svn commit: r307231 - head/lib/libgcc_s

[Dimitry Andric]

On 14 Oct 2016, at 04:06, Ed Maste <emaste at freebsd.org> wrote:
> 
> Hi Bruce, thank you for the detailed response.
> 
> On 14 October 2016 at 01:53, Bruce Evans <brde at optusnet.com.au> wrote:
>>> compiler-rt's complex division support routines contain calls to
>>> compiler builtins such as `__builtin_scalbnl`.  Unfortunately Clang
>>> turns these back into a call to `scalbnl`.
>> 
>> gcc-4.2 has the same bug.
> 
> Oh, interesting. Do you know off hand if it's resolved in later GCC?

Not for x86, in any case (I think I talked about this with you on IRC
before).  For example, using the latest gcc 7.0.0 20160911 on this
program:

float f1(float x, int n)
{
	return __builtin_scalbnf(x, n);
}

double f2(double x, int n)
{
	return __builtin_scalbn(x, n);
}

long double f3(long double x, long n)
{
	return __builtin_scalbnl(x, n);
}

gives this assembly:

	.p2align 4,,15
	.globl	f1
	.type	f1, @function
f1:
.LFB0:
	.cfi_startproc
	jmp	scalbnf
	.cfi_endproc
.LFE0:
	.size	f1, .-f1
	.p2align 4,,15
	.globl	f2
	.type	f2, @function
f2:
.LFB1:
	.cfi_startproc
	jmp	scalbn
	.cfi_endproc
.LFE1:
	.size	f2, .-f2
	.p2align 4,,15
	.globl	f3
	.type	f3, @function
f3:
.LFB2:
	.cfi_startproc
	jmp	scalbnl
	.cfi_endproc
.LFE2:
	.size	f3, .-f3

E.g., it looks like the __builtin_scalbn functions were always a red
herring. :-)


> It seems particularly unfortunate for the compiler to report (by
> whatever mechanism) that a builtin exists, and then just turn that
> builtin into a library call.

Yes, at best this is misleading.

-Dimitry

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 194 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.freebsd.org/pipermail/svn-src-head/attachments/20161014/5d35385c/attachment.sig>