michele.scandale added a comment.

In D136786#3893485 <https://reviews.llvm.org/D136786#3893485>, @andrew.w.kaylor 
wrote:

> My understanding is this:
>
>   unsafe-fp-math = exception_behavior(ignore) +
>                    fenv_access(off) +
>                    no_signed_zeros(on) +
>                    allow_reciprocal(on) +
>                    allow_approximate_fns(on) +
>                    allow_reassociation(on) +
>                    fp_contract(fast)
>
> The first two of those settings are the default behavior. The others can be 
> turned on by individual command line options. If all of those semantic modes 
> are in the states above, then we are in "unsafe-fp-math" mode. That's my 
> proposal.

I've updated the patch to use the low level options as you suggested. In the 
current version of the patch I'm not currently checking explicitly 
`exception_behavior(ignore) &&  fenv_access(off)` given that the other modes of 
these properties seems to work only if overall there is the "precise" profile. 
I don't know if that's acceptable, or if instead the CC1 shouldn't make any 
assumption about the compiler driver doing the right thing.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D136786/new/

https://reviews.llvm.org/D136786

_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to