aboutsummaryrefslogtreecommitdiff
path: root/lib/CodeGen/SelectionDAG
AgeCommit message (Collapse)Author
2010-07-30Fix for bug reported by Evzen Muller on llvm-commits: make sure to correctlyEli Friedman
check the range of the constant when optimizing a comparison between a constant and a sign_extend_inreg node. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109854 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-27Fix a crash in the dag combiner caused by ↵Nate Begeman
ConstantFoldBIT_CONVERTofBUILD_VECTOR calling itself recursively and returning a SCALAR_TO_VECTOR node, but assuming the input was always a BUILD_VECTOR. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109519 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-27It's better to have the arrays, which would trigger the creation of stackBill Wendling
protectors, to be near the stack protectors on the stack. Accomplish this by tagging the stack object with a predicate that indicates that it would trigger this. In the prolog-epilog inserter, assign these objects to the stack after the stack protector but before the other objects. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109481 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-26The "excess register pressure" returned by HighRegPressure() is not accurate ↵Evan Cheng
enough to factor into scheduling priority. Eliminate it and add early exits to speed up scheduling. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109449 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-26Handle Values with no value in getCopyFromRegs.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109415 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-26Pacify gcc-4.5 which wrongly thinks that RExcess (passed as the Excess ↵Duncan Sands
parameter) may be used uninitialized in the callers of HighRegPressure. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109393 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-25Add comments.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109383 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-25Fix crashes when scheduling a CopyToReg node -- getMachineOpcode asserts onBob Wilson
those. Radar 8231572. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109367 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-24Add an ILP scheduler. This is a register pressure aware scheduler that'sEvan Cheng
appropriate for targets without detailed instruction iterineries. The scheduler schedules for increased instruction level parallelism in low register pressure situation; it schedules to reduce register pressure when the register pressure becomes high. On x86_64, this is a win for all tests in CFP2000. It also sped up 256.bzip2 by 16%. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109300 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-23- Allow target to specify when is register pressure "too high". In most cases,Evan Cheng
it's too late to start backing off aggressive latency scheduling when most of the registers are in use so the threshold should be a bit tighter. - Correctly handle live out's and extract_subreg etc. - Enable register pressure aware scheduling by default for hybrid scheduler. For ARM, this is almost always a win on # of instructions. It's runtime neutral for most of the tests. But for some kernels with high register pressure it can be a huge win. e.g. 464.h264ref reduced number of spills by 54 and sped up by 20%. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109279 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-23Use the proper type for shift counts. This fixes a bootstrap error.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109265 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-23DAGCombine (shl (anyext x, c)) to (anyext (shl x, c)) if the high bitsDan Gohman
are not demanded. This often allows the anyext to be folded away. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109242 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-23Make SDNode::dump() print a newline at the end.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109234 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-2380-col.Eric Christopher
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109205 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-22keep in 80 colsGabor Greif
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109122 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-22mass elimination of reliance on automatic iterator dereferencingGabor Greif
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109103 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-22Re-apply r109079 with fix.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109083 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-22Revert r109079, which broke a lot of CodeGen tests.Owen Anderson
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109082 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-22Initialize RegLimit only when register pressure is being tracked.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109079 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-21More register pressure aware scheduling work.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109064 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-21Teach bottom up pre-ra scheduler to track register pressure. Work in progress.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108991 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-19After a custom inserter, in a block which has constant instructions,Dan Gohman
update the current basic block in addition to the current insert position, so that they remain consistent. This fixes rdar://8204072. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108765 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-19ARM has to provide its own TargetLowering::findRepresentativeClass because ↵Evan Cheng
its scalar floating point registers alias its vector registers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108761 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-19Teach computeRegisterProperties() to compute "representative" register class ↵Evan Cheng
for legal value types. A "representative" register class is the largest legal super-reg register class for a value type. e.g. On i386, GR32 is the rep register class for i8 / i16 / i32; on x86_64 it would be GR64. This property will be used by the register pressure tracking instruction scheduler. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108735 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-19Remove r108639 now that it is handled by InstCombine instead.Owen Anderson
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108688 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-18Add a DAGCombine xform to fold away redundant float->double->float ↵Owen Anderson
conversions around sqrt instructions. I am assured by people more knowledgeable than me that there are no rounding issues in eliminating this. This fixed <rdar://problem/8197504>. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108639 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-17Propagate alloca alignment information via variable size object frameEric Christopher
information. No functional change yet. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108583 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-16Revert r108369, sorting llvm.dbg.declare information by source position,Dan Gohman
