[llvm] r199508 - Add two new calling conventions for runtime calls
Andrew Trick
atrick at apple.com
Fri Jan 17 15:42:59 PST 2014
On Jan 17, 2014, at 2:20 PM, Hal Finkel <hfinkel at anl.gov> wrote:
> Andy,
>
> Out of curiosity, is it worth not considering such calls to be scheduling barriers?
Maybe ;)
There has to be some strong motivation to add the risk and complexity, which I haven’t seen yet.
I’m not saying we can’t do it, just that it has to be really important for some target.
It’s quite strange for the scheduler to reason about latency across calls. *If* we had an earlier MI pass that performed coarse grained global code motion, which I think we should have ideally, then that would be a better place to hoist high latency ops across calls.
-Andy
> -Hal
>
> ----- Original Message -----
>> From: "Juergen Ributzka" <juergen at apple.com>
>> To: "Richard Osborne" <richard at xmos.com>
>> Cc: "LLVM Commits" <llvm-commits at cs.uiuc.edu>
>> Sent: Friday, January 17, 2014 4:03:03 PM
>> Subject: Re: [llvm] r199508 - Add two new calling conventions for runtime calls
>>
>> Hi Richard,
>>
>> right now the cold calling convention is very very similar to the
>> PreserveMost calling convention. The only differences are that
>> PreserveMost doesn’t preserve R11 for X86-64 and the inlining
>> heuristic never inlines cold calls. I think these are the only
>> semantic differences. Besides the semantic differences the name
>> “coldcc” represents the exact opposite of what the preserve_*cc are
>> used for. I will extend LangRef a little more to show the
>> difference.
>>
>> Cheers,
>> Juergen
>>
>> On Jan 17, 2014, at 1:48 PM, Richard Osborne <richard at xmos.com>
>> wrote:
>>
>>> As target-independent calling conventions how do they compare to
>>> the cold calling convention? It’s clear from the language ref
>>> description that preserve_all preserves more registers than the
>>> preserve_most but I’m not sure where coldcc fits in. Would it be
>>> possible to clarify this?
>>>
>>> On 17 Jan 2014, at 19:47, Juergen Ributzka <juergen at apple.com>
>>> wrote:
>>>
>>>> Author: ributzka
>>>> Date: Fri Jan 17 13:47:03 2014
>>>> New Revision: 199508
>>>>
>>>> URL: http://llvm.org/viewvc/llvm-project?rev=199508&view=rev
>>>> Log:
>>>> Add two new calling conventions for runtime calls
>>>>
>>>> This patch adds two new target-independent calling conventions for
>>>> runtime
>>>> calls - PreserveMost and PreserveAll.
>>>> The target-specific implementation for X86-64 is defined as
>>>> following:
>>>> - Arguments are passed as for the default C calling convention
>>>> - The same applies for the return value(s)
>>>> - PreserveMost preserves all GPRs - except R11
>>>> - PreserveAll preserves all GPRs and all XMMs/YMMs - except R11
>>>>
>>>> Reviewed by Lang and Philip
>>>>
>>>> Added:
>>>> llvm/trunk/test/CodeGen/X86/preserve_allcc64.ll
>>>> llvm/trunk/test/CodeGen/X86/preserve_mostcc64.ll
>>>> Modified:
>>>> llvm/trunk/docs/BitCodeFormat.rst
>>>> llvm/trunk/docs/LangRef.rst
>>>> llvm/trunk/include/llvm/IR/CallingConv.h
>>>> llvm/trunk/lib/AsmParser/LLLexer.cpp
>>>> llvm/trunk/lib/AsmParser/LLParser.cpp
>>>> llvm/trunk/lib/AsmParser/LLToken.h
>>>> llvm/trunk/lib/IR/AsmWriter.cpp
>>>> llvm/trunk/lib/Target/X86/X86CallingConv.td
>>>> llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp
>>>>
>>>> Modified: llvm/trunk/docs/BitCodeFormat.rst
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/BitCodeFormat.rst?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/docs/BitCodeFormat.rst (original)
>>>> +++ llvm/trunk/docs/BitCodeFormat.rst Fri Jan 17 13:47:03 2014
>>>> @@ -739,6 +739,8 @@ function. The operand fields are:
>>>> * ``coldcc``: code 9
>>>> * ``webkit_jscc``: code 12
>>>> * ``anyregcc``: code 13
>>>> + * ``preserve_mostcc``: code 14
>>>> + * ``preserve_allcc``: code 15
>>>> * ``x86_stdcallcc``: code 64
>>>> * ``x86_fastcallcc``: code 65
>>>> * ``arm_apcscc``: code 66
>>>>
>>>> Modified: llvm/trunk/docs/LangRef.rst
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/LangRef.rst?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/docs/LangRef.rst (original)
>>>> +++ llvm/trunk/docs/LangRef.rst Fri Jan 17 13:47:03 2014
>>>> @@ -359,6 +359,49 @@ added in the future:
>>>> allocated. This can currently only be used with calls to
>>>> llvm.experimental.patchpoint because only this intrinsic
>>>> records
>>>> the location of its arguments in a side table. See
>>>> :doc:`StackMaps`.
