| Commit message (Collapse) | Author | Age | Files | Lines | |
|---|---|---|---|---|---|
| * | Materialize global addresses via movt/movw pair, this is always better | Anton Korobeynikov | 2009-11-24 | 1 | -0/+20 |
| | | | | | | | | | | | | | | than doing the same via constpool: 1. Load from constpool costs 3 cycles on A9, movt/movw pair - just 2. 2. Load from constpool might stall up to 300 cycles due to cache miss. 3. Movt/movw does not use load/store unit. 4. Less constpool entries => better compiler performance. This is only enabled on ELF systems, since darwin does not have needed relocations (yet). llvm-svn: 89720 | ||||
| * | move fconst[sd] to UAL. <rdar://7414913> | Jim Grosbach | 2009-11-23 | 2 | -5/+5 |
| | | | | | llvm-svn: 89700 | ||||
| * | Convert ARM tests to FileCheck for PR5307. | Edward O'Callaghan | 2009-11-22 | 7 | -35/+137 |
| | | | | | llvm-svn: 89593 | ||||
| * | Forgot to alter RUN line when converting to FileCheck. | Edward O'Callaghan | 2009-11-22 | 1 | -1/+1 |
| | | | | | llvm-svn: 89588 | ||||
| * | Fix for bad FileCheck converts in revision 89584. | Edward O'Callaghan | 2009-11-22 | 1 | -3/+2 |
| | | | | | llvm-svn: 89586 | ||||
| * | Convert a few tests to FileCheck for PR5307. | Edward O'Callaghan | 2009-11-22 | 4 | -6/+18 |
| | | | | | llvm-svn: 89584 | ||||
| * | Revert 89562. We're being sneakier than I was giving us credit for, and this | Jim Grosbach | 2009-11-21 | 3 | -4/+2 |
| | | | | | | | isn't necessary. llvm-svn: 89568 | ||||
| * | Darwin requires a frame pointer for all non-leaf functions to support correct | Jim Grosbach | 2009-11-21 | 3 | -2/+4 |
| | | | | | | | backtraces. llvm-svn: 89562 | ||||
| * | Remat VLDRD from constpool. Clean up some instruction property specifications. | Evan Cheng | 2009-11-20 | 1 | -0/+65 |
| | | | | | llvm-svn: 89478 | ||||
| * | Fix codegen of conditional move of immediates. We were not making use of the ↵ | Evan Cheng | 2009-11-20 | 2 | -1/+49 |
| | | | | | | | immediate forms of cmov instructions at all. llvm-svn: 89423 | ||||
| * | Fix buildbots. | Bob Wilson | 2009-11-18 | 1 | -1/+1 |
| | | | | | llvm-svn: 89274 | ||||
| * | Tail duplication still needs to iterate. Duplicating new instructions onto | Bob Wilson | 2009-11-18 | 1 | -0/+64 |
| | | | | | | | the tail of a block may make that block a new candidate for duplication. llvm-svn: 89264 | ||||
| * | Forgot to commit test fixes | Anton Korobeynikov | 2009-11-17 | 1 | -2/+2 |
| | | | | | llvm-svn: 89138 | ||||
| * | Detect need for autoalignment of the stack earlier to catch spills more | Jim Grosbach | 2009-11-15 | 1 | -2/+3 |
| | | | | | | | | conservatively. eliminateFrameIndex() machinery adjust to handle addr mode 6 (vld1/vst1) used for spills. Fix tests to expect aligned Q-reg spilling llvm-svn: 88874 | ||||
| * | - Change TargetInstrInfo::reMaterialize to pass in TargetRegisterInfo. | Evan Cheng | 2009-11-14 | 1 | -1/+2 |
| | | | | | | | | | - If destination is a physical register and it has a subreg index, use the sub-register instead. This fixes PR5423. llvm-svn: 88745 | ||||
