summaryrefslogtreecommitdiffstats
path: root/llvm/lib/CodeGen
Commit message (Collapse)AuthorAgeFilesLines
* Fix another bug legalizing calls!Chris Lattner2005-01-071-1/+1
| | | | llvm-svn: 19350
* Fix handling of dead PHI nodes.Chris Lattner2005-01-071-26/+30
| | | | llvm-svn: 19349
* Fix a bug legalizing callsChris Lattner2005-01-071-1/+1
| | | | llvm-svn: 19348
* After legalizing a DAG, delete dead nodes to save space.Chris Lattner2005-01-071-6/+1
| | | | llvm-svn: 19346
* Implement RemoveDeadNodesChris Lattner2005-01-071-0/+119
| | | | llvm-svn: 19345
* Teach legalize how to handle condbranchesChris Lattner2005-01-071-0/+10
| | | | llvm-svn: 19339
* Initial implementation of the SelectionDAGISel class. This contains mostChris Lattner2005-01-071-0/+853
| | | | | | of the code for lowering from LLVM code to a SelectionDAG. llvm-svn: 19331
* This file is obsoleteChris Lattner2005-01-071-278/+0
| | | | llvm-svn: 19330
* Initial implementation of the DAG legalization. This still has a long wayChris Lattner2005-01-071-0/+703
| | | | | | to go, but it does work for some non-trivial cases now. llvm-svn: 19329
* Complete rewrite of the SelectionDAG class.Chris Lattner2005-01-071-106/+758
| | | | llvm-svn: 19327
* Make the 2-address instruction lowering pass smarter in two ways:Chris Lattner2005-01-021-17/+61
| | | | | | | | | | | | | | | | | 1. If we are two-addressing a commutable instruction and the LHS is not the last use of the variable, see if the instruction is the last use of the RHS. If so, commute the instruction, allowing us to avoid a register-register copy in many cases for common instructions like ADD, OR, AND, etc on X86. 2. If #1 doesn't hold, and if this is an instruction that also existing in 3-address form, promote the instruction to a 3-address instruction to avoid the register-register copy. We can do this for several common instructions in X86, including ADDrr, INC, DEC, etc. This patch implements test/Regression/CodeGen/X86/commute-two-addr.ll, overlap-add.ll, and overlap-shift.ll when I check in the X86 support for it. llvm-svn: 19245
* Move virtual method call out of loopChris Lattner2004-12-152-6/+6
| | | | llvm-svn: 18955
* Move virtual function call out of loop to speed up getFreePhysReg by aboutNate Begeman2004-12-071-2/+2
| | | | | | 20%, shaving 0.1s off hbd compile time on my g5. Yay. llvm-svn: 18592
* For PR387:\Reid Spencer2004-12-072-2/+2
| | | | | | | Make only one print method to avoid overloaded virtual warnings when \ compiled with -Woverloaded-virtual llvm-svn: 18589
* Prevent accessing past the end of the intervals vector, this fixesChris Lattner2004-12-041-1/+3
| | | | | | Prolang-C/bison in the JIT llvm-svn: 18477
* Fix SingleSource/UnitTests/2004-11-28-GlobalBoolLayout.c, and hopefullyChris Lattner2004-11-281-1/+1
| | | | | | PR449 llvm-svn: 18306
* Fix the FIXME, nuke the JIT specific forceCompilationOf method.Chris Lattner2004-11-221-18/+0
| | | | llvm-svn: 18131
* These methods are obsoleteChris Lattner2004-11-221-8/+0
| | | | llvm-svn: 18129
* Adjust to changed interfacesChris Lattner2004-11-201-7/+7
| | | | llvm-svn: 18064
* Add getCurrentPCOffset() and addRelocation() methods.Chris Lattner2004-11-201-0/+12
| | | | llvm-svn: 18034
* Match change in MachineCodeEmitter prototype.Chris Lattner2004-11-191-2/+2
| | | | llvm-svn: 18009
* * There is no reason for SpillWeights to be an instance varChris Lattner2004-11-181-69/+71
| | | | | | | | * Do not put fixed registers into the unhandled set. This means they will never find their way into the inactive, active, or handled sets, so we can simplify a bunch of code. llvm-svn: 17945
* There is no need to check to see if j overflowed in this loop as we're onlyChris Lattner2004-11-181-1/+3
| | | | | | incrementing i. llvm-svn: 17944
* Moderate head scratching reveals that this conditional is not needed. IfChris Lattner2004-11-181-4/+0
| | | | | | i->start == j->start, then certainly i->end > j->start. llvm-svn: 17943
* Fix a couple of bugs where we considered physregs past their range as possiblyChris Lattner2004-11-181-6/+11
| | | | | | intersecting an interval. llvm-svn: 17939
* Fix typeoChris Lattner2004-11-181-1/+1
| | | | llvm-svn: 17938
* Start using the iterators in the fixed_ intervals to avoid having to binaryChris Lattner2004-11-181-5/+12
| | | | | | | search physreg intervals every time we access it. This takes another half second off of linscan. llvm-svn: 17937
* Take another .7 seconds off of linear scan time.Chris Lattner2004-11-181-3/+5
| | | | llvm-svn: 17936
* Add a counter for the number of times linscan has to backtrack. Start usingChris Lattner2004-11-181-22/+36