>>>> +"``preserve_mostcc``" - The `PreserveMost` calling convention
>>>> + This calling convention attempts to make the code in the
>>>> caller as little
>>>> + intrusive as possible. This calling convention behaves
>>>> identical to the `C`
>>>> + calling convention on how arguments and return values are
>>>> passed, but it
>>>> + uses a different set of caller/callee-saved registers. This
>>>> alleviates the
>>>> + burden of saving and recovering a large register set before
>>>> and after the
>>>> + call in the caller.
>>>> +
>>>> + - On X86-64 the callee preserves all general purpose
>>>> registers, except for
>>>> + R11. R11 can be used as a scratch register. Floating-point
>>>> registers
>>>> + (XMMs/YMMs) are not preserved and need to be saved by the
>>>> caller.
>>>> +
>>>> + The idea behind this convention is to support calls to
>>>> runtime functions
>>>> + that have a hot path and a cold path. The hot path is usually
>>>> a small piece
>>>> + of code that doesn't many registers. The cold path might need
>>>> to call out to
>>>> + another function and therefore only needs to preserve the
>>>> caller-saved
>>>> + registers, which haven't already been saved by the caller.
>>>> +
>>>> + This calling convention will be used by a future version of
>>>> the ObjectiveC
>>>> + runtime and should therefore still be considered experimental
>>>> at this time.
>>>> + Although this convention was created to optimize certain
>>>> runtime calls to
>>>> + the ObjectiveC runtime, it is not limited to this runtime and
>>>> might be used
>>>> + by other runtimes in the future too. The current
>>>> implementation only
>>>> + supports X86-64, but the intention is to support more
>>>> architectures in the
>>>> + future.
>>>> +"``preserve_allcc``" - The `PreserveAll` calling convention
>>>> + This calling convention attempts to make the code in the
>>>> caller even less
>>>> + intrusive than the `PreserveMost` calling convention. This
>>>> calling
>>>> + convention also behaves identical to the `C` calling
>>>> convention on how
>>>> + arguments and return values are passed, but it uses a
>>>> different set of
>>>> + caller/callee-saved registers. This removes the burden of
>>>> saving and
>>>> + recovering a large register set before and after the call in
>>>> the caller.
>>>> +
>>>> + - On X86-64 the callee preserves all general purpose
>>>> registers, except for
>>>> + R11. R11 can be used as a scratch register. Furthermore it
>>>> also preserves
>>>> + all floating-point registers (XMMs/YMMs).
>>>> +
>>>> + The idea behind this convention is to support calls to
>>>> runtime functions
>>>> + that don't need to call out to any other functions.
>>>> +
>>>> + This calling convention, like the `PreserveMost` calling
>>>> convention, will be
>>>> + used by a future version of the ObjectiveC runtime and should
>>>> be considered
>>>> + experimental at this time.
