aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--lib/CodeGen/BranchFolding.cpp42
-rw-r--r--test/CodeGen/Generic/2007-02-16-BranchFold.ll71
-rw-r--r--test/CodeGen/PowerPC/branch-opt.ll3
-rw-r--r--test/CodeGen/X86/2006-10-19-SwitchUnnecessaryBranching.ll2
4 files changed, 103 insertions, 15 deletions
diff --git a/lib/CodeGen/BranchFolding.cpp b/lib/CodeGen/BranchFolding.cpp
index 10262f2..9145418 100644
--- a/lib/CodeGen/BranchFolding.cpp
+++ b/lib/CodeGen/BranchFolding.cpp
@@ -431,6 +431,9 @@ bool BranchFolder::TailMergeBlocks(MachineFunction &MF) {
bool BranchFolder::OptimizeBranches(MachineFunction &MF) {
MadeChange = false;
+ // Make sure blocks are numbered in order
+ MF.RenumberBlocks();
+
for (MachineFunction::iterator I = ++MF.begin(), E = MF.end(); I != E; ) {
MachineBasicBlock *MBB = I++;
OptimizeBlock(MBB);
@@ -849,22 +852,35 @@ void BranchFolder::OptimizeBlock(MachineBasicBlock *MBB) {
PriorTBB, PriorFBB, PriorCond)) {
// Now we know that there was no fall-through into this block, check to
// see if it has a fall-through into its successor.
- if (!CanFallThrough(MBB, CurUnAnalyzable, CurTBB, CurFBB, CurCond)) {
- // Check all the predecessors of this block. If one of them has no fall
- // throughs, move this block right after it.
- for (MachineBasicBlock::pred_iterator PI = MBB->pred_begin(),
- E = MBB->pred_end(); PI != E; ++PI) {
- // Analyze the branch at the end of the pred.
- MachineBasicBlock *PredBB = *PI;
- MachineFunction::iterator PredFallthrough = PredBB; ++PredFallthrough;
- std::vector<MachineOperand> PredCond;
- if (PredBB != MBB && !CanFallThrough(PredBB)) {
- MBB->moveAfter(PredBB);
- MadeChange = true;
- return OptimizeBlock(MBB);
+ bool CurFallsThru = CanFallThrough(MBB, CurUnAnalyzable, CurTBB, CurFBB,
+ CurCond);
+
+ // Check all the predecessors of this block. If one of them has no fall
+ // throughs, move this block right after it.
+ for (MachineBasicBlock::pred_iterator PI = MBB->pred_begin(),
+ E = MBB->pred_end(); PI != E; ++PI) {
+ // Analyze the branch at the end of the pred.
+ MachineBasicBlock *PredBB = *PI;
+ MachineFunction::iterator PredFallthrough = PredBB; ++PredFallthrough;
+ if (PredBB != MBB && !CanFallThrough(PredBB)
+ && (!CurFallsThru || MBB->getNumber() >= PredBB->getNumber())) {
+ // If the current block doesn't fall through, just move it.
+ // If the current block can fall through and does not end with a
+ // conditional branch, we need to append an unconditional jump to
+ // the (current) next block. To avoid a possible compile-time
+ // infinite loop, move blocks only backward in this case.
+ if (CurFallsThru) {
+ MachineBasicBlock *NextBB = next(MachineFunction::iterator(MBB));
+ CurCond.clear();
+ TII->InsertBranch(*MBB, NextBB, 0, CurCond);
}
+ MBB->moveAfter(PredBB);
+ MadeChange = true;
+ return OptimizeBlock(MBB);
}
+ }
+ if (!CurFallsThru) {
// Check all successors to see if we can move this block before it.
for (MachineBasicBlock::succ_iterator SI = MBB->succ_begin(),
E = MBB->succ_end(); SI != E; ++SI) {
diff --git a/test/CodeGen/Generic/2007-02-16-BranchFold.ll b/test/CodeGen/Generic/2007-02-16-BranchFold.ll
new file mode 100644
index 0000000..13ed652
--- /dev/null
+++ b/test/CodeGen/Generic/2007-02-16-BranchFold.ll
@@ -0,0 +1,71 @@
+; RUN: llvm-as < %s | llc &&
+; RUN: llvm-as < %s | llc | grep jmp | wc -l | grep 0
+
+; ModuleID = 'bugpoint.test.bc'
+target datalayout = "e-p:32:32"
+target triple = "i686-apple-darwin8"
+ %struct.FILE = type { i8*, i32, i32, i16, i16, %struct.__sbuf, i32, i8*, i32 (i8*)*, i32 (i8*, i8*, i32)*, i64 (i8*, i64, i32)*, i32 (i8*, i8*, i32)*, %struct.__sbuf, %struct.__sFILEX*, i32, [3 x i8], [1 x i8], %struct.__sbuf, i32, i64 }
+ %struct.Index_Map = type { i32, %struct.item_set** }
+ %struct.Item = type { [4 x i16], %struct.rule* }
+ %struct.__sFILEX = type opaque
+ %struct.__sbuf = type { i8*, i32 }
+ %struct.dimension = type { i16*, %struct.Index_Map, %struct.mapping*, i32, %struct.plankMap* }
+ %struct.item_set = type { i32, i32, %struct.operator*, [2 x %struct.item_set*], %struct.item_set*, i16*, %struct.Item*, %struct.Item* }
+ %struct.list = type { i8*, %struct.list* }
+ %struct.mapping = type { %struct.list**, i32, i32, i32, %struct.item_set** }
+ %struct.nonterminal = type { i8*, i32, i32, i32, %struct.plankMap*, %struct.rule* }
+ %struct.operator = type { i8*, i8, i32, i32, i32, i32, %struct.table* }
+ %struct.pattern = type { %struct.nonterminal*, %struct.operator*, [2 x %struct.nonterminal*] }
+ %struct.plank = type { i8*, %struct.list*, i32 }
+ %struct.plankMap = type { %struct.list*, i32, %struct.stateMap* }
+ %struct.rule = type { [4 x i16], i32, i32, i32, %struct.nonterminal*, %struct.pattern*, i8 }
+ %struct.stateMap = type { i8*, %struct.plank*, i32, i16* }
+ %struct.table = type { %struct.operator*, %struct.list*, i16*, [2 x %struct.dimension*], %struct.item_set** }
+@outfile = external global %struct.FILE* ; <%struct.FILE**> [#uses=1]
+@str1 = external global [11 x i8] ; <[11 x i8]*> [#uses=1]
+
+implementation ; Functions:
+
+declare i32 @fprintf(%struct.FILE*, i8*, ...)
