[llvm-commits] CVS: llvm/include/llvm/BasicBlock.h
Changes in directory llvm/include/llvm: BasicBlock.h updated: 1.58 -> 1.59 --- Log message: New method BasicBlock::getFirstNonPHI. --- Diffs of the changes: (+8 -1) BasicBlock.h |9 - 1 files changed, 8 insertions(+), 1 deletion(-) Index: llvm/include/llvm/BasicBlock.h diff -u llvm/include/llvm/BasicBlock.h:1.58 llvm/include/llvm/BasicBlock.h:1.59 --- llvm/include/llvm/BasicBlock.h:1.58 Mon Jun 5 11:29:06 2006 +++ llvm/include/llvm/BasicBlock.h Thu Jun 8 10:46:17 2006 @@ -91,7 +91,14 @@ /// TerminatorInst *getTerminator(); const TerminatorInst *const getTerminator() const; - + + /// Returns a pointer to the first instructon in this block that is not a + /// PHINode instruction. When adding instruction to the beginning of the + /// basic block, they should be added before the returned value, not before + /// the first instruction, which might be PHI. + /// Returns 0 is there's no non-PHI instruction. + Instruction* getFirstNonPHI(); + /// removeFromParent - This method unlinks 'this' from the containing /// function, but does not delete it. /// ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/VMCore/BasicBlock.cpp
Changes in directory llvm/lib/VMCore:
BasicBlock.cpp updated: 1.68 -> 1.69
---
Log message:
New method BasicBlock::getFirstNonPHI.
---
Diffs of the changes: (+11 -0)
BasicBlock.cpp | 11 +++
1 files changed, 11 insertions(+)
Index: llvm/lib/VMCore/BasicBlock.cpp
diff -u llvm/lib/VMCore/BasicBlock.cpp:1.68 llvm/lib/VMCore/BasicBlock.cpp:1.69
--- llvm/lib/VMCore/BasicBlock.cpp:1.68 Fri Aug 12 17:14:06 2005
+++ llvm/lib/VMCore/BasicBlock.cpp Thu Jun 8 10:46:18 2006
@@ -121,6 +121,17 @@
return dyn_cast(&InstList.back());
}
+Instruction* BasicBlock::getFirstNonPHI()
+{
+BasicBlock::iterator i = begin(), e = end();
+// All valid basic blocks should have a terminator,
+// which is not a PHINode. If we have invalid basic
+// block we'll get assert when dereferencing past-the-end
+// iterator.
+while (isa(i)) ++i;
+return &*i;
+}
+
void BasicBlock::dropAllReferences() {
for(iterator I = begin(), E = end(); I != E; ++I)
I->dropAllReferences();
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Re: [llvm-commits] CVS: llvm/lib/VMCore/BasicBlock.cpp
Vladimir,The e = end() is no longer needed.Cheers,-- JimOn Jun 8, 2006, at 12:46 PM, Vladimir Prus wrote:+Instruction* BasicBlock::getFirstNonPHI() +{ + BasicBlock::iterator i = begin(), e = end(); + // All valid basic blocks should have a terminator, + // which is not a PHINode. If we have invalid basic + // block we'll get assert when dereferencing past-the-end + // iterator. + while (isa(i)) ++i; + return &*i; +} +
smime.p7s
Description: S/MIME cryptographic signature
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/VMCore/BasicBlock.cpp
Changes in directory llvm/lib/VMCore:
BasicBlock.cpp updated: 1.69 -> 1.70
---
Log message:
Remove dead code.
---
Diffs of the changes: (+1 -1)
BasicBlock.cpp |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm/lib/VMCore/BasicBlock.cpp
diff -u llvm/lib/VMCore/BasicBlock.cpp:1.69 llvm/lib/VMCore/BasicBlock.cpp:1.70
--- llvm/lib/VMCore/BasicBlock.cpp:1.69 Thu Jun 8 10:46:18 2006
+++ llvm/lib/VMCore/BasicBlock.cpp Thu Jun 8 11:03:13 2006
@@ -123,7 +123,7 @@
Instruction* BasicBlock::getFirstNonPHI()
{
-BasicBlock::iterator i = begin(), e = end();
+BasicBlock::iterator i = begin();
// All valid basic blocks should have a terminator,
// which is not a PHINode. If we have invalid basic
// block we'll get assert when dereferencing past-the-end
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/System/Win32/Path.inc
Changes in directory llvm/lib/System/Win32:
Path.inc updated: 1.43 -> 1.44
---
Log message:
For PR804: http://llvm.org/PR804 :
Change the file size field of StatusInfo to be uint64_t instead of size_t
so that we know it is always 64 bits. This prevents some overflow on
systems where size_t is 32 bits when it ought to be 64.
