summaryrefslogtreecommitdiffstats
path: root/llvm
Commit message (Collapse)AuthorAgeFilesLines
* don't eliminate address-taken blocks here.Chris Lattner2010-03-151-3/+15
| | | | llvm-svn: 98550
* SIGN_EXTEND from the same type as the dest is valid.Chris Lattner2010-03-151-1/+1
| | | | llvm-svn: 98548
* sink the call to VT.getSizeInBits() down into its uses,Chris Lattner2010-03-151-7/+5
| | | | | | not all unary nodes necessarily have a simple result type. llvm-svn: 98547
* Treat copysignl like the other copysign functions.Duncan Sands2010-03-151-1/+1
| | | | llvm-svn: 98542
* Don't save a temporary string into a StringRef field.Jeffrey Yasskin2010-03-151-1/+1
| | | | llvm-svn: 98538
* fix MCSectionELF to not leak memory, just like I did for MCSymbol.Chris Lattner2010-03-153-9/+12
| | | | | | | MCSectionMachO is already fine (yay for fixed size arrays?), MCSectionCOFF still leaks. llvm-svn: 98537
* fix a memory leak yjasskin pointed out: MCSymbol is bump pointerChris Lattner2010-03-153-19/+22
| | | | | | | | | | allocated and thus not freed. This is cool except that it contains and std::string so the string data didn't get freed. In any case there is no reason to redundantly store the string data in the MCSymbol anyway, just make the MCSymbol ref the string data in the MCContext StringMap. llvm-svn: 98536
* eliminate some #if 0 code I added in r96905, type inference Chris Lattner2010-03-151-18/+0
| | | | | | now enforces that input/output named values have hte same type. llvm-svn: 98535
* Completely rewrite tblgen's type inference mechanism,Chris Lattner2010-03-157-458/+733
| | | | | | | | | | | | | | | | | | | | | | | | changing the primary datastructure from being a "std::vector<unsigned char>" to being a new TypeSet class that actually has (gasp) invariants! This changes more things than I remember, but one major innovation here is that it enforces that named input values agree in type with their output values. This also eliminates code that transparently assumes (in some cases) that SDNodeXForm input/output types are the same, because this is wrong in many case. This also eliminates a bug which caused a lot of ambiguous patterns to go undetected, where a register class would sometimes pick the first possible type, causing an ambiguous pattern to get arbitrary results. With all the recent target changes, this causes no functionality change! llvm-svn: 98534
* MachineMove ctor doesn't need to to mutate input, add 'const'Chris Lattner2010-03-151-1/+2
| | | | llvm-svn: 98533
* do some serious surgery on CellSPU to get it back into a worldChris Lattner2010-03-154-38/+50
| | | | | | where it uses types consistently. llvm-svn: 98532
* fix a few more ambiguous types.Chris Lattner2010-03-152-4/+4
| | | | llvm-svn: 98531
* add some missing typesChris Lattner2010-03-151-2/+2
| | | | llvm-svn: 98530
* Tell Valgrind when we modify already-executed machine code so it knowsJeffrey Yasskin2010-03-1511-13/+290
| | | | | | | to re-instrument the code. We depend on the system valgrind.h to avoid adding a new license. llvm-svn: 98529
* various cleanups from danielChris Lattner2010-03-151-8/+8
| | | | llvm-svn: 98528
* fix an ambiguous pattern, contrary to expectations, scalar_to_vectorChris Lattner2010-03-151-1/+1
| | | | | | | doesn't have a type constraint on the scalar because we don't have an 'sAny' type. llvm-svn: 98527
* remove dead method.Chris Lattner2010-03-152-11/+0
| | | | llvm-svn: 98526
* don't forget to close a FD on an error condition, found byChris Lattner2010-03-141-1/+2
| | | | | | cppcheck, PR6617. Patch by Ettl Martin! llvm-svn: 98525
* tidy up and expound more on how half-float works.Chris Lattner2010-03-141-14/+21
| | | | llvm-svn: 98524
* tidy indentationChris Lattner2010-03-141-3/+3
| | | | llvm-svn: 98523
* no really, all 64-bit cpu's have cmov support. This shouldChris Lattner2010-03-141-1/+6
