| Commit message (Collapse) | Author | Age | Files | Lines | |
|---|---|---|---|---|---|
| * | Use df_ext_iterator to capture the reachable set without allocating an extra ↵ | Owen Anderson | 2008-07-02 | 1 | -17/+14 |
| | | | | | | | | | | set. Also, move large sets and vectors out of instance variables and onto the stack, and give them more reasonable sizes. llvm-svn: 53044 | ||||
| * | Avoid a redundant call. | Owen Anderson | 2008-07-02 | 1 | -4/+4 |
| | | | | | llvm-svn: 53040 | ||||
| * | Add support to ADCE for pruning unreachable blocks. This addresses the final | Owen Anderson | 2008-07-02 | 1 | -1/+43 |
| | | | | | | | part of PR2509. llvm-svn: 53038 | ||||
| * | Add a new getMergeValues method that does not need | Duncan Sands | 2008-07-02 | 8 | -83/+57 |
| | | | | | | | | | | | to be passed the list of value types, and use this where appropriate. Inappropriate places are where the value type list is already known and may be long, in which case the existing method is more efficient. llvm-svn: 53035 | ||||
| * | Use DenseSet rather than SmallPtrSet for the alive set. Using SmallPtrSet | Owen Anderson | 2008-07-02 | 1 | -2/+2 |
| | | | | | | | with a huge "size" parameter is actually quite inefficient. llvm-svn: 53034 | ||||
| * | A better fix for PR2503 that doesn't pessimize GVN in the presence of ↵ | Owen Anderson | 2008-07-02 | 2 | -13/+5 |
| | | | | | | | unreachable blocks. llvm-svn: 53032 | ||||
| * | Fixed problem in EmitStackConvert where the source and target type | Mon P Wang | 2008-07-02 | 2 | -12/+19 |
| | | | | | | | | have different alignment by creating a stack slot with the max alignment of source and target type. llvm-svn: 53031 | ||||
| * | instead of aborting on shifts of i1, just implicitly fold them. | Chris Lattner | 2008-07-02 | 1 | -1/+7 |
| | | | | | | | The dag combiner can produce a shift of i1 when folding icmp i1's. llvm-svn: 53030 | ||||
| * | Fix typo compounded by a cut-and-pasto. | Duncan Sands | 2008-07-02 | 1 | -4/+4 |
| | | | | | llvm-svn: 53012 | ||||
| * | Let AnalyzeNewNode take care of calling ExpungeNode. | Duncan Sands | 2008-07-02 | 2 | -86/+79 |
| | | | | | | | | This makes sure that all new nodes are expunged, not just those the top node of a new subtree. llvm-svn: 53011 | ||||
| * | - Use a faster priority comparison function if -fast. | Evan Cheng | 2008-07-02 | 2 | -228/+304 |
| | | | | | | | - Code clean up. llvm-svn: 53010 | ||||
| * | Add a new (simple) StringMap::clear method, patch by Pratik | Chris Lattner | 2008-07-02 | 1 | -4/+1 |
| | | | | | | | Solanki! llvm-svn: 53008 | ||||
| * | reuse vectors. | Devang Patel | 2008-07-02 | 1 | -5/+9 |
| | | | | | llvm-svn: 53007 | ||||
| * | Fix comment. | Devang Patel | 2008-07-02 | 1 | -1/+1 |
| | | | | | llvm-svn: 53006 | ||||
| * | Preserve loop data so that it is not fetched everytime it is needed. | Devang Patel | 2008-07-02 | 1 | -69/+85 |
| | | | | | | | Keep track of currentLoop. llvm-svn: 53005 | ||||
| * | Use the canonical form for getting an empty structure. | Bill Wendling | 2008-07-02 | 1 | -10/+5 |
| | | | | | llvm-svn: 53003 | ||||
| * | Sorry. I couldn't sleep at night knowing I put these ugly casts into the ↵ | Bill Wendling | 2008-07-02 | 2 | -6/+5 |
| | | | | | | | source tree. llvm-svn: 53001 | ||||
