summaryrefslogtreecommitdiffstats
path: root/llvm/lib/CodeGen
Commit message (Collapse)AuthorAgeFilesLines
* Change std::list into a std::vector for IntervalSets. This reducesAlkis Evlogimenos2004-07-211-4/+5
| | | | | | compile time for 176.gcc from 5.6 secs to 4.7 secs. llvm-svn: 15072
* Improve file comment.Alkis Evlogimenos2004-07-211-1/+7
| | | | llvm-svn: 15069
* Add Iterative scan register allocator.Alkis Evlogimenos2004-07-212-4/+479
| | | | llvm-svn: 15068
* Linearscan is no longer experimental.Alkis Evlogimenos2004-07-211-1/+1
| | | | llvm-svn: 15067
* Add function to clear all virtual->physical mappings but not assignedAlkis Evlogimenos2004-07-201-0/+5
| | | | | | stack slots. This is in preparation for the iterative linear scan. llvm-svn: 15032
* Remove unneeded functor. LiveInterval has a < operator.Alkis Evlogimenos2004-07-201-11/+1
| | | | llvm-svn: 15031
* Remove dead code.Alkis Evlogimenos2004-07-191-17/+0
| | | | llvm-svn: 15011
* Fix a bug that occurs when the last instruction in a range is deadChris Lattner2004-07-191-3/+6
| | | | llvm-svn: 15005
* When joining intervals, join intervals in deeply nested loops first. ThisChris Lattner2004-07-191-4/+34
| | | | | | | is a simple change, but seems to improve code a little. For example, on 256.bzip2, we went from 75.0s -> 73.33s (2% speedup). llvm-svn: 15004
* Split joinIntervals into two methodsChris Lattner2004-07-192-84/+89
| | | | llvm-svn: 15003
* There is no need to store the MBB along with the MI any more, we can nowChris Lattner2004-07-193-16/+16
| | | | | | ask instructions for their parent. llvm-svn: 14998
* Simplify the interface to LiveVariables::addVirtualRegister(Killed|Dead)Chris Lattner2004-07-194-9/+10
| | | | llvm-svn: 14997
* Remove the DefBlock element of VarInfo. DefBlock is always DefInst->getParent()Chris Lattner2004-07-191-4/+4
| | | | llvm-svn: 14996
* Two changes, both very significant:Chris Lattner2004-07-191-11/+2
| | | | | | | | | | * vreg <-> vreg joining now works, enable it unconditionally when joining is enabled (which is the default). * Fix a serious pessimization of spill code where we were saying that a spilled DEF operand was live into the subsequent instruction. This allows for substantially better code when spilling starts to happen. llvm-svn: 14993
* See comments. The live intervals were not coming out of the spiller in sortedChris Lattner2004-07-191-1/+20
| | | | | | | | | order, causing the inactive list in the linearscan list to get unsorted, which basically fuxored everything up severely. These seems to fix the joiner, so with more testing I will enable it by default. llvm-svn: 14992
* Fix assertion to not dereference end!Chris Lattner2004-07-191-2/+2
| | | | llvm-svn: 14991
* Add some asserts that the list of intervals returned by addIntervalsForSpillsChris Lattner2004-07-191-6/+30
| | | | | | | is sorted. This is not the case currently, which is causing no end of problems. llvm-svn: 14990
* remove the mbbi2mbbMap_, which was just keeping track of mbb order anyway.Chris Lattner2004-07-191-54/+73
| | | | | | | | | | Heavily refactor handleVirtualRegisterDef, adding comments and making it more efficient. It is also much easier to follow and convince ones self that it is correct :) Add -debug output to the joine, showing the result of joining the intervals. llvm-svn: 14989
* fill comment to 80 colsChris Lattner2004-07-191-11/+7
| | | | | | remove map that is not needed llvm-svn: 14988
* It looks like physref->virtreg joining is working fine. Enable it by defaultChris Lattner2004-07-171-10/+19
| | | | | | but make virtreg->virtreg joining stay off by default llvm-svn: 14916
* Fix IA64 compatibilityChris Lattner2004-07-162-2/+2
| | | | llvm-svn: 14866
* Use new macroChris Lattner2004-07-161-1/+1
| | | | llvm-svn: 14865
* Fix for PR341Chris Lattner2004-07-151-2/+2
| | | | llvm-svn: 14845
* Add checks to ensure that there are no unreachable blocks in the functionChris Lattner2004-07-091-0/+8
| | | | llvm-svn: 14725
* * Doxygenify commentsMisha Brukman2004-07-091-104/+96
| | | | | | | | | | | * Fix spacing, grammar in comment * Make code layout consistent * Wrap code at 80 cols * Delete spurious blank lines No functional changes. llvm-svn: 14721
* Fix typo.Alkis Evlogimenos2004-07-091-1/+1
| | | | llvm-svn: 14720
* Improve code comments.Alkis Evlogimenos2004-07-091-5/+11
| | | | llvm-svn: 14719