>>>> "``cc <n>``" - Numbered convention
>>>> Any calling convention may be specified by number, allowing
>>>> target-specific calling conventions to be used. Target specific
>>>>
>>>> Modified: llvm/trunk/include/llvm/IR/CallingConv.h
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/IR/CallingConv.h?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/include/llvm/IR/CallingConv.h (original)
>>>> +++ llvm/trunk/include/llvm/IR/CallingConv.h Fri Jan 17 13:47:03
>>>> 2014
>>>> @@ -58,6 +58,14 @@ namespace CallingConv {
>>>> // stackmap and patchpoint intrinsics).
>>>> AnyReg = 13,
>>>>
>>>> + // PreserveMost - Calling convention for runtime calls that
>>>> preserves most
>>>> + // registers.
>>>> + PreserveMost = 14,
>>>> +
>>>> + // PreserveAll - Calling convention for runtime calls that
>>>> preserves
>>>> + // (almost) all registers.
>>>> + PreserveAll = 15,
>>>> +
>>>> // Target - This is the start of the target-specific calling
>>>> conventions,
>>>> // e.g. fastcall and thiscall on X86.
>>>> FirstTargetCC = 64,
>>>>
>>>> Modified: llvm/trunk/lib/AsmParser/LLLexer.cpp
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/AsmParser/LLLexer.cpp?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/lib/AsmParser/LLLexer.cpp (original)
>>>> +++ llvm/trunk/lib/AsmParser/LLLexer.cpp Fri Jan 17 13:47:03 2014
>>>> @@ -563,6 +563,8 @@ lltok::Kind LLLexer::LexIdentifier() {
>>>> KEYWORD(x86_64_win64cc);
>>>> KEYWORD(webkit_jscc);
>>>> KEYWORD(anyregcc);
>>>> + KEYWORD(preserve_mostcc);
>>>> + KEYWORD(preserve_allcc);
>>>>
>>>> KEYWORD(cc);
>>>> KEYWORD(c);
>>>>
>>>> Modified: llvm/trunk/lib/AsmParser/LLParser.cpp
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/AsmParser/LLParser.cpp?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/lib/AsmParser/LLParser.cpp (original)
>>>> +++ llvm/trunk/lib/AsmParser/LLParser.cpp Fri Jan 17 13:47:03 2014
>>>> @@ -1369,6 +1369,8 @@ bool LLParser::ParseOptionalDLLStorageCl
>>>> /// ::= 'x86_64_win64cc'
>>>> /// ::= 'webkit_jscc'
>>>> /// ::= 'anyregcc'
>>>> +/// ::= 'preserve_mostcc'
>>>> +/// ::= 'preserve_allcc'
>>>> /// ::= 'cc' UINT
>>>> ///
>>>> bool LLParser::ParseOptionalCallingConv(CallingConv::ID &CC) {
>>>> @@ -1393,6 +1395,8 @@ bool LLParser::ParseOptionalCallingConv(
>>>> case lltok::kw_x86_64_win64cc: CC = CallingConv::X86_64_Win64;
>>>> break;
>>>> case lltok::kw_webkit_jscc: CC = CallingConv::WebKit_JS;
>>>> break;
>>>> case lltok::kw_anyregcc: CC = CallingConv::AnyReg; break;
>>>> + case lltok::kw_preserve_mostcc:CC = CallingConv::PreserveMost;
>>>> break;
>>>> + case lltok::kw_preserve_allcc: CC = CallingConv::PreserveAll;
>>>> break;
>>>> case lltok::kw_cc: {
>>>> unsigned ArbitraryCC;
>>>> Lex.Lex();
>>>>
>>>> Modified: llvm/trunk/lib/AsmParser/LLToken.h