| * | Add a version of AddString that takes a const char* so we can avoid extraneous | Owen Anderson | 2008-07-01 | 1 | -0/+38 |
| | | | | | | | conversions to std::string. llvm-svn: 52995 | ||||
| * | Darwin doesn't need exception handling information for the "move" info when | Bill Wendling | 2008-07-01 | 4 | -2/+5 |
| | | | | | | | debug information is being output, because it's leet! llvm-svn: 52994 | ||||
| * | Avoid creating expensive comment string if it's not going to be printed. | Evan Cheng | 2008-07-01 | 3 | -25/+38 |
| | | | | | llvm-svn: 52992 | ||||
| * | No need to use std::distance. We can just count the number of operands | Owen Anderson | 2008-07-01 | 1 | -1/+4 |
| | | | | | | | much more cheaply. llvm-svn: 52990 | ||||
| * | Simplify addRegisterKilled and addRegisterDead. | Evan Cheng | 2008-07-01 | 1 | -18/+18 |
| | | | | | llvm-svn: 52988 | ||||
| * | - Update comments. | Bill Wendling | 2008-07-01 | 2 | -36/+25 |
| | | | | | | | - Don't use GlobalVariable::LinkageTypes when unsigned works. llvm-svn: 52987 | ||||
| * | Fix longstanding thinko: don't exclude | Dale Johannesen | 2008-07-01 | 1 | -2/+1 |
| | | | | | | | | predessors of exit blocks from tail merging consideration. llvm-svn: 52985 | ||||
| * | Disable dom info verifier by default. | Devang Patel | 2008-07-01 | 1 | -1/+1 |
| | | | | | llvm-svn: 52983 | ||||
| * | Eliminate a compile time warning. | Evan Cheng | 2008-07-01 | 1 | -1/+1 |
| | | | | | llvm-svn: 52982 | ||||
| * | Add a version of AsmPrinter::EOL that takes a const char* so that we don't ↵ | Owen Anderson | 2008-07-01 | 1 | -0/+11 |
| | | | | | | | | | have to do as many implicit std::string constructions. Unfortunately, this doesn't appear to translate to a real speedup in practice. llvm-svn: 52981 | ||||
| * | Remove warning about initialization order. | Bill Wendling | 2008-07-01 | 1 | -2/+2 |
| | | | | | llvm-svn: 52980 | ||||
| * | Fix typos in comments. | Devang Patel | 2008-07-01 | 2 | -6/+6 |
| | | | | | | | Thanks for the feedback! llvm-svn: 52978 | ||||
| * | Prune a few dependencies on MachineFunction.h. | Dan Gohman | 2008-07-01 | 2 | -1/+1 |
| | | | | | llvm-svn: 52976 | ||||
| * | Do run ComputeLiveOutVRegInfo with -fast. | Evan Cheng | 2008-07-01 | 1 | -1/+1 |
| | | | | | llvm-svn: 52975 | ||||
| * | Do not use computationally expensive scheduling heuristics with -fast. | Evan Cheng | 2008-07-01 | 4 | -56/+61 |
| | | | | | llvm-svn: 52971 | ||||
| * | Apply Chris' suggestion. | Evan Cheng | 2008-07-01 | 1 | -32/+36 |
| | | | | | llvm-svn: 52970 | ||||
| * | Add dom info verifier. | Devang Patel | 2008-07-01 | 2 | -1/+54 |
| | | | | | llvm-svn: 52967 | ||||
| * | Make the subregister hashtable output more readable by wrapping the lines, | Owen Anderson | 2008-07-01 | 1 | -2/+4 |
| | | | | | | | and mark it const along with the associated changes to TargetRegisterInfo. llvm-svn: 52966 | ||||
| * | Highlight that getMergeValues optimization is | Duncan Sands | 2008-07-01 | 1 | -1/+2 |
| | | | | | | | being suppressed here. llvm-svn: 52952 | ||||
| * | Properly handle cases where a predecessor of the block being queried on is ↵ | Owen Anderson | 2008-07-01 | 1 | -0/+13 |