since it doesn't work for front-ends which don't emit column information (which includes llvm-gcc in its present configuration), and doesn't work for clang for K&R style variables where the variables are declared in a different order from the parameter list. Instead, make a separate pass through the instructions to collect the llvm.dbg.declare instructions in order. This ensures that the debug information for variables is emitted in this order. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108538 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-16Use the source-order scheduler instead of the "fast" scheduler at -O0,Dan Gohman
because it's more likely to keep debug line information in its original order. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108496 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-16The SelectionDAGBuilder's handling of debug info, on rareDale Johannesen
occasions, caused code to be generated in a different order. All cases I've seen involved float softening in the type legalizer, and this could be perhaps be fixed there, but it's better not to generate things differently in the first place. 7797940 (6/29/2010..7/15/2010). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108484 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-15Revert. This isn't the correct way to go.Bill Wendling
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108478 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-15Handle code gen for the unreachable instruction if it's the only instruction inBill Wendling
the function. We'll just turn it into a "trap" instruction instead. The problem with not handling this is that it might generate a prologue without the equivalent epilogue to go with it: $ cat t.ll define void @foo() { entry: unreachable } $ llc -o - t.ll -relocation-model=pic -disable-fp-elim -unwind-tables .section __TEXT,__text,regular,pure_instructions .globl _foo .align 4, 0x90 _foo: ## @foo Leh_func_begin0: ## BB#0: ## %entry pushq %rbp Ltmp0: movq %rsp, %rbp Ltmp1: Leh_func_end0: ... The unwind tables then have bad data in them causing all sorts of problems. Fixes <rdar://problem/8096481>. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108473 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-15Split -enable-finite-only-fp-math to two options:Evan Cheng
-enable-no-nans-fp-math and -enable-no-infs-fp-math. All of the current codegen fp math optimizations only care whether the fp arithmetics arguments and results can never be NaN. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108465 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-15Fix crash reported in PR7653.Devang Patel
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108441 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-1480-col.Eric Christopher
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108381 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-14Properly restore DebugLoc after leaving the local constant area.Dan Gohman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108364 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-14Delete fast-isel's trivial load optimization; it breaks debugging becauseDan Gohman
it can look past points where a debugger might modify user variables. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108336 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-14Don't propagate debug locations to instructions for materializingDan Gohman
constants, since they may not be emited near the other instructions which get the same line, and this confuses debug info. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108302 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-13In inline asm treat indirect 'X' constraint as 'm'.Dale Johannesen
This may not be right in all cases, but it's better than asserting which it was doing before. PR 7528. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108268 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-12Fix a typo and fit in 80 columns. Found by Bob Wilson.Rafael Espindola
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108164 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-12Convert some tab stops into spaces.Duncan Sands
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108130 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-11Use COPY for fast-isel bitconvert, but don't create cross-class copies.Jakob Stoklund Olesen
This doesn't change the behavior of SelectBitcast for X86. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108073 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-11Fix va_arg for doubles. With this patch VAARG nodes always contain theRafael Espindola
correct alignment information, which simplifies ExpandRes_VAARG a bit. The patch introduces a new alignment information to TargetLoweringInfo. This is needed since the two natural candidates cannot be used: * The 's' in target data: If this is set to the minimal alignment of any argument, getCallFrameTypeAlignment would return 4 for doubles on ARM for example. * The getTransientStackAlignment method. It is possible for an architecture to have argument less aligned than what we maintain the stack pointer. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108072 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-11Use COPY for extracting ImplicitDef'ed values from fast-isel instructions.Jakob Stoklund Olesen
This assumes that the registers can be copied which is probably a safe assumption. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108070 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-11Use COPY in FastISel everywhere it is safe and trivial.Jakob Stoklund Olesen
The remaining copyRegToReg calls actually check the return value (shock!), so we cannot trivially replace them with COPY instructions. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108069 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-10Fix a bug in the code which re-inserts DBG_VALUE nodes after scheduling;Dan Gohman
if a block is split (by a custom inserter), the insert point may be in a different block than it was originally. This fixes 32-bit llvm-gcc bootstrap builds, and I haven't been able to reproduce it otherwise. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108060 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-10Emit COPY instructions instead of using copyRegToReg in InstrEmitter,Jakob Stoklund Olesen
ScheduleDAGEmit, TwoAddressLowering, and PHIElimination. This switches the bulk of register copies to using COPY, but many less used copyRegToReg calls remain. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108050 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-10Insert IMPLICIT_DEF instructions at the current insert position, notDan Gohman
at the end of the block. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108045 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-10Reapply bottom-up fast-isel, with several fixes for x86-32:Dan Gohman
- Check getBytesToPopOnReturn(). - Eschew ST0 and ST1 for return values. - Fix the PIC base register initialization so that it doesn't ever fail to end up the top of the entry block. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108039 91177308-0d34-0410-b5e6-96231b3b80d8
2010-07-09Clarify what mysterious check means.Bill Wendling
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@108005 91177308-0d34-0410-b5e6-96231b3b80d8
generated by cgit v1.2.3-18-g5258 (git 2.32.0) at 2025-07-07 21:05:42 +0000