summaryrefslogtreecommitdiffstats
path: root/llvm/lib/CodeGen/SplitKit.h
Commit message (Collapse)AuthorAgeFilesLines
...
* Stop using dead function.Jakob Stoklund Olesen2011-04-131-4/+0
| | | | llvm-svn: 129442
* SparseBitVector is SLOW.Jakob Stoklund Olesen2011-04-121-0/+3
| | | | | | | Use a Bitvector instead, we didn't need the smaller memory footprint anyway. This makes the greedy register allocator 10% faster. llvm-svn: 129390
* Create new intervals for isolated blocks during region splitting.Jakob Stoklund Olesen2011-04-121-0/+5
| | | | | | | | | This merges the behavior of splitSingleBlocks into splitAroundRegion, so the RS_Region and RS_Block register stages can be coalesced. That means the leftover intervals after region splitting go directly to spilling instead of a second pass of per-block splitting. llvm-svn: 129379
* Add SplitKit API to query and select the current interval being worked on.Jakob Stoklund Olesen2011-04-121-1/+9
| | | | | | This makes it possible to target multiple registers in one pass. llvm-svn: 129374
* Build the Hopfield network incrementally when splitting global live ranges.Jakob Stoklund Olesen2011-04-091-4/+9
| | | | | | | | | It is common for large live ranges to have few basic blocks with register uses and many live-through blocks without any uses. This approach grows the Hopfield network incrementally around the use blocks, completely avoiding checking interference for some through blocks. llvm-svn: 129188
* Analyze blocks with uses separately from live-through blocks without uses.Jakob Stoklund Olesen2011-04-061-5/+15
| | | | | | | | About 90% of the relevant blocks are live-through without uses, and the only information required about them is their number. This saves memory and enables later optimizations that need to look at only the use-blocks. llvm-svn: 128985
* Use std::unique instead of a SmallPtrSet to ensure unique instructions in ↵Jakob Stoklund Olesen2011-04-051-16/+0
| | | | | | | | | | | UseSlots. This allows us to always keep the smaller slot for an instruction which is what we want when a register has early clobber defines. Drop the UsingInstrs set and the UsingBlocks map. They are no longer needed. llvm-svn: 128886
* Stop precomputing last split points, query the SplitAnalysis cache on demand.Jakob Stoklund Olesen2011-04-051-2/+0
| | | | llvm-svn: 128875
* Cache the fairly expensive last split point computation and provide a fastJakob Stoklund Olesen2011-04-051-4/+18
| | | | | | | | | inlined path for the common case. Most basic blocks don't contain a call that may throw, so the last split point os simply the first terminator. llvm-svn: 128874
* Stop caching basic block index ranges now that SlotIndexes can keep up.Jakob Stoklund Olesen2011-04-041-2/+0
| | | | llvm-svn: 128821
* Delete leftover data members.Jakob Stoklund Olesen2011-04-041-4/+0
| | | | llvm-svn: 128820
* Rewrite instructions as part of ConnectedVNInfoEqClasses::Distribute.Jakob Stoklund Olesen2011-03-171-7/+0
| | | | llvm-svn: 127779
* Delete dead code after rematerializing.Jakob Stoklund Olesen2011-03-081-0/+3
| | | | | | | | LiveRangeEdit::eliminateDeadDefs() will eventually be used by coalescing, splitting, and spilling for dead code elimination. It can delete chains of dead instructions as long as there are no dependency loops. llvm-svn: 127287
* Work around a coalescer bug.Jakob Stoklund Olesen2011-03-051-1/+1
| | | | | | | | | | | | The coalescer can in very rare cases leave too large live intervals around after rematerializing cheap-as-a-move instructions. Linear scan doesn't really care, but live range splitting gets very confused when a live range is killed by a ghost instruction. I will fix this properly in the coalescer after 2.9 branches. llvm-svn: 127096
* Use an IndexedMap instead of a DenseMap for the live-out cache.Jakob Stoklund Olesen2011-03-041-1/+7
| | | | | | | This speeds up updateSSA() so it only accounts for 5% of the live range splitting time. llvm-svn: 126972
* Cache basic block bounds instead of asking SlotIndexes::getMBBRange all the ↵Jakob Stoklund Olesen2011-03-031-0/+2
| | | | | | | | | time. This speeds up the greedy register allocator by 15%. DenseMap is not as fast as one might hope. llvm-svn: 126921
* Change the SplitEditor interface to a single instance can be shared for ↵Jakob Stoklund Olesen2011-03-031-3/+3
| | | | | | multiple splits. llvm-svn: 126912
* Turn the Edit member into a pointer so it can change dynamically.Jakob Stoklund Olesen2011-03-021-1/+1
| | | | | | No functional change. llvm-svn: 126898
* Transfer simply defined values directly without recomputing liveness and SSA.Jakob Stoklund Olesen2011-03-021-1/+5
