been no protests so far, so we should move from __CORTEX__ to GCC's build-in macros. Note, that ARMv7 does have several variants and also many things in common. But as far as I can say, __ARM_ARCH_7M__ is the only architecture that's currently supported by Nut/OS for the Cortex family.
More information about the En-Nut-Discussion