On Wed, 8 May 2024, Richard Biener wrote: > So it's reasonable to require -fnon-call-exceptions (which now enables > -fexceptions) and -fno-delete-dead-exceptions to have GCC preserve > a change of control flow side-effect of x*x? We do not preserve > FP exception bits set by otherwise unused operations, that is, we > do not consider that side-effect to be observable even with > -ftrapping-math. In fact I most uses of flag_trapping_math > are related to a possible control flow side-effect of FP math.
I think lots are about avoiding changing exceptions raised (flags set) by an operation that is seen as being used. > Exact preservation of FP exception flags will likely have to disable > all FP optimization if one considers FE_INEXACT and FE_UNDERFLOW. It's very likely that as per previous discussions we need some different set of options from the current ones to properly reflect the various ways in which preserving how operations interact with the floating-point environments (exception flags, exceptions changing flow of control, rounding modes; various of which may also depend on whether the results of operations appear to the compiler to be used) may inhibit optimization. -- Joseph S. Myers josmy...@redhat.com
