From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <775b8d190711191200g6fce7fb6i26ef414a0dca9cd6@mail.gmail.com> Date: Tue, 20 Nov 2007 07:00:27 +1100 From: "Bruce Ellis" To: "Fans of the OS Plan 9 from Bell Labs" <9fans@cse.psu.edu> Subject: Re: [9fans] 8c "out of fixed registers" In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline References: <9cd4aaed941882632a7f43cfc9811297@quanstro.net> Topicbox-Message-UUID: 020cc3fc-ead3-11e9-9d60-3106f5b1d025 it's my fault. 8c is designed not to run out of registers and when i added the division by constant optimization i thought i could get away with a temporary. it works nearly all the time :-) a simple fix is to check for register squeeze and bypass the optimization if it's going to fail. there are other smarter, and not too complex, alternatives. brucee On Nov 20, 2007 5:14 AM, Charles Forsyth wrote: > >does the function that causes 8c to quit declare a large > >number of "register int" variables? > > it ignores you, and works that out itself. > > usually the troublesome expression really does use > so many registers it needs to spill to memory, but doesn't.