Commits

Show all
Author Commit Message Labels Comments Date
Bill Wendling
$ svn merge -c 85934 https://llvm.org/svn/llvm-project/llvm/trunk --- Merging r85934 into '.': U lib/CodeGen/ScheduleDAGInstrs.cpp $ svn merge -c 85939 https://llvm.org/svn/llvm-project/llvm/trunk --- Merging r85939 into '.': U include/llvm/CodeGen/LatencyPriorityQueue.h U include/llvm/CodeGen/ScheduleDAG.h U lib/CodeGen/CriticalAntiDepBreaker.h U lib/CodeGen/ScheduleDAG.cpp U lib/CodeGen/CriticalAntiDepBreaker.cpp U lib/CodeGen/LatencyPriorityQueue.cpp U lib/CodeGen/ExactHazardRecognizer.cpp U lib/Cod…
Tags
llvmCore-2317
Branches
Apple/Leela
Bill Wendling
$ svn merge -c 85746 https://llvm.org/svn/llvm-project/llvm/trunk --- Merging r85746 into '.': $ svn merge -c 85762 https://llvm.org/svn/llvm-project/llvm/trunk --- Merging r85762 into '.': $ svn merge -c 85764 https://llvm.org/svn/llvm-project/llvm/trunk --- Merging r85764 into '.': $ svn merge -c 85765 https://llvm.org/svn/llvm-project/llvm/trunk --- Merging r85765 into '.': $ svn merge -c 85766 https://llvm.org/svn/llvm-project/llvm/trunk --- Merging r85766 into '.': $ svn merge -c 8…
Branches
Apple/Leela
Bill Wendling
Creating Leela from ToT.
Branches
Apple/Leela
Chris Lattner
remove a bunch of locking from LLVMContextImpl. Since only one thread can be banging on a context at a time, this isn't needed. Owen, please review.
Chris Lattner
improve comment.
Chris Lattner
add a comment about why we don't allow inlining indbr.
Evan Cheng
Fix tests.
Chris Lattner
the verifier shouldn't modify the IR.
Douglas Gregor
Reverting 85714, 85715, 85716, which are breaking the build
Dan Gohman
Add a function to Passes.h to allow clients to create instances of the ScalarEvolution pass without needing to #include ScalarEvolution.h.
Dan Gohman
Don't #include Pass.h from CallGraph.h.
Dan Gohman
Remove the #include of Pass.h from PassManager.h. This breaks a significant #include dependency, as frontends commonly pull in PassManager.h.
Chris Lattner
teach ipsccp and ipconstprop that a blockaddress doesn't 'take the address' of a function in a way that should prevent ip constprop. This allows clang/test/CodeGen/indirect-goto.c to pass with the new indirect goto lowering.
Chris Lattner
change llvm::MergeBlockIntoPredecessor to not merge two blocks BB1->BB2 when BB2 has its address taken. Since it ends up doing BB2->rauw(BB1), this can cause the address of the entry block to be taken. Since it is generally undesirable to nuke blocks whose address is taken, even when we can, just unconditionally stop this xform.
Chris Lattner
strengthen an assumption: RevectorBlockTo knows that PredBB ended in an uncond branch because the pass requires BreakCriticalEdges. However, BCE doesn't eliminate critical adges from indbrs.
Chris Lattner
fix an issue where the verifier would reject a function whose entry block had its address taken even if the blockaddress was dead.
Chris Lattner
if CostMetrics says to never duplicate some code, don't unswitch a loop. This prevents unswitching from duplicating indbr's.
Chris Lattner
constant fold indirectbr(blockaddress(%bb)) -> br label %bb.
Chris Lattner
improve x86 codegen support for blockaddress. We now compile the testcase into: _test1: ## @test1 ## BB#0: ## %entry leaq L_test1_bb6(%rip), %rax jmpq *%rax L_test1_bb: ## Address Taken LBB1_1: ## %bb movb $1, %al ret L_test1_bb6: …
Chris Lattner
pull check for return inst out of loop, never inline a callee that contains an indirectbr.
Chris Lattner
Fix BlockAddress::replaceUsesOfWithOnConstant to correctly maintain the block use count in SubclassData.
Chris Lattner
implement linker support for BlockAddress.
Chris Lattner
Revert 85678/85680. The decision is to stay with the current form of indirectbr, thus we don't need "blockaddr(@func, null)". Eliminate it for simplicity.
Evan Cheng
Use cbz and cbnz instructions.
Jim Grosbach
vml[as].f32 cause stalls in following advanced SIMD instructions. Avoid using them for scalar floating point operations for now.
Jim Grosbach
Consolidate test files
Jim Grosbach
Change to use FileCheck
Jim Grosbach
Make tests more explicit about which instructions are expected.
Jim Grosbach
Grammar tweak to comments
Chris Lattner
Make sure PRE doesn't split crit edges from indirectbr.
  1. Prev
  2. Next