| | | | | | | | | | unreachable. This fixes PR2503, though we should also fix other passes not to emit this kind of code. llvm-svn: 52946 | ||||
| * | Replace the dynamically computed std::set lookup method for subregisters ↵ | Owen Anderson | 2008-07-01 | 1 | -10/+0 |
| | | | | | | | | | with a hashtable-based version that is computed by tblgen at the time LLVM is compiled. llvm-svn: 52945 | ||||
| * | Minimize duplicated code in AsmPrinter::printLabel. | Dan Gohman | 2008-07-01 | 1 | -2/+1 |
| | | | | | llvm-svn: 52944 | ||||
| * | Split ISD::LABEL into ISD::DBG_LABEL and ISD::EH_LABEL, eliminating | Dan Gohman | 2008-07-01 | 26 | -71/+119 |
| | | | | | | | | | | | | | | | | | the need for a flavor operand, and add a new SDNode subclass, LabelSDNode, for use with them to eliminate the need for a label id operand. Change instruction selection to let these label nodes through unmodified instead of creating copies of them. Teach the MachineInstr emitter how to emit a MachineInstr directly from an ISD label node. This avoids the need for allocating SDNodes for the label id and flavor value, as well as SDNodes for each of the post-isel label, label id, and label flavor. llvm-svn: 52943 | ||||
| * | Suppress compiler warning. | Evan Cheng | 2008-06-30 | 1 | -1/+1 |
| | | | | | llvm-svn: 52934 | ||||
| * | Don't run stack slot coloring if -fast. | Evan Cheng | 2008-06-30 | 1 | -2/+4 |
| | | | | | llvm-svn: 52933 | ||||
| * | Don't use ISD namespace opcodes for MachineInstrs. | Dan Gohman | 2008-06-30 | 1 | -2/+2 |
| | | | | | llvm-svn: 52932 | ||||
| * | Use a simpler but equivalent form of RecordSource. | Dan Gohman | 2008-06-30 | 1 | -2/+1 |
| | | | | | llvm-svn: 52931 | ||||
| * | Add timing report for various sub-passes under SelectionDAGISel. | Evan Cheng | 2008-06-30 | 1 | -0/+26 |
| | | | | | llvm-svn: 52930 | ||||
| * | std::ostream and std::string microoptimizations for asm printing. | Dan Gohman | 2008-06-30 | 2 | -137/+137 |
| | | | | | llvm-svn: 52929 | ||||
| * | Use reserve. | Dan Gohman | 2008-06-30 | 1 | -0/+1 |
| | | | | | | | | | | | SelectionDAG::allnodes_size is linear, but that doesn't appear to outweigh the benefit of reducing heap traffic. If it does become a problem, we should teach SelectionDAG to keep a count of how many nodes are live, because there are several other places where that information would be useful as well. llvm-svn: 52926 | ||||
| * | Update comments to new-style syntax. | Dan Gohman | 2008-06-30 | 1 | -3/+3 |
| | | | | | llvm-svn: 52925 | ||||
| * | Rename ISD::LOCATION to ISD::DBG_STOPPOINT to better reflect its | Dan Gohman | 2008-06-30 | 13 | -55/+58 |
| | | | | | | | | | | | | | | | | | | purpose, and give it a custom SDNode subclass so that it doesn't need to have line number, column number, filename string, and directory string, all existing as individual SDNodes to be the operands. This was the only user of ISD::STRING, StringSDNode, etc., so remove those and some associated code. This makes stop-points considerably easier to read in -view-legalize-dags output, and reduces overhead (creating new nodes and copying std::strings into them) on code containing debugging information. llvm-svn: 52924 | ||||
| * | Split scheduling from instruction selection. | Evan Cheng | 2008-06-30 | 10 | -65/+55 |
| | | | | | llvm-svn: 52923 | ||||

