Re: [avr-gcc-list] Most reliable version of avr-gcc?

[Bob Paddock]

On Wed, Dec 17, 2008 at 6:37 PM, David Carr <address@hidden> wrote:

By reliability, I mean least probability of undetected errors in machine code generation.  IE: The machine code conforms to the source code.

"How to verify your compiler for use in IEC 61508 safety-critical applications; A step-by-step guide to what's necessary to satisfy the IEC61508 safety-critical requirements for compiler verification."

http://www.embedded.com/design/opensource/202800510

You have to disassemble the HEX file to be sure what you are putting in the device,
is really what you wanted.  I have personally been bitten by an assembler that made
a good listing and bad HEX file.  That was fun to debug.  As was the CPU (Non-AVR)
that had a bad XOR instruction, but only on certain bit patterns.  This why the
new "White Goods" standards is harder to pass self-test, than some the things that
really could kill you.  Gets even more fun when the standard says the device must
be in operational mode in under one second...

Is there some standard you are aiming for like FDA or DO178D?

See if there is anything that might answer your question at my site:
http://www.softwaresafety.net/

-- 
http://www.wearablesmartsensors.com/
http://www.softwaresafety.net/
http://www.designer-iii.com/
http://www.unusualresearch.com/