[llvm-dev] clang 4.0.0: Invalid code for builtin floating point function with -mfloat-abi=hard -ffast-math (ARM)

[Friedman, Eli via llvm-dev]

On 3/22/2017 1:42 PM, Peter Jakubek wrote:
> Am 22.03.2017 um 02:38 schrieb Friedman, Eli:
>>
>> That is probably unintentional.  Granted, using -mfloat-abi like this is
>> kind of weird, but I think clang's behavior is supposed to be
>> gcc-compatible.  See https://reviews.llvm.org/rL291909 and
>> https://bugs.llvm.org/show_bug.cgi?id=30543 for the most recent work in
>> this area.
>
> why is using -mfloat-abi weird?

Err, oh, I think I misunderstood how it's supposed to work the last time 
I looked. I guess it's supposed to be the same as using --target?

> And I do not agree that whether functions are treated as builtin 
> should depend on -ffast-math.
> (it does not with gcc)

Err, oh, sorry, that isn't really what I meant.  We treat it as a C 
library function either way; we just model it in the IR differently 
because we don't want to model modifications to errno in fast-math mode.

-Eli

-- 
Employee of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project