Ah - ARMEL and ARMEB tests were already there. Sorry, missed them the first time.
I'd like some 'NOT' checks. It may be best to do these as second tests in init.c, so that you don't confine the check to the range between two matches. I'd like to see that __ARM_BIG_ENDIAN and __ARMEB__ are not defined in the 'arm' case, and that __ARMEL__ is not defined in the armeb case. My question about -target armeb-none-darwin still stands (re inline comment from before). Other than those points, LGTM, with the disclaimer that I'm no expert on the various platform triples you're setting up. More generally, I think I'd like to see a CLI option to select endianness - I would prefer that as the 'user interface' over relying on another special triple name. That doesn't have to be part of this patch, though. (I haven't - yet - looked to see what other endian-bending targets do.) http://llvm-reviews.chandlerc.com/D3096 _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
