[M3devel] closure marker

Jay K jay.krell at cornell.edu
Wed Nov 28 04:48:05 CET 2012


I finally did some research.. on PowerPC, 4 bytes 0xFF is a valid opcode.4 bytes of 0 is not.  Perhaps the closure marker should be target-dependent?Default to -1?And 0 on PowerPC?  Granted, -1 isn't all that common, it is, I think fnmadd. -- floating point negative multiply-add, recording form, operating all on the same register, fr31. And non-leaf functions typically start with "mflr" -- move from link register.  But my old 601 manual specifically calls out 0 as invalid opcode -- by design.   I still haven't done this research for SPARC, IA64, MIPS, x86, etc.(I'm still susicipious that for IA64, a larger marker should be used, since instructions are in 128bit bundles. As well, for SPARC64, MIPS64, ALPHA64, PPC64, the marker probably need be only 4 bytes, and the aligned check might be more efficient.)   - Jay


 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://m3lists.elegosoft.com/pipermail/m3devel/attachments/20121128/ba8811bc/attachment-0001.html>


More information about the M3devel mailing list