- 36019aa Make the AND elimination operation recursive and significantly more powerful, by Chris Lattner · 20 years ago
- 16ac709 Change codegen for setcc to read the bit directly out of the condition by Nate Begeman · 20 years ago
- d91ff7c Add support for targets that require stubs for external functions. by Chris Lattner · 20 years ago
- 477d1de Handle ExternalSymbol operands in the PPC JIT by Chris Lattner · 20 years ago
- f8b0294 Make pattern isel default for ppc by Nate Begeman · 20 years ago
- 36d1273 a new simple pass, which will be extended to be more useful in the future. by Chris Lattner · 20 years ago
- 1e0d9bd fix calls by Andrew Lenharth · 20 years ago
- 3ae1829 a 21264 fix, and fix the operator precidence on an and -> zap check (should fix hundreds of test cases by Andrew Lenharth · 20 years ago
- 1ce0c01 print negative 64 bit immediates as negative numbers, makes things a little by Duraid Madina · 20 years ago
- 3eb7150 oops, this stopped us turning movl r4=0xFFFFFFFF;; and rX, r4 into zxt4 by Duraid Madina · 20 years ago
- 7bfba7d Implement multi-way branches through logical ops on condition registers. by Nate Begeman · 20 years ago
- 72ea281 Add a couple missing transforms in getSetCC that were triggering assertions by Nate Begeman · 20 years ago
- c4ccc2d we have zextloads, not sextloads! by Duraid Madina · 20 years ago
- ef7288c Add the necessary support to codegen condition register logical ops with by Nate Begeman · 20 years ago
- 1b7f7fb Start allocating condition registers. Almost all explicit uses of CR0 are by Nate Begeman · 20 years ago
- 3664cef Implement the fold shift X, zext(Y) -> shift X, Y at the target level, by Nate Begeman · 20 years ago
- eea805e Disbale the broken fold of shift + sz[ext] for now by Nate Begeman · 20 years ago
- 97e001d fix an infinite loop by Chris Lattner · 20 years ago
- fd8d389 fix some serious miscompiles on ia64, alpha, and ppc by Chris Lattner · 20 years ago
- 51679c4 avoid work when possible, perhaps fix the problem nate and andrew are seeing by Chris Lattner · 20 years ago
- c24b537 WOW, function calls still seem to work after this. by Andrew Lenharth · 20 years ago
- 556c44e prepare for func call optimization by Andrew Lenharth · 20 years ago
- ed09502 * add the shladd instruction * fold left shifts of 1, 2, 3 or 4 bits into adds by Duraid Madina · 20 years ago
- 4f7cba5 add matches for SxADDL and company, as well as simplify the SxADDQ code by Andrew Lenharth · 20 years ago
- e9c35e7 Implement expansion of unsigned i64 -> FP. by Chris Lattner · 20 years ago
- c02780e * if ANDing with a constant of the form: by Duraid Madina · 20 years ago
- 483f22d added all flavors of zap for anding by Andrew Lenharth · 20 years ago
- ffe284c Make expansion of uint->fp cast assert out instead of infinitely recurse. by Chris Lattner · 20 years ago
- 5bf2686 Fix some mysteriously missing {}'s which cause the miscompilation of by Chris Lattner · 20 years ago
- e666fcf add back the optimization that Nate added for shl X, (zext_inreg y) by Chris Lattner · 20 years ago
- 8a8dbfa Oops, remove these too. by Chris Lattner · 20 years ago
- 519f40b remove one more occurance of this that snuck in by Chris Lattner · 20 years ago
- 2942e9c Remove support for ZERO_EXTEND_INREG. This pessimizes code, genering stuff by Chris Lattner · 20 years ago
- 6ac614a Remove special handling of ZERO_EXTEND_INREG. This pessimizes code, causing by Chris Lattner · 20 years ago
- c951d87 Elimate handling of ZERO_EXTEND_INREG. This causes the PPC backend to emit by Chris Lattner · 20 years ago
- 91302a1 Z_E_I is gone by Chris Lattner · 20 years ago
- 2399356 Instead of making ZERO_EXTEND_INREG nodes, use the helper method in by Chris Lattner · 20 years ago
- 0f2287b Remove all foldings of ZERO_EXTEND_INREG, moving them to work for AND nodes by Chris Lattner · 20 years ago
- db81eba Fold shift x, [sz]ext(y) -> shift x, y by Nate Begeman · 20 years ago
- b882752 Fold shift by size larger than type size to undef by Nate Begeman · 20 years ago
- 9765c25 Implement setcc op, -1 sequences by Nate Begeman · 20 years ago
- 1c51c6a promote extload i1 -> extload i8 by Chris Lattner · 20 years ago
- 3464547 add an argument to allow avoiding deleting phi nodes. by Chris Lattner · 20 years ago
- 151c80b Get rid of this for_each loop by Chris Lattner · 20 years ago
- e8fd25f * OK, after changing to use liveIn/liveOut instead of IDEFs, by Duraid Madina · 20 years ago
- 0b04b5d Get rid of idefs for arguments (oops) by Andrew Lenharth · 20 years ago
- e1c5a00 Get rid of idefs for arguments by Andrew Lenharth · 20 years ago
- 9c24ba6 Put out* into the allocation order, allowing the register allocator to by Chris Lattner · 20 years ago
- ea6f770 Make sure to realize that calls use their argument regs by Chris Lattner · 20 years ago
- ca494fd stop emitting IDEFs for args - change to using liveIn/liveOut by Duraid Madina · 20 years ago
- 7af0248 Initial support for allocation condition registers by Nate Begeman · 20 years ago
