Re: [PATCH 1/3] protect some clobbered variables with volatile

[Jens Gustedt <jens.gustedt@inria.fr>]

[-- Attachment #1: Type: text/plain, Size: 1165 bytes --]

Hello Rich,
AFAIR these both occurred when compiling with -flto. I would not trust
the static analysis of the compiler at 100 % either, but this is just
an argument to have the volatile in.

The compiler is telling us that here he thinks that an optimization
might hide a intermittent write, but that by the aliasing rules it
would be OK to save a read. So better be careful and trust the
compiler that he is willing to do bad things :)

Maybe LTO is broken, yet, in that it might do some optimizations that
it shouldn't. Or maybe this exposes possible problems that were
previously hidden because the write was in a different compilation
unit than the read. Who knows.

In any case, the performance hit should be limited, and my personal
preference would tend towards safety and calming down the compiler.

Jens

-- 
:: INRIA Nancy Grand Est :: http://www.loria.fr/~gustedt/   ::
:: AlGorille ::::::::::::::: office Nancy : +33 383593090   ::
:: ICube :::::::::::::: office Strasbourg : +33 368854536   ::
:: ::::::::::::::::::::::::::: gsm France : +33 651400183   ::
:: :::::::::::::::::::: gsm international : +49 15737185122 ::




[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]