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/AsmParser/LLToken.h?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/lib/AsmParser/LLToken.h (original)
>>>> +++ llvm/trunk/lib/AsmParser/LLToken.h Fri Jan 17 13:47:03 2014
>>>> @@ -92,6 +92,7 @@ namespace lltok {
>>>> kw_spir_kernel, kw_spir_func,
>>>> kw_x86_64_sysvcc, kw_x86_64_win64cc,
>>>> kw_webkit_jscc, kw_anyregcc,
>>>> + kw_preserve_mostcc, kw_preserve_allcc,
>>>>
>>>> // Attributes:
>>>> kw_attributes,
>>>>
>>>> Modified: llvm/trunk/lib/IR/AsmWriter.cpp
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/IR/AsmWriter.cpp?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/lib/IR/AsmWriter.cpp (original)
>>>> +++ llvm/trunk/lib/IR/AsmWriter.cpp Fri Jan 17 13:47:03 2014
>>>> @@ -73,6 +73,8 @@ static void PrintCallingConv(unsigned cc
>>>> case CallingConv::Cold: Out << "coldcc"; break;
>>>> case CallingConv::WebKit_JS: Out << "webkit_jscc"; break;
>>>> case CallingConv::AnyReg: Out << "anyregcc"; break;
>>>> + case CallingConv::PreserveMost: Out << "preserve_mostcc";
>>>> break;
>>>> + case CallingConv::PreserveAll: Out << "preserve_allcc";
>>>> break;
>>>> case CallingConv::X86_StdCall: Out << "x86_stdcallcc"; break;
>>>> case CallingConv::X86_FastCall: Out << "x86_fastcallcc"; break;
>>>> case CallingConv::X86_ThisCall: Out << "x86_thiscallcc"; break;
>>>>
>>>> Modified: llvm/trunk/lib/Target/X86/X86CallingConv.td
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86CallingConv.td?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/lib/Target/X86/X86CallingConv.td (original)
>>>> +++ llvm/trunk/lib/Target/X86/X86CallingConv.td Fri Jan 17
>>>> 13:47:03 2014
>>>> @@ -620,6 +620,16 @@ def CSR_64EHRet : CalleeSavedRegs<(add R
>>>> def CSR_Win64 : CalleeSavedRegs<(add RBX, RBP, RDI, RSI, R12, R13,
>>>> R14, R15,
>>>> (sequence "XMM%u", 6, 15))>;
>>>>
>>>> +// All GPRs - except r11
>>>> +def CSR_64_RT_MostRegs : CalleeSavedRegs<(add CSR_64, RAX, RCX,
>>>> RDX, RSI, RDI,
>>>> + R8, R9, R10, RSP)>;
>>>> +
>>>> +// All registers - except r11
>>>> +def CSR_64_RT_AllRegs : CalleeSavedRegs<(add
>>>> CSR_64_RT_MostRegs,
>>>> + (sequence
>>>> "XMM%u", 0, 15))>;
>>>> +def CSR_64_RT_AllRegs_AVX : CalleeSavedRegs<(add
>>>> CSR_64_RT_MostRegs,
>>>> + (sequence
>>>> "YMM%u", 0, 15))>;
>>>> +
>>>> def CSR_64_MostRegs : CalleeSavedRegs<(add RBX, RCX, RDX, RSI,
>>>> RDI, R8, R9, R10,
>>>> R11, R12, R13, R14, R15,
>>>> RBP,
>>>> (sequence "XMM%u", 0,
>>>> 15))>;
>>>>
>>>> Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp?rev=199508&r1=199507&r2=199508&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp (original)
>>>> +++ llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Fri Jan 17
>>>> 13:47:03 2014
>>>> @@ -245,6 +245,12 @@ X86RegisterInfo::getCalleeSavedRegs(cons
>>>> if (HasAVX)
