YAMAMOTO Takashi wrote:
>> at the moment I don't see a serious benefit to changing the code,
> 
> reduce code duplication?
> 
>> and I
>> tend to prefer the explicit version (what we have now).
> 
> why?

in the example you provided, can you elaborate on what you mean
by "unifdef -D"? iiuc, the secmodel code *we* ship (the one in our
cvs repository) will have those #ifdefs.

also, there are requests that have different handling depends on the
arch. for example, access to unmanaged memory on alpha requires code
in the securelevel listener, but for powerpc requires code in the
suser listener. how do you suggest to handle that?

-e.