| | | | | | | the iterator hints we have to speed up overlaps(). This speeds linscan up by about .2s (out of 8.7) on 175.vpr for PPC. llvm-svn: 17935
* Add ability to give hints to the overlaps routines.Chris Lattner2004-11-182-7/+27
| | | | llvm-svn: 17934
* * Improve comments/documentation substantiallyChris Lattner2004-11-181-128/+172
| | | | | | | | | | | | | | | | | | * Eliminate the releaseMemory method, this is not an analysis * Change the fixed, active, and inactive lists of intervals to maintain an iterator for the current position in the interval. This allows us to do constant time increments of the iterator instead of having to do a binary search to find our liverange in our liveinterval all of the time, which substantially speeds up cases where LiveIntervals have many LiveRanges - which is very common for physical registers. On targets with many physregs, this can make a noticable difference. With a release build of LLC for PPC, this halves the time in processInactiveIntervals and processActiveIntervals, from 1.5s to .75s. This also lays the ground for more to come. llvm-svn: 17933
* Add new advanceTo methodChris Lattner2004-11-181-0/+13
| | | | llvm-svn: 17932
* Fix a minor bug in expiredAt. endNumber() is the first number that is not ↵Chris Lattner2004-11-181-1/+1
| | | | | | valid. llvm-svn: 17931
* Rename some methods, use 'begin' instead of 'start', add new LiveIntervalChris Lattner2004-11-183-22/+29
| | | | | | iterator/begin/end members. llvm-svn: 17930
* Give a better message for a common assertion failure.Brian Gaeke2004-11-161-1/+2
| | | | llvm-svn: 17887
* Do not make i have bigger scope that we needChris Lattner2004-11-051-2/+2
| | | | llvm-svn: 17483
* Change Library Names Not To Conflict With Others When InstalledReid Spencer2004-10-272-2/+2
| | | | llvm-svn: 17286
* Move method bodies that depend on <algorithm> from MBB.h to MBB.cppChris Lattner2004-10-261-14/+38
| | | | llvm-svn: 17253
* Clean up the MachineBasicBlock.h file, percolating #includes into this file.Chris Lattner2004-10-262-0/+2
| | | | | | Patch contributed by Morten Ofstad llvm-svn: 17251
* Reduce usage of MRegisterInfo::getRegClassChris Lattner2004-10-261-5/+6
| | | | llvm-svn: 17238
* Do not use variable sized arrays in C++, they are non-portable. PatchChris Lattner2004-10-251-5/+5
| | | | | | contributed by Morten Ofstad llvm-svn: 17217
* Patch to support MSVC better, contributed by Morten OfstadChris Lattner2004-10-251-1/+1
| | | | llvm-svn: 17215
* We won't use automakeReid Spencer2004-10-224-1520/+0
| | | | llvm-svn: 17155
* Initial automake generated Makefile templateReid Spencer2004-10-182-0/+1466
| | | | llvm-svn: 17136
* Add support for undefChris Lattner2004-10-161-2/+2
| | | | llvm-svn: 17055
* Allow machine operands to represent global variables with offsets. This isChris Lattner2004-10-151-5/+9
| | | | | | | | | | | | | | | | | useful when you have a reference like: int A[100]; void foo() { A[10] = 1; } In this case, &A[10] is a single constant and should be treated as such. Only MO_GlobalAddress and MO_ExternalSymbol are allowed to use this field, no other operand type is. This is another fine patch contributed by Jeff Cohen!! llvm-svn: 17007
* This patch fixes the nasty bug that caused 175.vpr to fail for X86 last night.Chris Lattner2004-10-151-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | The problem occurred when trying to reload this instruction: MOV32mr %reg2326, 8, %reg2297, 4, %reg2295 The value of reg2326 was available in EBX, so it was reused from there, instead of reloading it into EDX. The value of reg2297 was available in EDX, so it was reused from there, instead of reloading it into EDI. The value of reg2295 was not available, so we tried reloading it into EBX, its assigned register. However, we checked and saw that we already reloaded something into EBX, so we chose what reg2326 was assigned to (EDX) and reloaded into that register instead. Unfortunately EDX had already been used by reg2297, so reloading into EDX clobbered the value used by the reg2326 operand, breaking the program. The fix for this is to check that the newly picked register is ok. In this case we now find that EDX is already used and try using EDI, which succeeds. llvm-svn: 17006
* This patch adds and improves debugging output. No functionality changes.Chris Lattner2004-10-151-3/+7
| | | | llvm-svn: 17005
* Update to reflect changes in Makefile rules.Reid Spencer2004-10-132-8/+9
| | | | llvm-svn: 16950
* ModuloScheduling has moved to lib/Target/SparcV9Misha Brukman2004-10-102-2/+2
| | | | llvm-svn: 16906
OpenPOWER on IntegriCloud