| * | Add radar number. | Evan Cheng | 2009-11-14 | 1 | -0/+1 |
| | | | | | llvm-svn: 88739 | ||||
| * | Fix PR5412: Fix an inverted check and another missing sub-register check. | Evan Cheng | 2009-11-14 | 2 | -0/+235 |
| | | | | | llvm-svn: 88738 | ||||
| * | Fix PR5411. Bug in UpdateKills. A reg def partially define its super-registers. | Evan Cheng | 2009-11-13 | 1 | -0/+42 |
| | | | | | llvm-svn: 88719 | ||||
| * | Fix PR5410: LiveVariables lost subreg def: | Evan Cheng | 2009-11-13 | 1 | -0/+20 |
| | | | | | | | | | | | | | | | D0<def,dead> = ... ... = S0<use, kill> S0<def> = ... ... D0<def> = The first D0 def is correctly marked dead, however, livevariables should have added an implicit def of S0 or we end up with a use without a def. llvm-svn: 88690 | ||||
| * | Use Unified Assembly Syntax for the ARM backend. | Jim Grosbach | 2009-11-09 | 35 | -116/+119 |
| | | | | | llvm-svn: 86494 | ||||
| * | It turns out that the testcase in question uncovered subreg-handling bug. | Anton Korobeynikov | 2009-11-07 | 1 | -0/+1 |
| | | | | | | | | Add assert in asmprinter to catch such cases and xfail the tests. PR is to be filled. llvm-svn: 86375 | ||||
| * | Honour subreg machine operands during asmprinting | Anton Korobeynikov | 2009-11-06 | 1 | -0/+64 |
| | | | | | llvm-svn: 86303 | ||||
| * | Print VMOV (immediate) operands as hexadecimal values. Apple's assembler | Bob Wilson | 2009-11-06 | 1 | -0/+20 |
| | | | | | | | | | | will not accept negative values for these. LLVM's default operand printing sign extends values, so that valid unsigned values appear as negative immediates. Print all VMOV immediate operands as hex values to resolve this. Radar 7372576. llvm-svn: 86301 | ||||
| * | Remove ARMPCLabelIndex from ARMISelLowering. Use ↵ | Evan Cheng | 2009-11-06 | 1 | -4/+4 |
| | | | | | | | ARMFunctionInfo::createConstPoolEntryUId() instead. llvm-svn: 86294 | ||||
| * | Update these tests for the new label names. | Dan Gohman | 2009-11-05 | 1 | -3/+3 |
| | | | | | llvm-svn: 86192 | ||||
| * | Attempt again to fix buildbot failures: make expected output less specific | Bob Wilson | 2009-11-05 | 1 | -12/+9 |
| | | | | | | | and compile with -mtriple to specify *-apple-darwin targets. llvm-svn: 86081 | ||||
| * | Fix broken test. | Bob Wilson | 2009-11-04 | 1 | -2/+2 |
| | | | | | llvm-svn: 86045 | ||||
| * | Add test for ARM indirectbr codegen. | Bob Wilson | 2009-11-04 | 1 | -0/+63 |
| | | | | | llvm-svn: 86042 | ||||
| * | fconsts / fconstd immediate should be proceeded with #. | Evan Cheng | 2009-11-03 | 1 | -4/+4 |
| | | | | | llvm-svn: 85952 | ||||
| * | Re-apply 85799. It turns out my code isn't buggy. | Evan Cheng | 2009-11-03 | 1 | -1/+1 |
| | | | | | llvm-svn: 85947 | ||||
| * | Fix PR5367. QPR_8 is the super regclass of DPR_8 and SPR_8. | Evan Cheng | 2009-11-03 | 2 | -1/+64 |
| | | | | | llvm-svn: 85871 | ||||
| * | Revert r85049, it is causing PR5367 | Anton Korobeynikov | 2009-11-03 | 1 | -1/+1 |
| | | | | | llvm-svn: 85847 | ||||
| * | Revert 85799 for now. It might be breaking llvm-gcc driver. | Evan Cheng | 2009-11-02 | 1 | -1/+1 |
| | | | | | llvm-svn: 85827 | ||||
| * | Initilize the machine LICM CSE map upon the first time an instruction is ↵ | Evan Cheng | 2009-11-02 | 1 | -1/+1 |
| | | | | | | | | | | | hoisted to the loop preheader. Add instructions which are already in the preheader block that may be common expressions of those that are hoisted out. These does get a few more instructions CSE'ed. llvm-svn: 85799 | ||||
| * | Remove an irrelevant and poorly reduced test case. | Evan Cheng | 2009-11-02 | 1 | -414/+0 |
| | | | | | llvm-svn: 85794 | ||||
| * | Handle splats of undefs properly. This includes the testcase for PR5364 as well. | Anton Korobeynikov | 2009-11-02 | 1 | -0/+20 |
| | | | | | llvm-svn: 85767 | ||||
| * | 64-bit FP loads & stores operate on both NEON and VFP pipelines. | Anton Korobeynikov | 2009-11-02 | 1 | -0/+37 |
| | | | | | llvm-svn: 85765 | ||||
| * | vml[as].f32 cause stalls in following advanced SIMD instructions. Avoid using | Jim Grosbach | 2009-10-31 | 2 | -3/+5 |
| | | | | | | | them for scalar floating point operations for now. llvm-svn: 85697 | ||||
| * | Update test to be more explicit about what instruction sequences are ↵ | Jim Grosbach | 2009-10-31 | 1 | -3/+16 |
| | | | | | | | expected for each operation. llvm-svn: 85689 | ||||
| * | Expand 64-bit logical shift right inline | Jim Grosbach | 2009-10-31 | 1 | -1/+1 |
| | | | | | llvm-svn: 85687 | ||||
| * | Expand 64-bit arithmetic shift right inline | Jim Grosbach | 2009-10-31 | 1 | -1/+1 |
| | | | | | llvm-svn: 85685 | ||||
| * | Expand 64 bit left shift inline rather than using the libcall. For now, this | Jim Grosbach | 2009-10-31 | 1 | -1/+1 |
| | | | | | | | | is unconditional. Making it still use the libcall when optimizing for size would be a good adjustment. llvm-svn: 85675 | ||||
| * | Add missing colons for FileCheck. | Benjamin Kramer | 2009-10-31 | 1 | -1/+1 |
| | | | | | llvm-svn: 85674 | ||||
| * | Convert to FileCheck | Jim Grosbach | 2009-10-31 | 1 | -5/+9 |
| | | | | | llvm-svn: 85673 | ||||
| * | This fixes functions like | Rafael Espindola | 2009-10-30 | 1 | -0/+17 |
| | | | | | | | | | | | | void f (int a1, int a2, int a3, int a4, int a5,...) In ARMTargetLowering::LowerFormalArguments if the function has 4 or more regular arguments we used to set VarArgsFrameIndex using an offset of 0, which is only correct if the function has exactly 4 regular arguments. llvm-svn: 85590 | ||||
| * | Use fconsts and fconstd to materialize small fp constants. | Evan Cheng | 2009-10-28 | 1 | -0/+33 |
| | | | | | llvm-svn: 85362 | ||||
| * | Add missing testcase. | Rafael Espindola | 2009-10-27 | 1 | -0/+14 |
| | | | | | llvm-svn: 85266 | ||||
| * | Fix the rest of the ARM failures by converting them to FileCheck. | Bob Wilson | 2009-10-27 | 7 | -39/+74 |
| | | | | | llvm-svn: 85208 | ||||
| * | Fix some more failures by converting to FileCheck. | Bob Wilson | 2009-10-27 | 7 | -23/+44 |
| | | | | | llvm-svn: 85207 | ||||
| * | Convert to FileCheck, fixing failure due to tab change in the process. | Bob Wilson | 2009-10-27 | 1 | -4/+3 |
| | | | | | llvm-svn: 85204 | ||||