>>>> return CSR_64_AllRegs_AVX_SaveList;
>>>> return CSR_64_AllRegs_SaveList;
>>>> + case CallingConv::PreserveMost:
>>>> + return CSR_64_RT_MostRegs_SaveList;
>>>> + case CallingConv::PreserveAll:
>>>> + if (HasAVX)
>>>> + return CSR_64_RT_AllRegs_AVX_SaveList;
>>>> + return CSR_64_RT_AllRegs_SaveList;
>>>> case CallingConv::Intel_OCL_BI: {
>>>> if (HasAVX512 && IsWin64)
>>>> return CSR_Win64_Intel_OCL_BI_AVX512_SaveList;
>>>> @@ -292,6 +298,12 @@ X86RegisterInfo::getCallPreservedMask(Ca
>>>> if (HasAVX)
>>>> return CSR_64_AllRegs_AVX_RegMask;
>>>> return CSR_64_AllRegs_RegMask;
>>>> + case CallingConv::PreserveMost:
>>>> + return CSR_64_RT_MostRegs_RegMask;
>>>> + case CallingConv::PreserveAll:
>>>> + if (HasAVX)
>>>> + return CSR_64_RT_AllRegs_AVX_RegMask;
>>>> + return CSR_64_RT_AllRegs_RegMask;
>>>> case CallingConv::Intel_OCL_BI: {
>>>> if (IsWin64 && HasAVX512)
>>>> return CSR_Win64_Intel_OCL_BI_AVX512_RegMask;
>>>>
>>>> Added: llvm/trunk/test/CodeGen/X86/preserve_allcc64.ll
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/preserve_allcc64.ll?rev=199508&view=auto
>>>> ==============================================================================
>>>> --- llvm/trunk/test/CodeGen/X86/preserve_allcc64.ll (added)
>>>> +++ llvm/trunk/test/CodeGen/X86/preserve_allcc64.ll Fri Jan 17
>>>> 13:47:03 2014
>>>> @@ -0,0 +1,104 @@
>>>> +; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7 |
>>>> FileCheck --check-prefix=SSE %s
>>>> +; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx |
>>>> FileCheck --check-prefix=AVX %s
>>>> +
>>>> +define preserve_allcc void @preserve_allcc1() nounwind {
>>>> +entry:
>>>> +;SSE-LABEL: preserve_allcc1
>>>> +;SSE: pushq %r10
>>>> +;SSE-NEXT: pushq %r9
>>>> +;SSE-NEXT: pushq %r8
>>>> +;SSE-NEXT: pushq %rdi
>>>> +;SSE-NEXT: pushq %rsi
>>>> +;SSE-NEXT: pushq %rdx
>>>> +;SSE-NEXT: pushq %rcx
>>>> +;SSE-NEXT: pushq %rax
>>>> +;SSE-NEXT: pushq %rbp
>>>> +;SSE-NEXT: pushq %r15
>>>> +;SSE-NEXT: pushq %r14
>>>> +;SSE-NEXT: pushq %r13
>>>> +;SSE-NEXT: pushq %r12
>>>> +;SSE-NEXT: pushq %rbx
>>>> +;SSE: movaps %xmm15
>>>> +;SSE-NEXT: movaps %xmm14
>>>> +;SSE-NEXT: movaps %xmm13
>>>> +;SSE-NEXT: movaps %xmm12
>>>> +;SSE-NEXT: movaps %xmm11
>>>> +;SSE-NEXT: movaps %xmm10
>>>> +;SSE-NEXT: movaps %xmm9
>>>> +;SSE-NEXT: movaps %xmm8
>>>> +;SSE-NEXT: movaps %xmm7
>>>> +;SSE-NEXT: movaps %xmm6
>>>> +;SSE-NEXT: movaps %xmm5
>>>> +;SSE-NEXT: movaps %xmm4
>>>> +;SSE-NEXT: movaps %xmm3
>>>> +;SSE-NEXT: movaps %xmm2
>>>> +;SSE-NEXT: movaps %xmm1
>>>> +;SSE-NEXT: movaps %xmm0
>>>> +;AVX-LABEL: preserve_allcc1
>>>> +;AVX: pushq %r10
>>>> +;AVX-NEXT: pushq %r9
>>>> +;AVX-NEXT: pushq %r8
>>>> +;AVX-NEXT: pushq %rdi
>>>> +;AVX-NEXT: pushq %rsi
>>>> +;AVX-NEXT: pushq %rdx
>>>> +;AVX-NEXT: pushq %rcx
>>>> +;AVX-NEXT: pushq %rax
