r262688 - [X86] Pass __m64 types via SSE registers for GCC compatibility
Robinson, Paul via cfe-commits
cfe-commits at lists.llvm.org
Fri Mar 4 08:54:02 PST 2016
> It'd be nice to have a comment here that mentions that the clang
> behavior which is being preserved for Darwin, FreeBSD, and PS4 is a
> *bug* which is being intentionally left unfixed. The previous clang
> behavior directly contradicts the x86_64 ABI document, which I believe
> all of these platforms claim to follow. :)
Well, PS4 uses x86_64 ABI as a base document, but we have a handful of
variances. We had already documented this one to our licensees. So,
from our perspective, it's not a bug, it's a feature. :-) Describing
it as a bug (at least for PS4) would be technically incorrect.
--paulr
>
> On Fri, Mar 4, 2016 at 2:03 AM, Robinson, Paul via cfe-commits
> <cfe-commits at lists.llvm.org> wrote:
> >> To: cfe-commits at lists.llvm.org
> >> Subject: r262688 - [X86] Pass __m64 types via SSE registers for GCC
> >> compatibility
> >>
> >> Author: majnemer
> >> Date: Thu Mar 3 23:26:16 2016
> >> New Revision: 262688
> >>
> >> URL: http://llvm.org/viewvc/llvm-project?rev=262688&view=rev
> >> Log:
> >> [X86] Pass __m64 types via SSE registers for GCC compatibility
> >>
> >> For compatibility with GCC, classify __m64 as SSE.
> >> However, clang is a platform compiler for certain targets; retain our
> >> old behavior on those targets: classify __m64 as integer.
> >
> > Thank you very much for that!
> > --paulr
> >
> >>
> >> This fixes PR26832.
> >>
> >> Modified:
> >> cfe/trunk/lib/CodeGen/TargetInfo.cpp
> >> cfe/trunk/test/CodeGen/3dnow-builtins.c
> >> cfe/trunk/test/CodeGen/x86_64-arguments.c
> >>
> >> Modified: cfe/trunk/lib/CodeGen/TargetInfo.cpp
> >> URL: http://llvm.org/viewvc/llvm-
> >>
> project/cfe/trunk/lib/CodeGen/TargetInfo.cpp?rev=262688&r1=262687&r2=26268
> >> 8&view=diff
> >>
> ==========================================================================
> >> ====
> >> --- cfe/trunk/lib/CodeGen/TargetInfo.cpp (original)
> >> +++ cfe/trunk/lib/CodeGen/TargetInfo.cpp Thu Mar 3 23:26:16 2016
> >> @@ -1857,6 +1857,17 @@ class X86_64ABIInfo : public ABIInfo {
> >> return !getTarget().getTriple().isOSDarwin();
> >> }
> >>
> >> + /// GCC classifies <1 x long long> as SSE but compatibility with
> older
> >> clang
> >> + // compilers require us to classify it as INTEGER.
> >> + bool classifyIntegerMMXAsSSE() const {
> >> + const llvm::Triple &Triple = getTarget().getTriple();
> >> + if (Triple.isOSDarwin() || Triple.getOS() == llvm::Triple::PS4)
> >> + return false;
> >> + if (Triple.isOSFreeBSD() && Triple.getOSMajorVersion() >= 10)
> >> + return false;
> >> + return true;
> >> + }
> >> +
> >> X86AVXABILevel AVXLevel;
> >> // Some ABIs (e.g. X32 ABI and Native Client OS) use 32 bit pointers
> on
> >> // 64-bit hardware.
> >> @@ -2298,15 +2309,20 @@ void X86_64ABIInfo::classify(QualType Ty
> >> if (EB_Lo != EB_Hi)
> >> Hi = Lo;
> >> } else if (Size == 64) {
> >> + QualType ElementType = VT->getElementType();
> >> +
> >> // gcc passes <1 x double> in memory. :(
> >> - if (VT->getElementType()-
> >> >isSpecificBuiltinType(BuiltinType::Double))
> >> + if (ElementType->isSpecificBuiltinType(BuiltinType::Double))
> >> return;
> >>
> >> - // gcc passes <1 x long long> as INTEGER.