- af2e3e0 Fix a crash analyzing MultiSource/Benchmarks/MallocBench/gs by Chris Lattner · 20 years ago
- fd1f1ee Remove some redundant checks, add a couple of new ones. This allows us to by Chris Lattner · 20 years ago
- 5b95ed6 Emit comparisons against the sign bit better. Codegen this: by Chris Lattner · 20 years ago
- 08b698e Emit long comparison against -1 better. Instead of this (x86): by Chris Lattner · 20 years ago
- 3b2c1d9 canonicalize x <u 1 -> x == 0. On this testcase: by Chris Lattner · 20 years ago
- 9f833d3 Implement bitfield clears Implement divide by negative power of two by Nate Begeman · 20 years ago
- ef9531e Update PPC readme. Remove things that are done or aren't ppc specific by Nate Begeman · 20 years ago
- bd0781e Teach the dag mechanism that this: by Chris Lattner · 20 years ago
- ee27f57 Fix expansion of shifts by exactly NVT bits on arch's (like X86) that have by Chris Lattner · 20 years ago
- 30e8243 IA64 supports this operation. by Chris Lattner · 20 years ago
- 5eef9f3 ORo sets CR0 by Chris Lattner · 20 years ago
- 6b4ea2c Revert the previous patch, which I didn't mean to check in. by Chris Lattner · 20 years ago
- 26d4fdb Fix a minor bug (ORo didn't mark that it set CR0). by Chris Lattner · 20 years ago
- 21478e5 hmm, should probably change addImm() to take 64-bit arguments one day anyway. by Duraid Madina · 20 years ago
- c7bd482 Add recording variants of ISD::AND and ISD::OR. This kills almost 1000 by Nate Begeman · 20 years ago
- 5ef2ec9 assorted fixes: by Duraid Madina · 20 years ago
- 01c5513 Eliminate tabs by Jeff Cohen · 20 years ago
- c7c1657 Fix libcall code to not pass a NULL Chain to LowerCallTo by Nate Begeman · 20 years ago
- 2bb6f41 Don't bother sign/zext_inreg'ing the result of an and operation if we know by Chris Lattner · 20 years ago
- 01ff721 Teach legalize to deal with targets that don't support some SEXTLOAD/ZEXTLOADs by Chris Lattner · 20 years ago
- 6841dec don't zextload fp values! by Chris Lattner · 20 years ago
- 709c806 Fix another fixme: factor out the constant fp generation code. by Nate Begeman · 20 years ago
- c5b1cd2 Fix 64 bit argument loading that straddles the args in regs / args on stack by Nate Begeman · 20 years ago
- 232ee95 Until we have a dag combiner, promote using zextload's instead of extloads. by Chris Lattner · 20 years ago
- 45278e3 Fold zext_inreg(zextload), likewise for sext's by Chris Lattner · 20 years ago
- edeecfc add a simple xform by Chris Lattner · 20 years ago
- 340f290 Remove unnecessary Implicit Defs. Since r0 is not in allocation, we do not by Nate Begeman · 20 years ago
- 27499e3 Make sure that BRCOND branches can be converted into long branches too. by Nate Begeman · 20 years ago
- a0e3e94 Don't hand ISD::CALL nodes off to SelectExprFP. This fixes siod. by Nate Begeman · 20 years ago
- dea29e2 Fix a thinko. If the operand is promoted, pass the promoted value into by Chris Lattner · 20 years ago
- 51d2ed9 rename getPPCOpcodeForSetCCNumber -> getPPCOpcodeForSetCCOpode to be more by Chris Lattner · 20 years ago
- 706471e fix ISD::BRCONDTWOWAY codegen to not deference the end() iterator by Nate Begeman · 20 years ago
- 9184bfb Fix CodeGen/Generic/2005-05-09-GlobalInPHI.ll, which was reduced from 254.gap. by Chris Lattner · 20 years ago
- a2daa8c add a little peephole optimization. This allows us to codegen: by Chris Lattner · 20 years ago
- 91277ea do not set the root to null if an argument is dead by Chris Lattner · 20 years ago
- cd08e4c Add rlwnm instruction for variable rotate by Nate Begeman · 20 years ago
- 27ee3a3 Fix a crash on 173.applu by asking for a constant bigger than 32-bits. by Chris Lattner · 20 years ago
- f429a3e Switch this instruction selector over to using liveins and liveouts, eliminating by Chris Lattner · 20 years ago
- 8e7d87b there is no need to remove this instruction, linscan does it already as it by Chris Lattner · 20 years ago
- 799a919 Adjust live intervals to support a livein set by Chris Lattner · 20 years ago
- 4c52f0e Use live out sets for return values instead of imp_defs, which is cleaner and faster. by Chris Lattner · 20 years ago
- d493b34 Consider the livein/out set for a function, allowing targets to not have to by Chris Lattner · 20 years ago
- e00e5de ok, the "ia64 has a boatload of registers" joke stopped being funny today ;) by Duraid Madina · 20 years ago
- af4ab1b Optimize FSEL a bit for fneg arguments. This fixes the recently added test by Nate Begeman · 20 years ago
- 67ac5f4 Fix CodeGen/SparcV9/2005-05-09-GEP-Crash.ll a crash on some specfp program by Chris Lattner · 20 years ago
- 59723e9 recognize some patterns as fabs operations, so that fabs at the source level by Chris Lattner · 20 years ago
- e7ccd4a Emit BRCONDTWOWAY when possible. by Chris Lattner · 20 years ago
- 411e888 Legalize BRCONDTWOWAY into a BRCOND/BR pair if a target doesn't support it. by Chris Lattner · 20 years ago
- ef847df print and fold BRCONDTWOWAY correctly by Chris Lattner · 20 years ago