>>>> +;AVX-NEXT: pushq %rbp
>>>> +;AVX-NEXT: pushq %r15
>>>> +;AVX-NEXT: pushq %r14
>>>> +;AVX-NEXT: pushq %r13
>>>> +;AVX-NEXT: pushq %r12
>>>> +;AVX-NEXT: pushq %rbx
>>>> +;AVX: vmovups %ymm15
>>>> +;AVX-NEXT: vmovups %ymm14
>>>> +;AVX-NEXT: vmovups %ymm13
>>>> +;AVX-NEXT: vmovups %ymm12
>>>> +;AVX-NEXT: vmovups %ymm11
>>>> +;AVX-NEXT: vmovups %ymm10
>>>> +;AVX-NEXT: vmovups %ymm9
>>>> +;AVX-NEXT: vmovups %ymm8
>>>> +;AVX-NEXT: vmovups %ymm7
>>>> +;AVX-NEXT: vmovups %ymm6
>>>> +;AVX-NEXT: vmovups %ymm5
>>>> +;AVX-NEXT: vmovups %ymm4
>>>> +;AVX-NEXT: vmovups %ymm3
>>>> +;AVX-NEXT: vmovups %ymm2
>>>> +;AVX-NEXT: vmovups %ymm1
>>>> +;AVX-NEXT: vmovups %ymm0
>>>> + call void asm sideeffect "",
>>>> "~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15},~{rbp},~{xmm0},~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15}"()
>>>> + ret void
>>>> +}
>>>> +
>>>> +; Make sure only R11 is saved before the call
>>>> +declare preserve_allcc void @bar(i64, i64, double, double)
>>>> +define void @preserve_allcc2() nounwind {
>>>> +entry:
>>>> +;SSE-LABEL: preserve_allcc2
>>>> +;SSE: movq %r11, [[REG:%[a-z0-9]+]]
>>>> +;SSE-NOT: movaps %xmm
>>>> +;SSE: movq [[REG]], %r11
>>>> + %a0 = call i64 asm sideeffect "", "={rax}"() nounwind
>>>> + %a1 = call i64 asm sideeffect "", "={rcx}"() nounwind
>>>> + %a2 = call i64 asm sideeffect "", "={rdx}"() nounwind
>>>> + %a3 = call i64 asm sideeffect "", "={r8}"() nounwind
>>>> + %a4 = call i64 asm sideeffect "", "={r9}"() nounwind
>>>> + %a5 = call i64 asm sideeffect "", "={r10}"() nounwind
>>>> + %a6 = call i64 asm sideeffect "", "={r11}"() nounwind
>>>> + %a10 = call <2 x double> asm sideeffect "", "={xmm2}"()
>>>> nounwind
>>>> + %a11 = call <2 x double> asm sideeffect "", "={xmm3}"()
>>>> nounwind
>>>> + %a12 = call <2 x double> asm sideeffect "", "={xmm4}"()
>>>> nounwind
>>>> + %a13 = call <2 x double> asm sideeffect "", "={xmm5}"()
>>>> nounwind
>>>> + %a14 = call <2 x double> asm sideeffect "", "={xmm6}"()
>>>> nounwind
>>>> + %a15 = call <2 x double> asm sideeffect "", "={xmm7}"()
>>>> nounwind
>>>> + %a16 = call <2 x double> asm sideeffect "", "={xmm8}"()
>>>> nounwind
>>>> + %a17 = call <2 x double> asm sideeffect "", "={xmm9}"()
>>>> nounwind
>>>> + %a18 = call <2 x double> asm sideeffect "", "={xmm10}"()
>>>> nounwind
>>>> + %a19 = call <2 x double> asm sideeffect "", "={xmm11}"()
>>>> nounwind
>>>> + %a20 = call <2 x double> asm sideeffect "", "={xmm12}"()
>>>> nounwind
>>>> + %a21 = call <2 x double> asm sideeffect "", "={xmm13}"()
>>>> nounwind
>>>> + %a22 = call <2 x double> asm sideeffect "", "={xmm14}"()
>>>> nounwind
>>>> + %a23 = call <2 x double> asm sideeffect "", "={xmm15}"()
>>>> nounwind
>>>> + call preserve_allcc void @bar(i64 1, i64 2, double 3.0, double
>>>> 4.0)
>>>> + call void asm sideeffect "",
>>>> "{rax},{rcx},{rdx},{r8},{r9},{r10},{r11},{xmm2},{xmm3},{xmm4},{xmm5},{xmm6},{xmm7},{xmm8},{xmm9},{xmm10},{xmm11},{xmm12},{xmm13},{xmm14},{xmm15}"(i64
>>>> %a0, i64 %a1, i64 %a2, i64 %a3, i64 %a4, i64 %a5, i64 %a6, <2 x
>>>> double> %a10, <2 x double> %a11, <2 x double> %a12, <2 x double>
>>>> %a13, <2 x double> %a14, <2 x double> %a15, <2 x double> %a16, <2
>>>> x double> %a17, <2 x double> %a18, <2 x double> %a19, <2 x
>>>> double> %a20, <2 x double> %a21, <2 x double> %a22, <2 x double>
>>>> %a23)
>>>> + ret void
>>>> +}
>>>>
>>>> Added: llvm/trunk/test/CodeGen/X86/preserve_mostcc64.ll
>>>> URL:
>>>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/preserve_mostcc64.ll?rev=199508&view=auto
>>>> ==============================================================================
>>>> --- llvm/trunk/test/CodeGen/X86/preserve_mostcc64.ll (added)
>>>> +++ llvm/trunk/test/CodeGen/X86/preserve_mostcc64.ll Fri Jan 17
>>>> 13:47:03 2014
>>>> @@ -0,0 +1,86 @@
>>>> +; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7 |
>>>> FileCheck --check-prefix=SSE %s
>>>> +; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx |
>>>> FileCheck --check-prefix=AVX %s
>>>> +
>>>> +; Every GPR should be saved - except r11
>>>> +define preserve_mostcc void @preserve_mostcc1() nounwind {
>>>> +entry:
>>>> +;SSE-LABEL: preserve_mostcc1
>>>> +;SSE: pushq %r10
>>>> +;SSE-NEXT: pushq %r9
>>>> +;SSE-NEXT: pushq %r8
>>>> +;SSE-NEXT: pushq %rdi
>>>> +;SSE-NEXT: pushq %rsi
>>>> +;SSE-NEXT: pushq %rdx
>>>> +;SSE-NEXT: pushq %rcx
>>>> +;SSE-NEXT: pushq %rax
>>>> +;SSE-NEXT: pushq %rbp
>>>> +;SSE-NEXT: pushq %r15
>>>> +;SSE-NEXT: pushq %r14
>>>> +;SSE-NEXT: pushq %r13
>>>> +;SSE-NEXT: pushq %r12
>>>> +;SSE-NEXT: pushq %rbx
>>>> +;AVX-LABEL: preserve_mostcc1
>>>> +;AVX: pushq %r10
>>>> +;AVX-NEXT: pushq %r9
>>>> +;AVX-NEXT: pushq %r8
>>>> +;AVX-NEXT: pushq %rdi
>>>> +;AVX-NEXT: pushq %rsi
>>>> +;AVX-NEXT: pushq %rdx
>>>> +;AVX-NEXT: pushq %rcx
>>>> +;AVX-NEXT: pushq %rax
>>>> +;AVX-NEXT: pushq %rbp
>>>> +;AVX-NEXT: pushq %r15
>>>> +;AVX-NEXT: pushq %r14
>>>> +;AVX-NEXT: pushq %r13
>>>> +;AVX-NEXT: pushq %r12
>>>> +;AVX-NEXT: pushq %rbx
>>>> + call void asm sideeffect "",
>>>> "~{rax},~{rbx},~{rcx},~{rdx},~{rsi},~{rdi},~{r8},~{r9},~{r10},~{r11},~{r12},~{r13},~{r14},~{r15},~{rbp},~{xmm0},~{xmm1},~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15}"()
>>>> + ret void
>>>> +}
>>>> +
>>>> +; Make sure R11 and XMMs are saved before the call
>>>> +declare preserve_mostcc void @foo(i64, i64, double, double)
>>>> +define void @preserve_mostcc2() nounwind {
>>>> +entry:
>>>> +;SSE-LABEL: preserve_mostcc2
>>>> +;SSE: movq %r11, [[REG:%[a-z0-9]+]]
>>>> +;SSE: movaps %xmm2
>>>> +;SSE: movaps %xmm3
>>>> +;SSE: movaps %xmm4
>>>> +;SSE: movaps %xmm5
>>>> +;SSE: movaps %xmm6
>>>> +;SSE: movaps %xmm7
>>>> +;SSE: movaps %xmm8
>>>> +;SSE: movaps %xmm9
>>>> +;SSE: movaps %xmm10
>>>> +;SSE: movaps %xmm11
>>>> +;SSE: movaps %xmm12
>>>> +;SSE: movaps %xmm13
>>>> +;SSE: movaps %xmm14
>>>> +;SSE: movaps %xmm15
>>>> +;SSE: movq [[REG]], %r11
>>>> + %a0 = call i64 asm sideeffect "", "={rax}"() nounwind
>>>> + %a1 = call i64 asm sideeffect "", "={rcx}"() nounwind
>>>> + %a2 = call i64 asm sideeffect "", "={rdx}"() nounwind
>>>> + %a3 = call i64 asm sideeffect "", "={r8}"() nounwind
>>>> + %a4 = call i64 asm sideeffect "", "={r9}"() nounwind
>>>> + %a5 = call i64 asm sideeffect "", "={r10}"() nounwind
>>>> + %a6 = call i64 asm sideeffect "", "={r11}"() nounwind
>>>> + %a10 = call <2 x double> asm sideeffect "", "={xmm2}"()
>>>> nounwind
>>>> + %a11 = call <2 x double> asm sideeffect "", "={xmm3}"()
>>>> nounwind
>>>> + %a12 = call <2 x double> asm sideeffect "", "={xmm4}"()
>>>> nounwind
>>>> + %a13 = call <2 x double> asm sideeffect "", "={xmm5}"()
>>>> nounwind
>>>> + %a14 = call <2 x double> asm sideeffect "", "={xmm6}"()
>>>> nounwind
>>>> + %a15 = call <2 x double> asm sideeffect "", "={xmm7}"()
>>>> nounwind
>>>> + %a16 = call <2 x double> asm sideeffect "", "={xmm8}"()
>>>> nounwind
>>>> + %a17 = call <2 x double> asm sideeffect "", "={xmm9}"()
>>>> nounwind
>>>> + %a18 = call <2 x double> asm sideeffect "", "={xmm10}"()
>>>> nounwind
>>>> + %a19 = call <2 x double> asm sideeffect "", "={xmm11}"()
>>>> nounwind
>>>> + %a20 = call <2 x double> asm sideeffect "", "={xmm12}"()
>>>> nounwind
>>>> + %a21 = call <2 x double> asm sideeffect "", "={xmm13}"()
>>>> nounwind
>>>> + %a22 = call <2 x double> asm sideeffect "", "={xmm14}"()
>>>> nounwind
>>>> + %a23 = call <2 x double> asm sideeffect "", "={xmm15}"()
>>>> nounwind
>>>> + call preserve_mostcc void @foo(i64 1, i64 2, double 3.0, double
>>>> 4.0)
>>>> + call void asm sideeffect "",
>>>> "{rax},{rcx},{rdx},{r8},{r9},{r10},{r11},{xmm2},{xmm3},{xmm4},{xmm5},{xmm6},{xmm7},{xmm8},{xmm9},{xmm10},{xmm11},{xmm12},{xmm13},{xmm14},{xmm15}"(i64
>>>> %a0, i64 %a1, i64 %a2, i64 %a3, i64 %a4, i64 %a5, i64 %a6, <2 x
>>>> double> %a10, <2 x double> %a11, <2 x double> %a12, <2 x double>
>>>> %a13, <2 x double> %a14, <2 x double> %a15, <2 x double> %a16, <2
>>>> x double> %a17, <2 x double> %a18, <2 x double> %a19, <2 x
>>>> double> %a20, <2 x double> %a21, <2 x double> %a22, <2 x double>
>>>> %a23)
>>>> + ret void
>>>> +}
>>>>
>>>>
>>>> _______________________________________________
>>>> llvm-commits mailing list
>>>> llvm-commits at cs.uiuc.edu
>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>
> --
> Hal Finkel
> Assistant Computational Scientist
> Leadership Computing Facility
> Argonne National Laboratory
More information about the llvm-commits
mailing list