---
Diffs of the changes: (+1 -1)
Path.inc |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm/lib/System/Win32/Path.inc
diff -u llvm/lib/System/Win32/Path.inc:1.43 llvm/lib/System/Win32/Path.inc:1.44
--- llvm/lib/System/Win32/Path.inc:1.43 Mon Jun 5 10:44:46 2006
+++ llvm/lib/System/Win32/Path.inc Thu Jun 8 12:00:08 2006
@@ -340,7 +340,7 @@
ThrowError("getStatusInfo():" + std::string(path) + ": Can't get status:
");
info.fileSize = fi.nFileSizeHigh;
- info.fileSize <<= 32;
+ info.fileSize <<= sizeof(fi.nFileSizeHigh)*8;
info.fileSize += fi.nFileSizeLow;
info.mode = fi.dwFileAttributes & FILE_ATTRIBUTE_READONLY ? 0555 : 0777;
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/include/llvm/System/Path.h
Changes in directory llvm/include/llvm/System:
Path.h updated: 1.28 -> 1.29
---
Log message:
For PR804: http://llvm.org/PR804 :
Change the file size field of StatusInfo to be uint64_t instead of size_t
so that we know it is always 64 bits. This prevents some overflow on
systems where size_t is 32 bits when it ought to be 64.
---
Diffs of the changes: (+1 -1)
Path.h |2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
Index: llvm/include/llvm/System/Path.h
diff -u llvm/include/llvm/System/Path.h:1.28
llvm/include/llvm/System/Path.h:1.29
--- llvm/include/llvm/System/Path.h:1.28Wed Feb 22 10:23:43 2006
+++ llvm/include/llvm/System/Path.h Thu Jun 8 12:00:08 2006
@@ -67,7 +67,7 @@
struct StatusInfo {
StatusInfo() : fileSize(0), modTime(0,0), mode(0777), user(999),
group(999), isDir(false) { }
-size_t fileSize; ///< Size of the file in bytes
+uint64_tfileSize; ///< Size of the file in bytes
TimeValue modTime;///< Time of file's modification
uint32_tmode; ///< Mode of the file, if applicable
uint32_tuser; ///< User ID of owner, if applicable
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/CodeGen/AsmPrinter.cpp
Changes in directory llvm/lib/CodeGen:
AsmPrinter.cpp updated: 1.78 -> 1.79
---
Log message:
Fix an assert-on-inline-inline-asm bug.
---
Diffs of the changes: (+18 -12)
AsmPrinter.cpp | 30 ++
1 files changed, 18 insertions(+), 12 deletions(-)
Index: llvm/lib/CodeGen/AsmPrinter.cpp
diff -u llvm/lib/CodeGen/AsmPrinter.cpp:1.78
llvm/lib/CodeGen/AsmPrinter.cpp:1.79
--- llvm/lib/CodeGen/AsmPrinter.cpp:1.78Fri May 12 12:50:35 2006
+++ llvm/lib/CodeGen/AsmPrinter.cpp Thu Jun 8 13:00:47 2006
@@ -656,24 +656,30 @@
// operand!
if (CurVariant == -1 || CurVariant == AsmPrinterVariant) {
unsigned OpNo = 1;
-
+
+bool Error = false;
+
// Scan to find the machine operand number for the operand.
for (; Val; --Val) {
+ if (OpNo >= MI->getNumOperands()) break;
unsigned OpFlags = MI->getOperand(OpNo).getImmedValue();
OpNo += (OpFlags >> 3) + 1;
}
-
-unsigned OpFlags = MI->getOperand(OpNo).getImmedValue();
-++OpNo; // Skip over the ID number.
-
-bool Error;
-AsmPrinter *AP = const_cast(this);
-if ((OpFlags & 7) == 4 /*ADDR MODE*/) {
- Error = AP->PrintAsmMemoryOperand(MI, OpNo, AsmPrinterVariant,
-Modifier[0] ? Modifier : 0);
+
+if (OpNo >= MI->getNumOperands()) {
+ Error = true;
} else {
- Error = AP->PrintAsmOperand(MI, OpNo, AsmPrinterVariant,
- Modifier[0] ? Modifier : 0);
+ unsigned OpFlags = MI->getOperand(OpNo).getImmedValue();
+ ++OpNo; // Skip over the ID number.
