Re: [avr-gcc-list] cbr and negative numbers

Ruud Vlaming <[EMAIL PROTECTED]> wrote: > Yeah, you are correct. And if i mange to dive into the problem and > fix it myself (which i doubt i can) the patch would probably not be > committed because of possible side effects. GCC has a fairly well developed set of tests. If your patch doesn't cau

Re: [avr-gcc-list] cbr and negative numbers

Ruud Vlaming <[EMAIL PROTECTED]> wrote: > The problem i am pointing out can (easily) be solved by letting gcc > emit only positive/hexadecimal operands or by a minimal change in > the way the cbr instruction calculates its operand: (0xFF-x) -> > (0xFF - (x & 0xFF)). If you've got a patch that doe

Re: [avr-gcc-list] cbr and negative numbers

You can file bug report against gcc. Then there is a chance that it will be fixed. (but no guarantee). mint8 bugs have been fixed before - particulary if they are easy. Apology if you have already submitted bug report. -- Sent from my Dingleberry wi

Re: [avr-gcc-list] cbr and negative numbers

On Wednesday 10 September 2008 23:45, Weddington, Eric wrote: > But I highly doubt that the other developers are going to fix a problem > that only shows up with -mint8. Yeah, you are correct. And if i mange to dive into the problem and fix it myself (which i doubt i can) the patch would proba

Re: [avr-gcc-list] Re: cbr and negative numbers

On Wednesday 10 September 2008 23:15, David Brown wrote: > Have you tried using "M" as the constraint rather than "i" ? I haven't > tried this myself, but according to > "M" is an > 8-bit integer constant in the range 0 to 255. Tha