Commits

Show all
Author Commit Message Labels Comments Date
Bill Wendling
--- Merging (from foreign repository) r68368 into '.': A test/CodeGen/X86/vec_i64.ll U lib/Target/X86/X86ISelLowering.cpp Added a x86 dag combine to increase the chances to use a movq for v2i64 on x86-32.
Tags
5 tags
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r67473 into '.': U include/llvm/ADT/SmallVector.h Fix PR3860 by correcting a predicate.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68277 into '.': U include/llvm/ADT/SmallVector.h fix overflow checks in SmallVector: "The code was doing "if (End+NumInputs > Capacity) ...". If End is close to 0xFFFFFFFF and NumInputs is large, it'll overflow, the condition will come out false, and the vector won't grow to accommodate the new elements, and the program will crash in memmove." Patch by Jeffrey Yasskin!
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68254 into '.': U include/llvm/PassAnalysisSupport.h U include/llvm/PassManagers.h U lib/VMCore/PassManager.cpp Clean up pass manager cache after each run.
Branches
Apple/Dib
Bill Wendling
Remove dead local variables.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68229 into '.': U tools/lto/LTOCodeGenerator.cpp Recognize arm triplets.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68054 into '.': U include/llvm/CodeGen/MachineBasicBlock.h U lib/CodeGen/MachineBasicBlock.cpp Constify arguments in isSuccessor and isLayoutSuccessor.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68142 into '.': A test/Transforms/InstCombine/select-2.ll U lib/Transforms/Scalar/InstructionCombining.cpp Throttle back "fold select into operand" transformation. InstCombine should not generate selects of two constants unless they are selects of 0 and 1. e.g. define i32 @t1(i32 %c, i32 %x) nounwind { %t1 = icmp eq i32 %c, 0 %t2 = lshr i32 %x, 18 %t3 = select i1 %t1, i32 %t2, i32 %x ret i32 %t3 } was turned into define i32 @t2(i32 %c, i32 %x) nounwind { %t1 = icmp…
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68133 into '.': U lib/Target/X86/X86ISelLowering.cpp i128 shift libcalls are not available on x86. --- Merging (from foreign repository) r68134 into '.': A test/CodeGen/X86/shift-i128.ll A test/CodeGen/X86/shift-i256.ll U lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp U lib/CodeGen/SelectionDAG/LegalizeTypes.h Fully general expansion of integer shift of any size.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68129 into '.': U include/llvm/CodeGen/MachineBasicBlock.h U lib/CodeGen/MachineBasicBlock.cpp U lib/Target/X86/AsmPrinter/X86ATTAsmPrinter.cpp Reapply 68073, with fixes. EH Landing-pad basic blocks are not entered via fall-through. Don't miss fallthroughs from blocks terminated by conditional branches. Also, move isOnlyReachableByFallthrough out of line.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68083 into '.': A test/CodeGen/X86/live-out-reg-info.ll U lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp Fix live-out reg logic to not insert over-aggressive AssertZExt instructions. This fixes lua.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68065 into '.': U test/CodeGen/X86/inline-asm-2addr.ll U lib/CodeGen/TwoAddressInstructionPass.cpp Turn a 2-address instruction into a 3-address one when it's profitable even if the two-address operand is killed. e.g. %reg1024<def> = MOV r1 %reg1025<def> = ADD %reg1024, %reg1026 r0 = MOV %reg1025 If it's not possible / profitable to commute ADD, then turning ADD into a LEA saves a copy.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68047 into '.': U lib/Analysis/AliasSetTracker.cpp getEntryFor() may invalidate DenseMap iterator. Walking an invalidated iterator is not a good idea.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r68066 into '.': U lib/Target/X86/X86Instr64bit.td U lib/Target/X86/X86InstrInfo.td U lib/Target/X86/X86ISelLowering.cpp U lib/Target/X86/X86ISelDAGToDAG.cpp U lib/Target/X86/X86ISelLowering.h When optimzing a mul by immediate into two, the resulting mul's should get a x86 specific node to avoid dag combiner from hacking on them further.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r67996 into '.': A test/CodeGen/Generic/2009-03-29-SoftFloatVectorExtract.ll U lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp U lib/CodeGen/SelectionDAG/LegalizeTypes.cpp U lib/CodeGen/SelectionDAG/LegalizeTypes.h Fix PR3899: add support for extracting floats from vectors when using -soft-float. Based on a patch by Jakob Stoklund Olesen.
Branches
Apple/Dib
Bill Wendling
Add a __builtin___memset_chk test.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r67917 into '.': A test/CodeGen/X86/imul-lea-2.ll U include/llvm/Target/TargetLowering.h U lib/CodeGen/SelectionDAG/DAGCombiner.cpp U lib/Target/X86/X86ISelLowering.cpp Optimize some 64-bit multiplication by constants into two lea's or one lea + shl since imulq is slow (latency 5). e.g. x * 40 => shlq $3, %rdi leaq (%rdi,%rdi,4), %rax This has the added benefit of allowing more multiply to be folded into addressing mode. e.g. a * 24 + b => leaq (%rdi,%rdi,2), %rax leaq (%rsi,%rax,8), %rax
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r67879 into '.': U lib/CodeGen/SelectionDAG/SelectionDAGBuild.h --- Merging (from foreign repository) r67881 into '.': U lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r67839 into '.': U lib/Transforms/Scalar/LICM.cpp Revert r67798.
Branches
Apple/Dib
Bill Wendling
--- Merging (from foreign repository) r67855 into '.': U Makefile.rules Use correct flag to create a dylib.
Branches
Apple/Dib
convert-repo
update tags
Dan Gohman
Fix this test so that it doesn't spuriously fail due to some unrelated debugging output happening to contain the string "store".
Rafael Avila de Espindola
Avoid hardcoding that X86 addresses have 4 operands.
Rafael Avila de Espindola
Use less hard coded constants to make the code less brittle.
Duncan Sands
Reapply r66415, which was reverted in r66426 for causing a bootstrap failure. Bootstraps here on x86-32-linux and x86-64-linux. Requested by the author Gabor Greif who says that a bug that might have been causing the failure has since been fixed.
Rafael Avila de Espindola
I am trying to add a segment to the X86 addresses matching to improve TLS support (see http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090309/075220.html), but that code is VERY brittle. This patch just makes it a bit more resistant.
Duncan Sands
Revert r67798: it breaks llvm-gcc bootstrap on x86-64-linux, presumably due to a miscompilation. make[4]: Entering directory `gcc-4.2.llvm-objects/x86_64-unknown-linux-gnu/libstdc++-v3/include' if [ ! -d "./x86_64-unknown-linux-gnu/bits/stdtr1c++.h.gch" ]; then \ mkdir -p ./x86_64-unknown-linux-gnu/bits/stdtr1c++.h.gch; \ fi; \ gcc-4.2.llvm-objects/./gcc/xgcc -shared-libgcc -Bgcc-4.2.llvm-objects/./gcc -nostdinc++ -Lgcc-4.2.llvm-objects/x86_64-unknown-linux-gnu/libst…
Mikhail Glushenkov
Documentation update. Expand a bit on various '--*-graph' options.
Mikhail Glushenkov
-write-graph now can be used with -o. Makes it possible to set the output file name.
Duncan Sands
Revert previous change in favour of an explanatory comment.
  1. Prev
  2. Next