+
+ AsmPrinter *AP = const_cast(this);
+ if ((OpFlags & 7) == 4 /*ADDR MODE*/) {
+Error = AP->PrintAsmMemoryOperand(MI, OpNo, AsmPrinterVariant,
+ Modifier[0] ? Modifier : 0);
+ } else {
+Error = AP->PrintAsmOperand(MI, OpNo, AsmPrinterVariant,
+Modifier[0] ? Modifier : 0);
+ }
}
if (Error) {
std::cerr << "Invalid operand found in inline asm: '"
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/System/Win32/Path.inc
Changes in directory llvm/lib/System/Win32: Path.inc updated: 1.44 -> 1.45 --- Log message: Squelch a warning about signed/unsigned. --- Diffs of the changes: (+1 -1) Path.inc |2 +- 1 files changed, 1 insertion(+), 1 deletion(-) Index: llvm/lib/System/Win32/Path.inc diff -u llvm/lib/System/Win32/Path.inc:1.44 llvm/lib/System/Win32/Path.inc:1.45 --- llvm/lib/System/Win32/Path.inc:1.44 Thu Jun 8 12:00:08 2006 +++ llvm/lib/System/Win32/Path.inc Thu Jun 8 13:08:43 2006 @@ -121,7 +121,7 @@ // Append a subdirectory passed on our process id so multiple LLVMs don't // step on each other's toes. - sprintf(pathname, "LLVM_%u", GetCurrentProcessId()); + sprintf(pathname, "LLVM_%u", unsigned(GetCurrentProcessId())); result.appendComponent(pathname); // If there's a directory left over from a previous LLVM execution that ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
Changes in directory llvm/lib/Target/X86:
X86ISelDAGToDAG.cpp updated: 1.73 -> 1.74
---
Log message:
Add support for "m" inline asm constraints.
---
Diffs of the changes: (+29 -1)
X86ISelDAGToDAG.cpp | 30 +-
1 files changed, 29 insertions(+), 1 deletion(-)
Index: llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
diff -u llvm/lib/Target/X86/X86ISelDAGToDAG.cpp:1.73
llvm/lib/Target/X86/X86ISelDAGToDAG.cpp:1.74
--- llvm/lib/Target/X86/X86ISelDAGToDAG.cpp:1.73Fri Jun 2 17:38:37 2006
+++ llvm/lib/Target/X86/X86ISelDAGToDAG.cpp Thu Jun 8 13:03:49 2006
@@ -128,7 +128,13 @@
bool TryFoldLoad(SDOperand P, SDOperand N,
SDOperand &Base, SDOperand &Scale,
SDOperand &Index, SDOperand &Disp);
-
+/// SelectInlineAsmMemoryOperand - Implement addressing mode selection for
+/// inline asm expressions.
+virtual bool SelectInlineAsmMemoryOperand(const SDOperand &Op,
+ char ConstraintCode,
+ std::vector &OutOps,
+ SelectionDAG &DAG);
+
void EmitSpecialCodeForMain(MachineBasicBlock *BB, MachineFrameInfo *MFI);
inline void getAddressOperands(X86ISelAddressMode &AM, SDOperand &Base,
@@ -876,6 +882,28 @@
#endif
}
+bool X86DAGToDAGISel::
+SelectInlineAsmMemoryOperand(const SDOperand &Op, char ConstraintCode,
+ std::vector &OutOps, SelectionDAG
&DAG){
+ SDOperand Op0, Op1, Op2, Op3;
+ switch (ConstraintCode) {
+ case 'o': // offsetable??
+ case 'v': // not offsetable??
+ default: return true;
+ case 'm': // memory
+if (!SelectAddr(Op, Op0, Op1, Op2, Op3))
+ return true;
+break;
+ }
+
+ OutOps.resize(4);
+ Select(OutOps[0], Op0);
+ Select(OutOps[1], Op1);
+ Select(OutOps[2], Op2);
+ Select(OutOps[3], Op3);
+ return false;
+}
+
/// createX86ISelDag - This pass converts a legalized DAG into a
/// X86-specific DAG, ready for instruction scheduling.
///
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/test/Regression/CodeGen/X86/inline-asm.ll
Changes in directory llvm/test/Regression/CodeGen/X86:
inline-asm.ll added (r1.1)
---
Log message:
New testcase, using "AX" as i32.
---
Diffs of the changes: (+8 -0)
inline-asm.ll |8
1 files changed, 8 insertions(+)
Index: llvm/test/Regression/CodeGen/X86/inline-asm.ll
diff -c /dev/null llvm/test/Regression/CodeGen/X86/inline-asm.ll:1.1
*** /dev/null Thu Jun 8 13:20:23 2006
--- llvm/test/Regression/CodeGen/X86/inline-asm.ll Thu Jun 8 13:20:13 2006
***
*** 0
--- 1,8
+ ; RUN: llvm-as < %s | llc -march=x86
+
+ int %test1() {
+ ; Dest is AX, dest type = i32.
+ %tmp4 = call int asm sideeffect "FROB %0", "={ax}"()
+ ret int %tmp4
+ }
+
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
Changes in directory llvm/lib/CodeGen/SelectionDAG:
SelectionDAGISel.cpp updated: 1.253 -> 1.254
---
Log message:
Fix Regression/CodeGen/X86/inline-asm.ll, a case where inline asm causes
implement extension of a register.
---
Diffs of the changes: (+13 -6)
SelectionDAGISel.cpp | 19 +--
1 files changed, 13 insertions(+), 6 deletions(-)
Index: llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
diff -u llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1.253
llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1.254
--- llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1.253Fri May 26
18:13:20 2006
+++ llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp Thu Jun 8 13:22:48 2006
@@ -1780,15 +1780,19 @@
return DAG.getNode(ISD::BUILD_PAIR, ValueVT, Hi, Val);
}
- // Otherwise, if the return value was promoted, truncate it to the
+ // Otherwise, if the return value was promoted or extended, truncate it to
the
// appropriate type.
if (RegVT == ValueVT)
return Val;
- if (MVT::isInteger(RegVT))
-return DAG.getNode(ISD::TRUNCATE, ValueVT, Val);
- else
+ if (MVT::isInteger(RegVT)) {
+if (ValueVT < RegVT)
+ return DAG.getNode(ISD::TRUNCATE, ValueVT, Val);
+else
+ return DAG.getNode(ISD::ANY_EXTEND, ValueVT, Val);
+ } else {
return DAG.getNode(ISD::FP_ROUND, ValueVT, Val);
+ }
}
/// getCopyToRegs - Emit a series of CopyToReg nodes that copies the
@@ -1895,14 +1899,17 @@
if (PhysReg.first) {
if (VT == MVT::Other)
ValueVT = *PhysReg.second->vt_begin();
-RegVT = VT;
+
+// Get the actual register value type. This is important, because the user
+// may have asked for (e.g.) the AX register in i32 type. We need to
+// remember that AX is actually i16 to get the right extension.
+RegVT = *PhysReg.second->vt_begin();
// This is a explicit reference to a physical register.
Regs.push_back(PhysReg.first);
// If this is an expanded reference, add the rest of the regs to Regs.
if (NumRegs != 1) {
- RegVT = *PhysReg.second->vt_begin();
TargetRegisterClass::iterator I = PhysReg.second->begin();
TargetRegisterClass::iterator E = PhysReg.second->end();
for (; *I != PhysReg.first; ++I)
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/test/Regression/CodeGen/X86/inline-asm.ll
Changes in directory llvm/test/Regression/CodeGen/X86:
inline-asm.ll updated: 1.1 -> 1.2
---
Log message:
Add new testcase
---
Diffs of the changes: (+7 -1)
inline-asm.ll |8 +++-
1 files changed, 7 insertions(+), 1 deletion(-)
Index: llvm/test/Regression/CodeGen/X86/inline-asm.ll
diff -u llvm/test/Regression/CodeGen/X86/inline-asm.ll:1.1
llvm/test/Regression/CodeGen/X86/inline-asm.ll:1.2
--- llvm/test/Regression/CodeGen/X86/inline-asm.ll:1.1 Thu Jun 8 13:20:13 2006
+++ llvm/test/Regression/CodeGen/X86/inline-asm.ll Thu Jun 8 13:26:48 2006
@@ -2,7 +2,13 @@
int %test1() {
; Dest is AX, dest type = i32.
-%tmp4 = call int asm sideeffect "FROB %0", "={ax}"()
+%tmp4 = call int asm sideeffect "FROB $0", "={ax}"()
ret int %tmp4
}
+void %test2(int %V) {
+ ; input is AX, in type = i32.
