Re: [ECOS] Order in which CDL scripts are processed

["Mike Sweeney" <msweeney77 at gmail dot com>]

On 6/3/07, Gary Thomas <gary@mlbassoc.com> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Mike Sweeney wrote:
> > While looking through the code for the CYGPKG_HAL_ARM_XSCALE_CORE
> > package I noticed that it exports some header files to cyg/hal in the
> > install tree. In particular, it exports hal_cache.h which contains
> > some cache size definitions and such. Furthermore, some of the generic
> > ARM platform HAL packages such as the CYGPKG_HAL_ARM_AEB export
> > hal_cache.h to the same directory. Now, granted since the generic ARM
> > targets won't include the XScale package there won't be a conflict,
> > but out of curiousity if I did export a hal_cache.h from an XScale
> > variant HAL package is there any guarantee as to which hal_cache.h
> > would end up in the install tree under include/cyg/hal, i.e. the one
> > from CYGPKG_HAL_ARM_XSCALE_CORE or the one from my XScale variant HAL
> > package?
> >
> > In a nutshell, if I export the same header file to the same directory
> > from two different packages, which one wins out or is it undefined
> > behavior?
>
> Undefined and probably no guaranteed behaviour, so don't do it.
>
> If you need platform and/or variant changes/additions, there are
> files for that (which are in turn included by the main hal_cache.h)
>
>
> - --
> - ------------------------------------------------------------
> Gary Thomas                 |  Consulting for the
> MLB Associates              |    Embedded world
> - ------------------------------------------------------------

Thanks, I wasn't going to do it. I was just curious because I thought
that XScale cores could have either 32KB or 16KB(read this in XScale
core developer's manual) of data and instruction cache based on
implementation but the cache size seems to be defined to be 32KB in
the XScale package which parents the XScale variant and platform HALs.

--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss