aboutsummaryrefslogtreecommitdiffstats
path: root/lib/Target/X86/X86Subtarget.cpp
diff options
context:
space:
mode:
authorEvan Cheng <evan.cheng@apple.com>2011-07-07 21:06:52 +0000
committerEvan Cheng <evan.cheng@apple.com>2011-07-07 21:06:52 +0000
commit18fb1d35db9e2160be3a5bd2950f7e0d206bdbb8 (patch)
tree1390922ecf4949a4bed04f2639da2c8aaab1862e /lib/Target/X86/X86Subtarget.cpp
parent4ae970b393f9518eb4a0c2e6b2c16ec3bee570d0 (diff)
downloadexternal_llvm-18fb1d35db9e2160be3a5bd2950f7e0d206bdbb8.zip
external_llvm-18fb1d35db9e2160be3a5bd2950f7e0d206bdbb8.tar.gz
external_llvm-18fb1d35db9e2160be3a5bd2950f7e0d206bdbb8.tar.bz2
Add Mode64Bit feature and sink it down to MC layer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@134641 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Target/X86/X86Subtarget.cpp')
-rw-r--r--lib/Target/X86/X86Subtarget.cpp115
1 files changed, 32 insertions, 83 deletions
diff --git a/lib/Target/X86/X86Subtarget.cpp b/lib/Target/X86/X86Subtarget.cpp
index 1c1a10d..cba0484 100644
--- a/lib/Target/X86/X86Subtarget.cpp
+++ b/lib/Target/X86/X86Subtarget.cpp
@@ -158,7 +158,7 @@ const char *X86Subtarget::getBZeroEntry() const {
/// IsLegalToCallImmediateAddr - Return true if the subtarget allows calls
/// to immediate address.
bool X86Subtarget::IsLegalToCallImmediateAddr(const TargetMachine &TM) const {
- if (Is64Bit)
+ if (In64BitMode)
return false;
return isTargetELF() || TM.getRelocationModel() == Reloc::Static;
}
@@ -174,73 +174,6 @@ unsigned X86Subtarget::getSpecialAddressLatency() const {
return 200;
}
-/// GetCpuIDAndInfo - Execute the specified cpuid and return the 4 values in the
-/// specified arguments. If we can't run cpuid on the host, return true.
-static bool GetCpuIDAndInfo(unsigned value, unsigned *rEAX,
- unsigned *rEBX, unsigned *rECX, unsigned *rEDX) {
-#if defined(__x86_64__) || defined(_M_AMD64) || defined (_M_X64)
- #if defined(__GNUC__)
- // gcc doesn't know cpuid would clobber ebx/rbx. Preseve it manually.
- asm ("movq\t%%rbx, %%rsi\n\t"
- "cpuid\n\t"
- "xchgq\t%%rbx, %%rsi\n\t"
- : "=a" (*rEAX),
- "=S" (*rEBX),
- "=c" (*rECX),
- "=d" (*rEDX)
- : "a" (value));
- return false;
- #elif defined(_MSC_VER)
- int registers[4];
- __cpuid(registers, value);
- *rEAX = registers[0];
- *rEBX = registers[1];
- *rECX = registers[2];
- *rEDX = registers[3];
- return false;
- #endif
-#elif defined(i386) || defined(__i386__) || defined(__x86__) || defined(_M_IX86)
- #if defined(__GNUC__)
- asm ("movl\t%%ebx, %%esi\n\t"
- "cpuid\n\t"
- "xchgl\t%%ebx, %%esi\n\t"
- : "=a" (*rEAX),
- "=S" (*rEBX),
- "=c" (*rECX),
- "=d" (*rEDX)
- : "a" (value));
- return false;
- #elif defined(_MSC_VER)
- __asm {
- mov eax,value
- cpuid
- mov esi,rEAX
- mov dword ptr [esi],eax
- mov esi,rEBX
- mov dword ptr [esi],ebx
- mov esi,rECX
- mov dword ptr [esi],ecx
- mov esi,rEDX
- mov dword ptr [esi],edx
- }
- return false;
- #endif
-#endif
- return true;
-}
-
-static void DetectFamilyModel(unsigned EAX, unsigned &Family, unsigned &Model) {
- Family = (EAX >> 8) & 0xf; // Bits 8 - 11
- Model = (EAX >> 4) & 0xf; // Bits 4 - 7
- if (Family == 6 || Family == 0xf) {
- if (Family == 0xf)
- // Examine extended family ID if family ID is F.
- Family += (EAX >> 20) & 0xff; // Bits 20 - 27
- // Examine extended model ID if family ID is 6 or F.
- Model += ((EAX >> 16) & 0xf) << 4; // Bits 16 - 19
- }
-}
-
void X86Subtarget::AutoDetectSubtargetFeatures() {
unsigned EAX = 0, EBX = 0, ECX = 0, EDX = 0;
union {
@@ -248,10 +181,10 @@ void X86Subtarget::AutoDetectSubtargetFeatures() {
char c[12];
} text;
- if (GetCpuIDAndInfo(0, &EAX, text.u+0, text.u+2, text.u+1))
+ if (X86_MC::GetCpuIDAndInfo(0, &EAX, text.u+0, text.u+2, text.u+1))
return;
- GetCpuIDAndInfo(0x1, &EAX, &EBX, &ECX, &EDX);
+ X86_MC::GetCpuIDAndInfo(0x1, &EAX, &EBX, &ECX, &EDX);
if ((EDX >> 15) & 1) HasCMov = true;
if ((EDX >> 23) & 1) X86SSELevel = MMX;
@@ -276,13 +209,13 @@ void X86Subtarget::AutoDetectSubtargetFeatures() {
// Determine if bit test memory instructions are slow.
unsigned Family = 0;
unsigned Model = 0;
- DetectFamilyModel(EAX, Family, Model);
+ X86_MC::DetectFamilyModel(EAX, Family, Model);
IsBTMemSlow = IsAMD || (Family == 6 && Model >= 13);
// If it's Nehalem, unaligned memory access is fast.
if (Family == 15 && Model == 26)
IsUAMemFast = true;
- GetCpuIDAndInfo(0x80000001, &EAX, &EBX, &ECX, &EDX);
+ X86_MC::GetCpuIDAndInfo(0x80000001, &EAX, &EBX, &ECX, &EDX);
HasX86_64 = (EDX >> 29) & 0x1;
HasSSE4A = IsAMD && ((ECX >> 6) & 0x1);
HasFMA4 = IsAMD && ((ECX >> 16) & 0x1);
@@ -291,7 +224,7 @@ void X86Subtarget::AutoDetectSubtargetFeatures() {
X86Subtarget::X86Subtarget(const std::string &TT, const std::string &CPU,
const std::string &FS,
- bool is64Bit, unsigned StackAlignOverride)
+ unsigned StackAlignOverride)
: X86GenSubtargetInfo(TT, CPU, FS)
, PICStyle(PICStyles::None)
, X86SSELevel(NoMMXSSE)
@@ -312,15 +245,26 @@ X86Subtarget::X86Subtarget(const std::string &TT, const std::string &CPU,
// FIXME: this is a known good value for Yonah. How about others?
, MaxInlineSizeThreshold(128)
, TargetTriple(TT)
- , Is64Bit(is64Bit) {
+ , In64BitMode(false) {
+ // Insert the architecture feature derived from the target triple into the
+ // feature string. This is important for setting features that are implied
+ // based on the architecture version.
+ std::string ArchFS = X86_MC::ParseX86Triple(TT);
+ if (!FS.empty()) {
+ if (!ArchFS.empty())
+ ArchFS = ArchFS + "," + FS;
+ else
+ ArchFS = FS;
+ }
+
+ std::string CPUName = CPU;
+ if (CPUName.empty())
+ CPUName = sys::getHostCPUName();
// Determine default and user specified characteristics
- if (!CPU.empty() || !FS.empty()) {
+ if (!CPUName.empty() || !ArchFS.empty()) {
// If feature string is not empty, parse features string.
- std::string CPUName = CPU;
- if (CPUName.empty())
- CPUName = sys::getHostCPUName();
- ParseSubtargetFeatures(CPUName, FS);
+ ParseSubtargetFeatures(CPUName, ArchFS);
// All X86-64 CPUs also have SSE2, however user might request no SSE via
// -mattr, so don't force SSELevel here.
if (HasAVX)
@@ -328,14 +272,19 @@ X86Subtarget::X86Subtarget(const std::string &TT, const std::string &CPU,
} else {
// Otherwise, use CPUID to auto-detect feature set.
AutoDetectSubtargetFeatures();
+
+ // If CPU is 64-bit capable, default to 64-bit mode if not specified.
+ In64BitMode = HasX86_64;
+
// Make sure SSE2 is enabled; it is available on all X86-64 CPUs.
- if (Is64Bit && !HasAVX && X86SSELevel < SSE2)
+ if (In64BitMode && !HasAVX && X86SSELevel < SSE2)
X86SSELevel = SSE2;
}
// If requesting codegen for X86-64, make sure that 64-bit features
// are enabled.
- if (Is64Bit) {
+ // FIXME: Remove this feature since it's not actually being used.
+ if (In64BitMode) {
HasX86_64 = true;
// All 64-bit cpus have cmov support.
@@ -345,7 +294,7 @@ X86Subtarget::X86Subtarget(const std::string &TT, const std::string &CPU,
DEBUG(dbgs() << "Subtarget features: SSELevel " << X86SSELevel
<< ", 3DNowLevel " << X863DNowLevel
<< ", 64bit " << HasX86_64 << "\n");
- assert((!Is64Bit || HasX86_64) &&
+ assert((!In64BitMode || HasX86_64) &&
"64-bit code requested on a subtarget that doesn't support it!");
// Stack alignment is 16 bytes on Darwin, FreeBSD, Linux and Solaris (both
@@ -353,6 +302,6 @@ X86Subtarget::X86Subtarget(const std::string &TT, const std::string &CPU,
if (StackAlignOverride)
stackAlignment = StackAlignOverride;
else if (isTargetDarwin() || isTargetFreeBSD() || isTargetLinux() ||
- isTargetSolaris() || Is64Bit)
+ isTargetSolaris() || In64BitMode)
stackAlignment = 16;
}