| | | | | | fix the rest of the buildbot failures on non-x86 hosts. llvm-svn: 98522
* filecheckize a test and mark these wiht a cpu so it passesChris Lattner2010-03-143-6/+13
| | | | | | on hosts without cmovs. llvm-svn: 98521
* all 64-bit cpus have cmov, this should fix CodeGen/X86/cmov.llChris Lattner2010-03-141-1/+2
| | | | | | (at least) on non-x86 builders. llvm-svn: 98520
* Remove pointless forward declaration, MSVC got confused by this.Benjamin Kramer2010-03-141-1/+0
| | | | llvm-svn: 98519
* Turn calls to copysignl into an FCOPYSIGN node. Handle FCOPYSIGN nodesDuncan Sands2010-03-146-1/+29
| | | | | | | with ppc_f128 type by having the type legalizer turn these back into a call to copysignl. llvm-svn: 98514
* Rename SDDbgValue.h to SDNodeDbgValue.h for consistency.Evan Cheng2010-03-144-6/+6
| | | | llvm-svn: 98513
* fix ShrinkDemandedOps to not leave dead nodes around,Chris Lattner2010-03-142-0/+25
| | | | | | fixing PR6607 llvm-svn: 98512
* rewrite ShrinkDemandedOps to be faster and indent less,Chris Lattner2010-03-141-48/+64
| | | | | | no functionality change. llvm-svn: 98511
* Fix jit encoding bugs.Evan Cheng2010-03-141-4/+4
| | | | llvm-svn: 98510
* make -view-isel-dags print after the 'ShrinkDemandedOps' pass.Chris Lattner2010-03-141-2/+2
| | | | llvm-svn: 98509
* don't have i386-specific tests in CodeGen/Generic, PR6601.Chris Lattner2010-03-1422-30/+0
| | | | llvm-svn: 98508
* fix PR6605, X86ISD::CMP always returns i32 (EFLAGS), notChris Lattner2010-03-143-3/+27
| | | | | | the operand type. llvm-svn: 98507
* Fix typoAnton Korobeynikov2010-03-141-1/+1
| | | | llvm-svn: 98506
* Document fp16 intrinsicsAnton Korobeynikov2010-03-141-0/+90
| | | | llvm-svn: 98505
* Feature test for half precision FP.Anton Korobeynikov2010-03-141-0/+32
| | | | llvm-svn: 98504
* Add substarget feature for FP16Anton Korobeynikov2010-03-144-1/+10
| | | | llvm-svn: 98503
* Add codegen support for FP16 on ARMAnton Korobeynikov2010-03-143-12/+54
| | | | llvm-svn: 98502
* Make default expansion for FP16 <-> FP32 nodes into libcallsAnton Korobeynikov2010-03-148-0/+51
| | | | llvm-svn: 98501
* Add DAG nodes to represent FP16 <-> FP32 intrinsicsAnton Korobeynikov2010-03-142-0/+14
| | | | llvm-svn: 98500
* CleanupAnton Korobeynikov2010-03-141-6/+4
| | | | llvm-svn: 98499
* The fp16 <-> fp32 intrinsics are pure, mark them so they can be CSE'd, etc.Anton Korobeynikov2010-03-141-2/+4
| | | | llvm-svn: 98498
* Add intrinsics to represent fp16 <-> fp32 conversionsAnton Korobeynikov2010-03-141-0/+8
| | | | llvm-svn: 98497
* add support for pentium class CPUs which do not have cmov,Chris Lattner2010-03-144-16/+64
| | | | | | PR4841. Patch by Craig Smith! llvm-svn: 98496
* fix AsmPrinter::GetBlockAddressSymbol to always return a uniqueChris Lattner2010-03-148-31/+37
| | | | | | | label instead of trying to form one based on the BB name (which causes collisions if the name is empty). This fixes PR6608 llvm-svn: 98495
* comment fix.Chris Lattner2010-03-141-1/+1
| | | | llvm-svn: 98494
* shrink 4-byte branches to 1-byte branches when lowering fromChris Lattner2010-03-141-0/+23
| | | | | | | MachineInstr -> MCInst. This is what the assembler backend wants, it relaxes from smaller to larger things. This fixes rdar://7750815 llvm-svn: 98493
* Skip over debug info when trying to merge two return BBs.Bill Wendling2010-03-141-4/+9
| | | | llvm-svn: 98491
* Make returns more consistent with others.Bill Wendling2010-03-141-2/+2
| | | | llvm-svn: 98490
* get MMI out of the label uniquing business, just go to MCContextChris Lattner2010-03-1413-46/+25
| | | | | | to get unique assembler temporary labels. llvm-svn: 98489
* fix these two get the mcsymbol operand instead of imm operand.Chris Lattner2010-03-141-5/+2
| | | | llvm-svn: 98487
OpenPOWER on IntegriCloud