| | | | | | | | Values that map to a single new value in a new interval after splitting don't need new PHIDefs, and if the parent value was never rematerialized the live range will be the same. llvm-svn: 126894
* Extract a method. No functional change.Jakob Stoklund Olesen2011-03-021-0/+4
| | | | llvm-svn: 126893
* Move extendRange() into SplitEditor and delete the LiveRangeMap class.Jakob Stoklund Olesen2011-03-021-60/+32
| | | | | | | | | Extract the updateSSA() method from the too long extendRange(). LiveOutCache can be shared among all the new intervals since there is at most one of the new ranges live out from each basic block. llvm-svn: 126818
* Rename mapValue to extendRange because that is its function now.Jakob Stoklund Olesen2011-03-021-10/+3
| | | | | | Simplify the signature - The return value and ParentVNI are no longer needed. llvm-svn: 126809
* Move LiveIntervalMap::extendTo into LiveInterval itself.Jakob Stoklund Olesen2011-03-021-5/+0
| | | | | | | This method could probably be used by LiveIntervalAnalysis::shrinkToUses, and now it can use extendIntervalEndTo() which coalesces ranges. llvm-svn: 126803
* Delete dead code.Jakob Stoklund Olesen2011-03-011-9/+0
| | | | llvm-svn: 126801
* Move the value map from LiveIntervalMap to SplitEditor.Jakob Stoklund Olesen2011-03-011-28/+23
| | | | | | | | | | | The value map is currently not used, all values are 'complex mapped' and LiveIntervalMap::mapValue is used to dig them out. This is the first step in a series changes leading to the removal of LiveIntervalMap. Its data structures can be shared among all the live intervals created by a split, so it is wasteful to create a copy for each. llvm-svn: 126800
* Delete dead code.Jakob Stoklund Olesen2011-03-011-9/+0
| | | | | | | Local live range splitting is better driven by interference. This code was just guessing. llvm-svn: 126799
* Add SplitKit::isOriginalEndpoint and use it to force live range splitting to ↵Jakob Stoklund Olesen2011-02-211-0/+7
| | | | | | | | | | | | | | terminate. An original endpoint is an instruction that killed or defined the original live range before any live ranges were split. When splitting global live ranges, avoid creating local live ranges without any original endpoints. We may still create global live ranges without original endpoints, but such a range won't be split again, and live range splitting still terminates. llvm-svn: 126151
* Give SplitAnalysis a VRM member to access VirtRegMap::getOriginal().Jakob Stoklund Olesen2011-02-191-1/+2
| | | | llvm-svn: 126005
* Missed member rename for naming convention.Jakob Stoklund Olesen2011-02-191-2/+2
| | | | llvm-svn: 126003
* Split local live ranges.Jakob Stoklund Olesen2011-02-171-0/+3
| | | | | | | | A local live range is live in a single basic block. If such a range fails to allocate, try to find a sub-range that would get a larger spill weight than its interference. llvm-svn: 125764
* Delete unused code for analyzing and splitting around loops.Jakob Stoklund Olesen2011-02-091-74/+0
| | | | | | | Loop splitting is better handled by the more generic global region splitting based on the edge bundle graph. llvm-svn: 125243
* Use the LiveBLocks array for SplitEditor::splitSingleBlocks() as well.Jakob Stoklund Olesen2011-02-091-0/+4
| | | | | | | This fixes a bug where splitSingleBlocks() could split a live range after a terminator instruction. llvm-svn: 125237
* Move calcLiveBlockInfo() and the BlockInfo struct into SplitAnalysis.Jakob Stoklund Olesen2011-02-091-0/+35
| | | | | | No functional changes intended. llvm-svn: 125231
* Add SplitEditor::overlapIntv() to create small ranges where both registers ↵Jakob Stoklund Olesen2011-02-081-0/+12
| | | | | | | | | | | | | | | | | | | | | | are live. If a live range is used by a terminator instruction, and that live range needs to leave the block on the stack or in a different register, it can be necessary to have both sides of the split live at the terminator instruction. Example: %vreg2 = COPY %vreg1 JMP %vreg1 Becomes after spilling %vreg2: SPILL %vreg1 JMP %vreg1 The spill doesn't kill the register as is normally the case. llvm-svn: 125102
* Return live range end points from SplitEditor::enter*/leave*.Jakob Stoklund Olesen2011-02-031-9/+14
| | | | | | | These end points come from the inserted copies, and can be passed directly to useIntv. This simplifies the coloring code. llvm-svn: 124799
* Reapply this.Eric Christopher2011-02-031-27/+36
| | | | llvm-svn: 124779
* Temporarily revert 124765 in an attempt to find the cycle breaking bootstrap.Eric Christopher2011-02-031-36/+27
| | | | llvm-svn: 124778
* Defer SplitKit value mapping until all defs are available.Jakob Stoklund Olesen2011-02-031-27/+36