+call void asm sideeffect "FROB $0", "{ax}"(int %V)
+ret void
+}
+
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
Changes in directory llvm/lib/CodeGen/SelectionDAG:
SelectionDAGISel.cpp updated: 1.254 -> 1.255
---
Log message:
Fix X86/inline-asm.ll:test2, a case where an input value was implicitly
truncated.
---
Diffs of the changes: (+6 -3)
SelectionDAGISel.cpp |9 ++---
1 files changed, 6 insertions(+), 3 deletions(-)
Index: llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
diff -u llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1.254
llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1.255
--- llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1.254Thu Jun 8
13:22:48 2006
+++ llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp Thu Jun 8 13:27:11 2006
@@ -1804,9 +1804,12 @@
// If there is a single register and the types differ, this must be
// a promotion.
if (RegVT != ValueVT) {
- if (MVT::isInteger(RegVT))
-Val = DAG.getNode(ISD::ANY_EXTEND, RegVT, Val);
- else
+ if (MVT::isInteger(RegVT)) {
+if (RegVT < ValueVT)
+ Val = DAG.getNode(ISD::TRUNCATE, RegVT, Val);
+else
+ Val = DAG.getNode(ISD::ANY_EXTEND, RegVT, Val);
+ } else
Val = DAG.getNode(ISD::FP_EXTEND, RegVT, Val);
}
Chain = DAG.getCopyToReg(Chain, Regs[0], Val, Flag);
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/include/llvm/Transforms/Scalar.h
Changes in directory llvm/include/llvm/Transforms: Scalar.h updated: 1.65 -> 1.66 --- Log message: Update some comments, and expose LCSSAID in preparation for having other passes require LCSSA. --- Diffs of the changes: (+1 -0) Scalar.h |1 + 1 files changed, 1 insertion(+) Index: llvm/include/llvm/Transforms/Scalar.h diff -u llvm/include/llvm/Transforms/Scalar.h:1.65 llvm/include/llvm/Transforms/Scalar.h:1.66 --- llvm/include/llvm/Transforms/Scalar.h:1.65 Fri May 26 08:58:26 2006 +++ llvm/include/llvm/Transforms/Scalar.h Thu Jun 8 15:02:53 2006 @@ -306,6 +306,7 @@ // This pass inserts phi nodes at loop boundaries to simplify other loop // optimizations. FunctionPass *createLCSSAPass(); +extern const PassInfo *LCSSAID; } // End llvm namespace ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/Transforms/Utils/LCSSA.cpp
Changes in directory llvm/lib/Transforms/Utils:
LCSSA.cpp updated: 1.15 -> 1.16
---
Log message:
Update some comments, and expose LCSSAID in preparation for having other passes
require LCSSA.
---
Diffs of the changes: (+8 -1)
LCSSA.cpp |9 -
1 files changed, 8 insertions(+), 1 deletion(-)
Index: llvm/lib/Transforms/Utils/LCSSA.cpp
diff -u llvm/lib/Transforms/Utils/LCSSA.cpp:1.15
llvm/lib/Transforms/Utils/LCSSA.cpp:1.16
--- llvm/lib/Transforms/Utils/LCSSA.cpp:1.15Mon Jun 5 23:36:36 2006
+++ llvm/lib/Transforms/Utils/LCSSA.cpp Thu Jun 8 15:02:53 2006
@@ -86,7 +86,9 @@
}
FunctionPass *llvm::createLCSSAPass() { return new LCSSA(); }
+const PassInfo *llvm::LCSSAID = X.getPassInfo();
+/// runOnFunction - Process all loops in the function, inner-most out.
bool LCSSA::runOnFunction(Function &F) {
bool changed = false;
LI = &getAnalysis();
@@ -100,6 +102,8 @@
return changed;
}
+/// visitSubloop - Recursively process all subloops, and then process the given
+/// loop if it has live-out values.
bool LCSSA::visitSubloop(Loop* L) {
for (Loop::iterator I = L->begin(), E = L->end(); I != E; ++I)
visitSubloop(*I);
@@ -131,7 +135,8 @@
return true;
}
-/// processInstruction -
+/// processInstruction - Given a live-out instruction, insert LCSSA Phi nodes,
+/// eliminate all out-of-loop uses.
void LCSSA::processInstruction(Instruction* Instr,
const std::vector& exitBlocks)
{
@@ -252,6 +257,8 @@
return AffectedValues;
}
+/// getValueDominatingBlock - Return the value within the potential dominators
+/// map that dominates the given block.