* Add viewCFG() and viewCFGOnly() APIs.Alkis Evlogimenos2004-07-081-1/+77
| | | | llvm-svn: 14679
* Disable coalescing.Alkis Evlogimenos2004-07-071-1/+1
| | | | llvm-svn: 14655
* Do not crash when joining two intervals of registers of differentAlkis Evlogimenos2004-07-061-1/+10
| | | | | | | classes: just ignore that move. Thanks to Vladimir Prus who found the bug! llvm-svn: 14644
* Fix a bug in the unreachable block elim pass. Dropping all references on aChris Lattner2004-07-061-4/+11
| | | | | | | | basic block clear()'s all of the operands lists, including phis. This caused removePredecessor to get confused later. Because of this, we just nuke (without prejudice) PHI nodes in unreachable blocks. llvm-svn: 14635
* Correctly compute the ration of iterations/#intervals.Alkis Evlogimenos2004-07-041-2/+6
| | | | llvm-svn: 14626
* Add #include <iostream> since Value.h does not #include it any more.Reid Spencer2004-07-0411-0/+16
| | | | llvm-svn: 14622
* Add efficiency statistic.Alkis Evlogimenos2004-07-041-1/+7
| | | | llvm-svn: 14590
* Revert patches 1.79 and 1.80 which had to do with dead MBB's. Now that theyChris Lattner2004-07-021-6/+4
| | | | | | don't exist, we don't have to pretend to handle them. llvm-svn: 14567
* Add a trivially simple pass to delete unreachable blocks from the CFG. ThisChris Lattner2004-07-021-0/+68
| | | | | | | | | | | | | pass is required to paper over problems in the code generator (primarily live variables and its clients) which doesn't really have any well defined semantics for unreachable code. The proper solution to this problem is to have instruction selectors not select blocks that are unreachable. Until we have a instruction selection framework available for use, however, we can't expect all instruction selector writers to do this. Until then, this pass should be used. llvm-svn: 14563
* Make sure that we destroy the MBB's, with all of their instructions, beforeChris Lattner2004-07-011-0/+1
| | | | | | any other data structures llvm-svn: 14524
* Now that MachineFunction/MachineBasicBlock keep a mapping of blocks to ID #'sChris Lattner2004-07-012-21/+1
| | | | | | use them instead of a local LiveVariables numbering llvm-svn: 14523
* Change MBB autonumber a bit to get the reverse mapping as well as a forwardChris Lattner2004-07-011-10/+6
| | | | | | mapping llvm-svn: 14521
* Instance var no longer existsChris Lattner2004-07-011-1/+1
| | | | llvm-svn: 14520
* Start using MBB numbers directly instead of going through the live variablesChris Lattner2004-07-013-4/+4
| | | | | | map. llvm-svn: 14518
* Instead of building a private numbering of MBB's use brg's nifty auto-numbering.Chris Lattner2004-07-011-5/+5
| | | | | | Also convert df_iterator -> df_ext_iterator for subsequent stuff I'm doing. llvm-svn: 14517
* In line with the previous patch, do not assert out if analyzing a dead basic ↵Chris Lattner2004-06-291-1/+4
| | | | | | block. llvm-svn: 14475
* Do not dereference end iterators. It hurts when you do that.Chris Lattner2004-06-291-3/+2
| | | | llvm-svn: 14474
* Do not find these ugly sparc-specific objects by using the annotation API onChris Lattner2004-06-271-21/+7
| | | | | | | | instructions. Instead, keep a map of instructions -> MCFI objects in the already sparc-specific class MachineFunctionInfo. This will slow down the sparc backend a bit, but it does not penalize the rest of LLVM! llvm-svn: 14438
* Made a fix so that you can print out MachineInstrs that belong to a ↵Tanya Lattner2004-06-256-19/+32
| | | | | | MachineBasicBlock that is not yet attached to a MachineFunction. This change includes changing the third operand (TargetMachine) to a pointer for the MachineInstr::print function. llvm-svn: 14389
* Convert tabs to spaces.Misha Brukman2004-06-241-45/+44
| | | | llvm-svn: 14373
* Making an archive version of the CodeGen library is unnecessary if we justMisha Brukman2004-06-221-1/+0
| | | | | | move InstructionLowering itself. llvm-svn: 14324
* Provide archive version of CodeGen library (for those tools that only needMisha Brukman2004-06-221-0/+1
| | | | | | IntrinsicLowering, for instance). llvm-svn: 14323
* Rename Interval class to LiveInterval to avoid conflicting with the alreadyChris Lattner2004-06-213-50/+54
| | | | | | | | | existing llvm::Interval class. Patch contributed by Vladimir Prus! http://mail.cs.uiuc.edu/pipermail/llvmbugs/2004-June/000710.html llvm-svn: 14281
OpenPOWER on IntegriCloud