- 4f0f86d Teach the selector to fold lo(g) into load instruction immediate fields by Chris Lattner · 20 years ago
- 860e886 Add an initial hack at legalizing GlobalAddress into the appropriate nodes by Chris Lattner · 20 years ago
- 2823b3e When lowering direct calls, lower them to use a targetglobaladress directly by Chris Lattner · 20 years ago
- 422b0ce Patch to clean up function call pseudos and support the BLA instruction, by Nate Begeman · 20 years ago
- 7d7b967 Don't emit "32" for unordered comparison by Chris Lattner · 20 years ago
- ed048c0 add a hack to get code with ordered comparisons working. This hack is by Chris Lattner · 20 years ago
- 6df2507 add support for branch on ordered/unordered. by Chris Lattner · 20 years ago
- 6e61ca6 autogen undef by Chris Lattner · 20 years ago
- 9c73f09 Autogen fsel by Chris Lattner · 20 years ago
- e6115b3 Autogen a few new ppc-specific nodes by Chris Lattner · 20 years ago
- 8ecedbe The dag isel generator generates this now by Chris Lattner · 20 years ago
- 99ea9da Be a bit more paranoid about calling SelectNodeTo by Chris Lattner · 20 years ago
- 3393e80 Fix a couple of minor bugs. The first fixes povray, the second fixes things by Chris Lattner · 20 years ago
- dabb829 Instead of aborting if not a case we can handle specially, break out and by Chris Lattner · 20 years ago
- 405e3ec Invert the TargetLowering flag that controls divide by consant expansion. by Nate Begeman · 20 years ago
- 6957523 Move the target constant divide optimization up into the dag combiner, so by Nate Begeman · 20 years ago
- 2d5aff7 Write patterns for the various shl and srl patterns that don't involve by Nate Begeman · 20 years ago
- 8be1fa5 Convert these cases to patterns by Chris Lattner · 20 years ago
- 8d94832 Woo, it kinda works. We now generate this atrociously bad, but correct, by Nate Begeman · 20 years ago
- da32c9e Make a new reg class for 64 bit regs that aliases the 32 bit regs. This by Nate Begeman · 20 years ago
- 4a95945 Add the ability to lower return instructions to TargetLowering. This by Nate Begeman · 20 years ago
- 1d9d742 First bits of 64 bit PowerPC stuff, currently disabled. A lot of this is by Nate Begeman · 20 years ago
- 21e463b More PPC32 -> PPC changes, as well as merging some classes that were by Nate Begeman · 20 years ago
- 4cb5a1b Remove some dead code: the ORI/ORIS cases are autogen'd. This makes by Chris Lattner · 20 years ago
- 75c9f67 These instructions are now autogenerated by Chris Lattner · 20 years ago
- 651dea7 remove dead code by Chris Lattner · 20 years ago
- 16e71f2 Rename PPC32*.h to PPC*.h by Chris Lattner · 20 years ago
- 2668959 Rename PowerPC*.h to PPC*.h by Chris Lattner · 20 years ago
- 4c7b43b Eliminate PowerPC.td and PPC32.td, consolidating them into PPC.td by Chris Lattner · 20 years ago
- 3f31d43 These are now autogenerated by Chris Lattner · 20 years ago
- 65a419a Disable formation of rlwinm instructions from SRA bases. This fixes by Chris Lattner · 20 years ago
- cf01a70 When preselecting, favor things that have low depth to select first. This by Chris Lattner · 20 years ago
- 6a16f6a Pull out Call, reducing stack frame size from 6032 bytes to 5184 bytes. by Chris Lattner · 20 years ago
- 222adac Pull out setcc, this reduces stack frame size from 7520 to 6032 bytes by Chris Lattner · 20 years ago
- 2b63e4c Pull two more methods out, reducing stack frame size from 8224 -> 7520 bytes by Chris Lattner · 20 years ago
- bd937b9 Add a recursive-iterative hybrid stage to attempt to reduce stack space, this by Chris Lattner · 20 years ago
- 43f07a4 another solution to the fsel issue. Instead of having 4 variants, just force by Chris Lattner · 20 years ago
- 867940d fsel can take a different FP type for the comparison and for the result. As such by Chris Lattner · 20 years ago
- ca0a477 Minor tweak to the branch selector. When emitting a two-way branch, and if by Chris Lattner · 20 years ago
- 2c1760f fix typo by Chris Lattner · 20 years ago
- 919c032 Modify the ppc backend to use two register classes for FP: F8RC and F4RC. by Chris Lattner · 20 years ago
- 242f255 Should be using flag and not chain. by Jim Laskey · 20 years ago
- 237733e Remove code for patterns that are autogenerated by Chris Lattner · 20 years ago
- d3d2cf5 Never rely on ReplaceAllUsesWith when selecting, use CodeGenMap instead. by Chris Lattner · 20 years ago
- d8ead9e Autogen MUL, move FP cases together by Chris Lattner · 20 years ago
- 88add10 disentangle FP from INT versions of div/mul by Chris Lattner · 20 years ago
- 4a7de21 Use the autogenerated matcher for ADD/SUB by Chris Lattner · 20 years ago
- 615c2d0 Add FP versions of the binary operators, keeping the int and fp worlds seperate. by Chris Lattner · 20 years ago
- 393e138 All (xor *) cases are autogenerated now by Chris Lattner · 20 years ago
- 1bd8b7b Implement PowerPC/eqv-andc-orc-nor.ll:EQV3 by Chris Lattner · 20 years ago
- d135fa4 These nodes are all autogenerated by Chris Lattner · 20 years ago
- 333bd83 Make sure to clear the CodeGenMap after each basic block is selected to avoid by Chris Lattner · 20 years ago
- 303b555 we don't need this proto any longer by Chris Lattner · 20 years ago
- af16538 move the #include for the generated code into the isel class body so we by Chris Lattner · 20 years ago
- e6ec9f2 PowerPC cannot truncstore i1 natively by Chris Lattner · 20 years ago
- 19c0907 Remove some cases handled by the generated portion of the isel by Chris Lattner · 20 years ago
- c09eeec Implement i64<->fp using the fctidz/fcfid instructions on PowerPC when we by Nate Begeman · 20 years ago
- 6cd40d5 include the dag isel fragment by Chris Lattner · 20 years ago
- 25dae72 Change the isel to not break out of the big giant switch. Instead, the by Chris Lattner · 20 years ago
- 75592e4 Implement dynamic allocas correctly. In particular, because we were copying by Chris Lattner · 20 years ago
- 393ecd6 Fix a bug where we were useing HA to get the high part, which seems like it by Chris Lattner · 20 years ago
- 50ff55c Do not select the operands being passed into SelectCC. IT does this itself by Chris Lattner · 20 years ago
- f760532 Move FCTIWZ handling out of the instruction selectors and into legalization, by Chris Lattner · 20 years ago
- 8346bb6 Remove dead code by Chris Lattner · 20 years ago
- 99296ff add assert zext/sext to the dag isel by Chris Lattner · 20 years ago
- 7a49fdc Fix 'ret long' to return the high and lo parts in the right registers. This by Chris Lattner · 20 years ago
- eb80fe8 now that physregs can exist in the same dag with multiple types, remove some by Chris Lattner · 20 years ago
- 2ea0c66 Fix type mismatches when passing f32 values to calls by Chris Lattner · 20 years ago
- 1368721 Fix some indentation (first hunks). by Chris Lattner · 20 years ago
- b551ba7 Fix a problem Nate found where we swapped the operands of SHL/SHR_PARTS. This by Chris Lattner · 20 years ago
- 14b86c7 codegen ADD_PARTS correctly: put the results in the right registers! This by Chris Lattner · 20 years ago
- 2501d5e add operands in the right order, fixing McCat/18-imp with the dag isel by Chris Lattner · 20 years ago
- 31ce12f Make sure the selector emits register register copies with flag operands by Chris Lattner · 20 years ago
- 1505573 The first operand to AND does not always have more than two operands. This by Chris Lattner · 20 years ago
- 8f83872 emit FMR instructions to convert f64<->f32 instructions, so things like by Chris Lattner · 20 years ago
- 8bbcc20 fix a crash in cfrac by Chris Lattner · 20 years ago
- 9c2dece Implement DYNAMIC_STACKALLOC, wrap some long lines by Chris Lattner · 20 years ago
- 7107c10 Fix a dumb bug of mine where we were mishandling the PPC ABI (undef handling). by Chris Lattner · 20 years ago
- efa6abc Fix a bug the last patch exposed in treeadd among others by Chris Lattner · 20 years ago
- 2fef809 A hack to fix a problem folding immedaites. This fixes Olden/power. by Chris Lattner · 20 years ago
- 2a06a5e Fix order of operands for copytoreg node when emitting calls. This fixes by Chris Lattner · 20 years ago
- b9efd14 add operands in the correct order by Chris Lattner · 20 years ago
- c8a89a1 Fix a bug in FP_EXTEND, implement FP_TO_SINT by Chris Lattner · 20 years ago
- 528f58e fix an assertion failure in treeadd by Chris Lattner · 20 years ago
- 8a2d3ca implement SELECT_CC fully for the DAG->DAG isel! by Chris Lattner · 20 years ago
- 0bbea95 Make fsel emission work with both the pattern and dag-dag selectors, by by Chris Lattner · 20 years ago
- 13794f5 implement the fold for: by Chris Lattner · 20 years ago
- 52987f4 Changes to adjust to new ReplaceAllUsesWith syntax. Change FP_EXTEND to by Chris Lattner · 20 years ago
- 28b9be3 now that fsel is formed during legalization, this code is dead by Chris Lattner · 20 years ago
- 5839bf2 Change ConstantPoolSDNode to actually hold the Constant itself instead of by Chris Lattner · 20 years ago
- 95e0682 Fix some warnings in an optimized build by Chris Lattner · 20 years ago
- 2bb06cd Fix a huge annoyance: SelectNodeTo took types before the opcode unlike by Chris Lattner · 20 years ago
- bb22df3 SUBFIC produces two results, not one. by Nate Begeman · 20 years ago
- 6660cd6 Implement SHL_PARTS and SRL_PARTS by Nate Begeman · 20 years ago
- b20c318 Emit the lo/hi parts in the right order :) by Chris Lattner · 20 years ago
- a9317ed implement support for 64-bit add/sub, fix a broken assertion for 64-bit by Chris Lattner · 20 years ago
- 047b952 Finish implementing SDIV/UDIV by copying over the majik constant code from by Chris Lattner · 20 years ago
- 957fcfb Implement setcc correctly for G5 and non-G5 systems by Chris Lattner · 20 years ago
- 64906a0 implement setcc on the G5. We're still missing the non-g5 specific bits, but by Chris Lattner · 20 years ago
- 8784a23 Add support for sdiv by 2^k and -2^k. Producing code like: by Chris Lattner · 20 years ago