> >> - if (VT->getElementType()-
> >> >isSpecificBuiltinType(BuiltinType::LongLong) ||
> >> - VT->getElementType()-
> >> >isSpecificBuiltinType(BuiltinType::ULongLong) ||
> >> - VT->getElementType()-
> >isSpecificBuiltinType(BuiltinType::Long)
> >> ||
> >> - VT->getElementType()-
> >> >isSpecificBuiltinType(BuiltinType::ULong))
> >> + // gcc passes <1 x long long> as SSE but clang used to
> >> unconditionally
> >> + // pass them as integer. For platforms where clang is the de
> facto
> >> + // platform compiler, we must continue to use integer.
> >> + if (!classifyIntegerMMXAsSSE() &&
> >> + (ElementType->isSpecificBuiltinType(BuiltinType::LongLong)
> ||
> >> + ElementType->isSpecificBuiltinType(BuiltinType::ULongLong)
> ||
> >> + ElementType->isSpecificBuiltinType(BuiltinType::Long) ||
> >> + ElementType->isSpecificBuiltinType(BuiltinType::ULong)))
> >> Current = Integer;
> >> else
> >> Current = SSE;
> >>
> >> Modified: cfe/trunk/test/CodeGen/3dnow-builtins.c
> >> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/3dnow-
> >> builtins.c?rev=262688&r1=262687&r2=262688&view=diff
> >>
> ==========================================================================
> >> ====
> >> --- cfe/trunk/test/CodeGen/3dnow-builtins.c (original)
> >> +++ cfe/trunk/test/CodeGen/3dnow-builtins.c Thu Mar 3 23:26:16 2016
> >> @@ -1,4 +1,5 @@
> >> -// RUN: %clang_cc1 %s -triple=x86_64-unknown-unknown -target-feature
> >> +3dnowa -emit-llvm -o - -Werror | FileCheck %s
> >> +// RUN: %clang_cc1 %s -triple=x86_64-unknown-unknown -target-feature
> >> +3dnowa -emit-llvm -o - -Werror | FileCheck %s -check-prefix=GCC -
> check-
> >> prefix=CHECK
> >> +// RUN: %clang_cc1 %s -triple=x86_64-scei-ps4 -target-feature +3dnowa
> -
> >> emit-llvm -o - -Werror | FileCheck %s -check-prefix=PS4 -check-
> >> prefix=CHECK
> >>
> >> // Don't include mm_malloc.h, it's system specific.
> >> #define __MM_MALLOC_H
> >> @@ -6,151 +7,176 @@
> >> #include <x86intrin.h>
> >>
> >> __m64 test_m_pavgusb(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pavgusb
> >> + // PS4-LABEL: define i64 @test_m_pavgusb
> >> + // GCC-LABEL: define double @test_m_pavgusb
> >> // CHECK: @llvm.x86.3dnow.pavgusb
> >> return _m_pavgusb(m1, m2);
> >> }
> >>
> >> __m64 test_m_pf2id(__m64 m) {
> >> - // CHECK-LABEL: define i64 @test_m_pf2id
> >> + // PS4-LABEL: define i64 @test_m_pf2id
> >> + // GCC-LABEL: define double @test_m_pf2id
> >> // CHECK: @llvm.x86.3dnow.pf2id
> >> return _m_pf2id(m);
> >> }
> >>
> >> __m64 test_m_pfacc(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfacc
> >> + // PS4-LABEL: define i64 @test_m_pfacc
> >> + // GCC-LABEL: define double @test_m_pfacc
> >> // CHECK: @llvm.x86.3dnow.pfacc
> >> return _m_pfacc(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfadd(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfadd
> >> + // PS4-LABEL: define i64 @test_m_pfadd
> >> + // GCC-LABEL: define double @test_m_pfadd
> >> // CHECK: @llvm.x86.3dnow.pfadd
> >> return _m_pfadd(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfcmpeq(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfcmpeq
> >> + // PS4-LABEL: define i64 @test_m_pfcmpeq
> >> + // GCC-LABEL: define double @test_m_pfcmpeq
> >> // CHECK: @llvm.x86.3dnow.pfcmpeq
> >> return _m_pfcmpeq(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfcmpge(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfcmpge
> >> + // PS4-LABEL: define i64 @test_m_pfcmpge
> >> + // GCC-LABEL: define double @test_m_pfcmpge
> >> // CHECK: @llvm.x86.3dnow.pfcmpge
> >> return _m_pfcmpge(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfcmpgt(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfcmpgt
> >> + // PS4-LABEL: define i64 @test_m_pfcmpgt
> >> + // GCC-LABEL: define double @test_m_pfcmpgt
> >> // CHECK: @llvm.x86.3dnow.pfcmpgt
> >> return _m_pfcmpgt(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfmax(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfmax
> >> + // PS4-LABEL: define i64 @test_m_pfmax
> >> + // GCC-LABEL: define double @test_m_pfmax
> >> // CHECK: @llvm.x86.3dnow.pfmax
> >> return _m_pfmax(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfmin(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfmin
> >> + // PS4-LABEL: define i64 @test_m_pfmin
> >> + // GCC-LABEL: define double @test_m_pfmin
> >> // CHECK: @llvm.x86.3dnow.pfmin
> >> return _m_pfmin(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfmul(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfmul
> >> + // PS4-LABEL: define i64 @test_m_pfmul
> >> + // GCC-LABEL: define double @test_m_pfmul
> >> // CHECK: @llvm.x86.3dnow.pfmul
> >> return _m_pfmul(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfrcp(__m64 m) {
> >> - // CHECK-LABEL: define i64 @test_m_pfrcp
> >> + // PS4-LABEL: define i64 @test_m_pfrcp
> >> + // GCC-LABEL: define double @test_m_pfrcp
> >> // CHECK: @llvm.x86.3dnow.pfrcp
> >> return _m_pfrcp(m);
> >> }
> >>
> >> __m64 test_m_pfrcpit1(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfrcpit1
> >> + // PS4-LABEL: define i64 @test_m_pfrcpit1
> >> + // GCC-LABEL: define double @test_m_pfrcpit1
> >> // CHECK: @llvm.x86.3dnow.pfrcpit1
> >> return _m_pfrcpit1(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfrcpit2(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfrcpit2
> >> + // PS4-LABEL: define i64 @test_m_pfrcpit2
> >> + // GCC-LABEL: define double @test_m_pfrcpit2
> >> // CHECK: @llvm.x86.3dnow.pfrcpit2
> >> return _m_pfrcpit2(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfrsqrt(__m64 m) {
> >> - // CHECK-LABEL: define i64 @test_m_pfrsqrt
> >> + // PS4-LABEL: define i64 @test_m_pfrsqrt
> >> + // GCC-LABEL: define double @test_m_pfrsqrt
> >> // CHECK: @llvm.x86.3dnow.pfrsqrt
> >> return _m_pfrsqrt(m);
> >> }
> >>
> >> __m64 test_m_pfrsqrtit1(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfrsqrtit1
> >> + // PS4-LABEL: define i64 @test_m_pfrsqrtit1
> >> + // GCC-LABEL: define double @test_m_pfrsqrtit1
> >> // CHECK: @llvm.x86.3dnow.pfrsqit1
> >> return _m_pfrsqrtit1(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfsub(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfsub
> >> + // PS4-LABEL: define i64 @test_m_pfsub
> >> + // GCC-LABEL: define double @test_m_pfsub
> >> // CHECK: @llvm.x86.3dnow.pfsub
> >> return _m_pfsub(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfsubr(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfsubr
> >> + // PS4-LABEL: define i64 @test_m_pfsubr
> >> + // GCC-LABEL: define double @test_m_pfsubr
> >> // CHECK: @llvm.x86.3dnow.pfsubr
> >> return _m_pfsubr(m1, m2);
> >> }
> >>
> >> __m64 test_m_pi2fd(__m64 m) {
> >> - // CHECK-LABEL: define i64 @test_m_pi2fd
> >> + // PS4-LABEL: define i64 @test_m_pi2fd