| | | | | | | | | | | | | | | | | | | The greedy register allocator revealed some problems with the value mapping in SplitKit. We would sometimes start mapping values before all defs were known, and that could change a value from a simple 1-1 mapping to a multi-def mapping that requires ssa update. The new approach collects all defs and register assignments first without filling in any live intervals. Only when finish() is called, do we compute liveness and mapped values. At this time we know with certainty which values map to multiple values in a split range. This also has the advantage that we can compute live ranges based on the remaining uses after rematerializing at split points. The current implementation has many opportunities for compile time optimization. llvm-svn: 124765
* Rename member variables to follow the rest of LLVM.Jakob Stoklund Olesen2011-01-261-76/+76
| | | | | | No functional change. llvm-svn: 124257
* Add LiveIntervalMap::dumpCache() to print out the cache used by the ssa ↵Jakob Stoklund Olesen2011-01-201-0/+3
| | | | | | update algorithm. llvm-svn: 123925
* Add RAGreedy methods for splitting live ranges around regions.Jakob Stoklund Olesen2011-01-181-0/+8
| | | | | | | | | | Analyze the live range's behavior entering and leaving basic blocks. Compute an interference pattern for each allocation candidate, and use SpillPlacement to find an optimal region where that register can be live. This code is still not enabled. llvm-svn: 123774
* Turn the EdgeBundles class into a stand-alone machine CFG analysis pass.Jakob Stoklund Olesen2011-01-041-37/+1
| | | | | | | | | | The analysis will be needed by both the greedy register allocator and the X86FloatingPoint pass. It only needs to be computed once when the CFG doesn't change. This pass is very fast, usually showing up as 0.0% wall time. llvm-svn: 122832
* Add EdgeBundles to SplitKit.Jakob Stoklund Olesen2010-12-211-1/+38
| | | | | | | | | Edge bundles is an annotation on the CFG that turns it into a bipartite directed graph where each basic block is connected to an outgoing and an ingoing bundle. These bundles are useful for identifying regions of the CFG for live range splitting. llvm-svn: 122301
* Detect and enumerate bypass loops.Jakob Stoklund Olesen2010-12-151-0/+9
| | | | | | | | Bypass loops have the current live range live through, but contain no uses or defs. Splitting around a bypass loop can free registers for other uses inside the loop by spilling the split range. llvm-svn: 121871
* Separate SplitAnalysis::getSplitLoops().Jakob Stoklund Olesen2010-12-151-0/+4
| | | | | | | This method returns the set of loops with uses that are candidates for splitting. llvm-svn: 121870
* Basic rematerialization during splitting.Jakob Stoklund Olesen2010-11-101-8/+10
| | | | | | | | | | | Whenever splitting wants to insert a copy, it checks if the value can be rematerialized cheaply instead. Missing features: - Delete instructions when all uses have been rematerialized. - Truncate live ranges to the remaining uses after rematerialization. llvm-svn: 118702
* When inserting copies during splitting, always use the parent register as theJakob Stoklund Olesen2010-11-011-6/+6
| | | | | | | | | | | | source, and let rewrite() clean it up. This way, kill flags on the inserted copies are fixed as well during rewrite(). We can't just assume that all the copies we insert are going to be kills since critical edges into loop headers sometimes require both source and dest to be live out of a block. llvm-svn: 117980
* Replace SplitKit SSA update with an iterative algorithm very similar to the oneJakob Stoklund Olesen2010-10-281-1/+23
| | | | | | | | | | | | | | | | in SSAUpdaterImpl.h Verifying live intervals revealed that the old method was completely wrong, and we need an iterative approach to calculating PHI placemant. Fortunately, we have MachineDominators available, so we don't have to compute that over and over like SSAUpdaterImpl.h must. Live-out values are cached between calls to mapValue() and computed in a greedy way, so most calls will be working with very small block sets. Thanks to Bob for explaining how this should work. llvm-svn: 117599
* Make MachineDominators available for SplitEditor. We are going to need it forJakob Stoklund Olesen2010-10-281-2/+6
| | | | | | | | | proper SSA updating. This doesn't cause MachineDominators to be recomputed since we are already requiring MachineLoopInfo which uses dominators as well. llvm-svn: 117598
* Handle critical loop predecessors by making both inside and outside registersJakob Stoklund Olesen2010-10-271-1/+5
| | | | | | | | | live out. This doesn't prevent us from inserting a loop preheader later on, if that is better. llvm-svn: 117424
OpenPOWER on IntegriCloud