+
+define i16 @main_bb_2E_i9_2E_i_2E_i932_2E_ce(%struct.list* %l_addr.01.0.i2.i.i929, %struct.operator** %tmp66.i62.i.out) {
+newFuncRoot:
+ br label %bb.i9.i.i932.ce
+
+bb36.i.i.exitStub: ; preds = %bb.i9.i.i932.ce
+ store %struct.operator* %tmp66.i62.i, %struct.operator** %tmp66.i62.i.out
+ ret i16 0
+
+bb.i14.i.exitStub: ; preds = %bb.i9.i.i932.ce
+ store %struct.operator* %tmp66.i62.i, %struct.operator** %tmp66.i62.i.out
+ ret i16 1
+
+bb12.i.i935.exitStub: ; preds = %bb.i9.i.i932.ce
+ store %struct.operator* %tmp66.i62.i, %struct.operator** %tmp66.i62.i.out
+ ret i16 2
+
+bb20.i.i937.exitStub: ; preds = %bb.i9.i.i932.ce
+ store %struct.operator* %tmp66.i62.i, %struct.operator** %tmp66.i62.i.out
+ ret i16 3
+
+bb28.i.i938.exitStub: ; preds = %bb.i9.i.i932.ce
+ store %struct.operator* %tmp66.i62.i, %struct.operator** %tmp66.i62.i.out
+ ret i16 4
+
+bb.i9.i.i932.ce: ; preds = %newFuncRoot
+ %tmp1.i3.i.i930 = getelementptr %struct.list* %l_addr.01.0.i2.i.i929, i32 0, i32 0 ; <i8**> [#uses=1]
+ %tmp2.i4.i.i931 = load i8** %tmp1.i3.i.i930 ; <i8*> [#uses=1]
+ %tmp66.i62.i = bitcast i8* %tmp2.i4.i.i931 to %struct.operator* ; <%struct.operator*> [#uses=7]
+ %tmp1.i6.i = getelementptr %struct.operator* %tmp66.i62.i, i32 0, i32 2 ; <i32*> [#uses=1]
+ %tmp2.i7.i = load i32* %tmp1.i6.i ; <i32> [#uses=1]
+ %tmp3.i8.i = load %struct.FILE** @outfile ; <%struct.FILE*> [#uses=1]
+ %tmp5.i9.i = call i32 (%struct.FILE*, i8*, ...)* @fprintf( %struct.FILE* %tmp3.i8.i, i8* getelementptr ([11 x i8]* @str1, i32 0, i32 0), i32 %tmp2.i7.i ) ; <i32> [#uses=0]
+ %tmp7.i10.i = getelementptr %struct.operator* %tmp66.i62.i, i32 0, i32 5 ; <i32*> [#uses=1]
+ %tmp8.i11.i = load i32* %tmp7.i10.i ; <i32> [#uses=1]
+ switch i32 %tmp8.i11.i, label %bb36.i.i.exitStub [
+ i32 -1, label %bb.i14.i.exitStub
+ i32 0, label %bb12.i.i935.exitStub
+ i32 1, label %bb20.i.i937.exitStub
+ i32 2, label %bb28.i.i938.exitStub
+ ]
+}
diff --git a/test/CodeGen/PowerPC/branch-opt.ll b/test/CodeGen/PowerPC/branch-opt.ll
index 7f40a2d..9048f0f 100644
--- a/test/CodeGen/PowerPC/branch-opt.ll
+++ b/test/CodeGen/PowerPC/branch-opt.ll
@@ -1,5 +1,6 @@
; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 &&
-; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep 'b LBB.*cond_next48.loopexit' | wc -l | grep 1
+; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 | grep 'b LBB.*' | wc -l | grep 5
+;;; FIXME, 4 unconditional branches is minimal
target endian = big
target pointersize = 32
diff --git a/test/CodeGen/X86/2006-10-19-SwitchUnnecessaryBranching.ll b/test/CodeGen/X86/2006-10-19-SwitchUnnecessaryBranching.ll
index 5103679..48c6a1a 100644
--- a/test/CodeGen/X86/2006-10-19-SwitchUnnecessaryBranching.ll
+++ b/test/CodeGen/X86/2006-10-19-SwitchUnnecessaryBranching.ll
@@ -1,4 +1,4 @@
-; RUN: llvm-upgrade < %s | llvm-as | llc -march=x86 | %prcontext je 1 | grep "BB1_4:"
+; RUN: llvm-upgrade < %s | llvm-as | llc -march=x86 | %prcontext je 1 | grep "BB1_1:"
%str = internal constant [14 x sbyte] c"Hello world!\0A\00" ; <[14 x sbyte]*> [#uses=1]
%str = internal constant [13 x sbyte] c"Blah world!\0A\00" ; <[13 x sbyte]*> [#uses=1]
generated by cgit v1.1 at 2025-02-07 03:35:43 +0000