Instruction *LCSSA::getValueDominatingBlock(BasicBlock *BB,
std::map& PotDoms)
{
DominatorTree::Node* bbNode = DT->getNode(BB);
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
Changes in directory llvm/lib/Transforms/Scalar: LoopStrengthReduce.cpp updated: 1.83 -> 1.84 --- Log message: RewriteExpr, either the new PHI node of induction variable or the post-increment value, should be first cast to the appropriated type (to the type of the common expr). Otherwise, the rewrite of a use based on (common + iv) may end up with an incorrect type. --- Diffs of the changes: (+3 -0) LoopStrengthReduce.cpp |3 +++ 1 files changed, 3 insertions(+) Index: llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp diff -u llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp:1.83 llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp:1.84 --- llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp:1.83 Wed Apr 12 14:28:15 2006 +++ llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp Thu Jun 8 19:12:42 2006 @@ -1074,6 +1074,9 @@ if (L->contains(User.Inst->getParent())) User.Inst->moveBefore(LatchBlock->getTerminator()); } + if (RewriteOp->getType() != ReplacedTy) +RewriteOp = SCEVExpander::InsertCastOfTo(RewriteOp, ReplacedTy); + SCEVHandle RewriteExpr = SCEVUnknown::get(RewriteOp); // Clear the SCEVExpander's expression map so that we are guaranteed ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/Target/X86/X86ISelLowering.cpp
Changes in directory llvm/lib/Target/X86: X86ISelLowering.cpp updated: 1.227 -> 1.228 --- Log message: Minor compilation speed improvement. --- Diffs of the changes: (+2 -2) X86ISelLowering.cpp |4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) Index: llvm/lib/Target/X86/X86ISelLowering.cpp diff -u llvm/lib/Target/X86/X86ISelLowering.cpp:1.227 llvm/lib/Target/X86/X86ISelLowering.cpp:1.228 --- llvm/lib/Target/X86/X86ISelLowering.cpp:1.227 Tue Jun 6 18:30:24 2006 +++ llvm/lib/Target/X86/X86ISelLowering.cpp Fri Jun 9 01:24:42 2006 @@ -3413,8 +3413,8 @@ MachineFunction &MF = DAG.getMachineFunction(); const Function* Fn = MF.getFunction(); if (Fn->hasExternalLinkage() && - Fn->getName() == "main" && - Subtarget->TargetType == X86Subtarget::isCygwin) + Subtarget->TargetType == X86Subtarget::isCygwin && + Fn->getName() == "main") MF.getInfo()->setForceFramePointer(true); unsigned CC = cast(Op.getOperand(1))->getValue(); ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[llvm-commits] CVS: llvm/lib/Target/X86/X86MachineFunctionInfo.h
Changes in directory llvm/lib/Target/X86:
X86MachineFunctionInfo.h updated: 1.1 -> 1.2
---
Log message:
Comments to appease sabre.
---
Diffs of the changes: (+7 -1)
X86MachineFunctionInfo.h |8 +++-
1 files changed, 7 insertions(+), 1 deletion(-)
Index: llvm/lib/Target/X86/X86MachineFunctionInfo.h
diff -u llvm/lib/Target/X86/X86MachineFunctionInfo.h:1.1
llvm/lib/Target/X86/X86MachineFunctionInfo.h:1.2
--- llvm/lib/Target/X86/X86MachineFunctionInfo.h:1.1Tue Jun 6 18:30:24 2006
+++ llvm/lib/Target/X86/X86MachineFunctionInfo.hFri Jun 9 01:25:10 2006
@@ -18,8 +18,14 @@
namespace llvm {
+/// X86FunctionInfo - This class is derived from MachineFunction private
+/// X86 target-specific information for each MachineFunction.
class X86FunctionInfo : public MachineFunctionInfo {
- bool ForceFramePointer; // Function requires use of frame pointer.
+ // ForceFramePointer - True if the function is required to use of frame
+ // pointer for reasons other than it containing dynamic allocation or
+ // that FP eliminatation is turned off. For example, Cygwin main function
+ // contains stack pointer re-alignment code which requires FP.
+ bool ForceFramePointer;
public:
X86FunctionInfo(MachineFunction& MF) : ForceFramePointer(false) {}
bool getForceFramePointer() const { return ForceFramePointer;}
___
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
