This is the mail archive of the
cgen@sourceware.org
mailing list for the CGEN project.
Re: subword semantics
- From: Dave Brolley <brolley at redhat dot com>
- To: Joern Rennecke <joernr at arc dot com>
- Cc: cgen at sources dot redhat dot com
- Date: Fri, 19 Jan 2007 10:30:02 -0500
- Subject: Re: subword semantics
- References: <20070119120733.GA2043@elsdt-razorfish.arc.com>
Joern Rennecke wrote:
What are the actual semantics of subword?
The documentation says that it is endian dependent, but the existing
bi-endian ports always use word 0 to refer to the most significant part.
You are correct. The semantics are not endian dependent. Similar to the
way shift operators are not endian dependent. The documentation also
says that dependence on endianness was to be revisited. Perhaps that's
what happend, or perhaps the intent was to recommend caution when using
a (mem ...) expression as the operand of (subword ...)
Dave