> >> + // GCC-LABEL: define double @test_m_pi2fd
> >> // CHECK: @llvm.x86.3dnow.pi2fd
> >> return _m_pi2fd(m);
> >> }
> >>
> >> __m64 test_m_pmulhrw(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pmulhrw
> >> + // PS4-LABEL: define i64 @test_m_pmulhrw
> >> + // GCC-LABEL: define double @test_m_pmulhrw
> >> // CHECK: @llvm.x86.3dnow.pmulhrw
> >> return _m_pmulhrw(m1, m2);
> >> }
> >>
> >> __m64 test_m_pf2iw(__m64 m) {
> >> - // CHECK-LABEL: define i64 @test_m_pf2iw
> >> + // PS4-LABEL: define i64 @test_m_pf2iw
> >> + // GCC-LABEL: define double @test_m_pf2iw
> >> // CHECK: @llvm.x86.3dnowa.pf2iw
> >> return _m_pf2iw(m);
> >> }
> >>
> >> __m64 test_m_pfnacc(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfnacc
> >> + // PS4-LABEL: define i64 @test_m_pfnacc
> >> + // GCC-LABEL: define double @test_m_pfnacc
> >> // CHECK: @llvm.x86.3dnowa.pfnacc
> >> return _m_pfnacc(m1, m2);
> >> }
> >>
> >> __m64 test_m_pfpnacc(__m64 m1, __m64 m2) {
> >> - // CHECK-LABEL: define i64 @test_m_pfpnacc
> >> + // PS4-LABEL: define i64 @test_m_pfpnacc
> >> + // GCC-LABEL: define double @test_m_pfpnacc
> >> // CHECK: @llvm.x86.3dnowa.pfpnacc
> >> return _m_pfpnacc(m1, m2);
> >> }
> >>
> >> __m64 test_m_pi2fw(__m64 m) {
> >> - // CHECK-LABEL: define i64 @test_m_pi2fw
> >> + // PS4-LABEL: define i64 @test_m_pi2fw
> >> + // GCC-LABEL: define double @test_m_pi2fw
> >> // CHECK: @llvm.x86.3dnowa.pi2fw
> >> return _m_pi2fw(m);
> >> }
> >>
> >> __m64 test_m_pswapdsf(__m64 m) {
> >> - // CHECK-LABEL: define i64 @test_m_pswapdsf
> >> + // PS4-LABEL: define i64 @test_m_pswapdsf
> >> + // GCC-LABEL: define double @test_m_pswapdsf
> >> // CHECK: @llvm.x86.3dnowa.pswapd
> >> return _m_pswapdsf(m);
> >> }
> >>
> >> __m64 test_m_pswapdsi(__m64 m) {
> >> - // CHECK-LABEL: define i64 @test_m_pswapdsi
> >> + // PS4-LABEL: define i64 @test_m_pswapdsi
> >> + // GCC-LABEL: define double @test_m_pswapdsi
> >> // CHECK: @llvm.x86.3dnowa.pswapd
> >> return _m_pswapdsi(m);
> >> }
> >>
> >> Modified: cfe/trunk/test/CodeGen/x86_64-arguments.c
> >> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/x86_64-
> >> arguments.c?rev=262688&r1=262687&r2=262688&view=diff
> >>
> ==========================================================================
> >> ====
> >> --- cfe/trunk/test/CodeGen/x86_64-arguments.c (original)
> >> +++ cfe/trunk/test/CodeGen/x86_64-arguments.c Thu Mar 3 23:26:16 2016
> >> @@ -261,12 +261,12 @@ void f33(va_list X) {
> >> typedef unsigned long long v1i64 __attribute__((__vector_size__(8)));
> >>
> >> // rdar://8359248
> >> -// CHECK-LABEL: define i64 @f34(i64 %arg.coerce)
> >> +// CHECK-LABEL: define double @f34(double %arg.coerce)
> >> v1i64 f34(v1i64 arg) { return arg; }
> >>
> >>
> >> // rdar://8358475
> >> -// CHECK-LABEL: define i64 @f35(i64 %arg.coerce)
> >> +// CHECK-LABEL: define double @f35(double %arg.coerce)
> >> typedef unsigned long v1i64_2 __attribute__((__vector_size__(8)));
> >> v1i64_2 f35(v1i64_2 arg) { return arg+arg; }
> >>
> >>
> >>
> >> _______________________________________________
> >> cfe-commits mailing list
> >> cfe-commits at lists.llvm.org
> >> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
> > _______________________________________________
> > cfe-commits mailing list
> > cfe-commits at lists.llvm.org
> > http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
More information about the cfe-commits
mailing list