| 2008-08-13 |
Owen Anderson | 1) Merge entire live intervals instead of parts of...
|
commit | commitdiff | tree |
| 2008-08-08 |
Owen Anderson | Reduce the entries in a phi before testing it for deadness...
|
commit | commitdiff | tree |
| 2008-08-07 |
Owen Anderson | Do a dominator walk when scheduling copies, rather...
|
commit | commitdiff | tree |
| 2008-08-06 |
Owen Anderson | SDISel's constant branch folding can fold away self...
|
commit | commitdiff | tree |
| 2008-08-06 |
Owen Anderson | Correct handle cases where two phis are coalesced together...
|
commit | commitdiff | tree |
| 2008-08-06 |
Owen Anderson | Oops, didn't mean to commit this.
|
commit | commitdiff | tree |
| 2008-08-06 |
Owen Anderson | We don't need to try to coalesce input vregs that are...
|
commit | commitdiff | tree |
| 2008-08-06 |
Owen Anderson | Only trim a live interval if the register is not used...
|
commit | commitdiff | tree |
| 2008-08-06 |
Owen Anderson | Only remap each VNInfo once when doing renumbering.
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Fix breakage on ARM/2008-04-10-ScavengerAssert.ll.
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Correctly handle replacement and removal of PHIs with...
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Oops, we were already checking for dead phis. Handle...
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | We don't need to update live intervals for dead PHIs.
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Remove the -disable-correct-folding option, which was...
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Update the remaining tests not to use -disable-correct...
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | One more -disable-correct-folding case removed.
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Remove another -disable-correct-folding use.
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Eliminate another use of -disable-correct-folding.
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | This check is unnecessary, and getting rid of it removes...
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Remove the need for -disable-correct-folding from this...
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | Remove unneeded iteration. Thanks to Dan for the feedback.
|
commit | commitdiff | tree |
| 2008-08-05 |
Owen Anderson | This option doesn't need to be a target option. It...
|
commit | commitdiff | tree |
| 2008-08-04 |
Owen Anderson | Update these tests to work by disabling the new correct...
|
commit | commitdiff | tree |
| 2008-08-04 |
Owen Anderson | - Fix SelectionDAG to generate correct CFGs.
|
commit | commitdiff | tree |
| 2008-07-30 |
Owen Anderson | Use existing LiveInterval methods to simplify live...
|
commit | commitdiff | tree |
| 2008-07-30 |
Owen Anderson | Value numbers whose def index is a special sentinel...
|
commit | commitdiff | tree |
| 2008-07-30 |
Owen Anderson | More fixes for corner cases when remapping live range...
|
commit | commitdiff | tree |
| 2008-07-30 |
Owen Anderson | When merging live intervals, we also need to merge...
|
commit | commitdiff | tree |
| 2008-07-29 |
Owen Anderson | When merging a PHI operand's live interval into the...
|
commit | commitdiff | tree |
| 2008-07-29 |
Owen Anderson | Don't decrement the BB remap when we don't need to.
|
commit | commitdiff | tree |
| 2008-07-28 |
Owen Anderson | Don't remove volatile loads. Thanks to Duncan for...
|
commit | commitdiff | tree |
| 2008-07-28 |
Owen Anderson | Add support for eliminating stores that store the same...
|
commit | commitdiff | tree |
| 2008-07-28 |
Owen Anderson | Fix a subtle bug when removing instructions from memdep...
|
commit | commitdiff | tree |
| 2008-07-25 |
Owen Anderson | Fix the issues originally addressed in r54070. After...
|
commit | commitdiff | tree |
| 2008-07-25 |
Owen Anderson | Revert my previous patch. In retrospect, this is completely...
|
commit | commitdiff | tree |
| 2008-07-25 |
Owen Anderson | Special cases are needed in renumbering when dealing...
|
commit | commitdiff | tree |
| 2008-07-25 |
Owen Anderson | In order to avoid reprocessing a register more than...
|
commit | commitdiff | tree |
| 2008-07-25 |
Owen Anderson | Remove live interval entries for an interval if we...
|
commit | commitdiff | tree |
| 2008-07-25 |
Owen Anderson | Properly remap live ranges whose end indices are the...
|
commit | commitdiff | tree |
| 2008-07-25 |
Owen Anderson | Make the remapping of interval indices (particularly...
|
commit | commitdiff | tree |
| 2008-07-24 |
Owen Anderson | Store the predecessor MBB in the PHIUnion, rather than...
|
commit | commitdiff | tree |
| 2008-07-23 |
Owen Anderson | Enable the insertion of empty indices into LiveInterals...
|
commit | commitdiff | tree |
| 2008-07-23 |
Owen Anderson | Fix a compile-time regression introduced by my heuristic...
|
commit | commitdiff | tree |
| 2008-07-22 |
Owen Anderson | Change the heuristics used in the coalescer, register...
|
commit | commitdiff | tree |
| 2008-07-18 |
Owen Anderson | Make PRE actually handle critical edges (by splitting...
|
commit | commitdiff | tree |
| 2008-07-18 |
Owen Anderson | Reapply r53735. My last patch fixed the failures Dan...
|
commit | commitdiff | tree |
| 2008-07-18 |
Owen Anderson | Add some checks that got lost in the shuffle. This...
|
commit | commitdiff | tree |
| 2008-07-17 |
Owen Anderson | Use MergeBlockIntoPredecessor to simplify some code.
|
commit | commitdiff | tree |
| 2008-07-17 |
Owen Anderson | Make MergeBlockIntoPredecessor more aggressive when...
|
commit | commitdiff | tree |
| 2008-07-17 |
Owen Anderson | Enable PRE. My last batch of changes fixed the miscompile.
|
commit | commitdiff | tree |
| 2008-07-17 |
Owen Anderson | Factor MergeBlockIntoPredecessor out into BasicBlockUtils.
|
commit | commitdiff | tree |
| 2008-07-16 |
Owen Anderson | There's no need to iterate block merging and PRE. ...
|
commit | commitdiff | tree |
| 2008-07-15 |
Owen Anderson | Revert this, as it seems to still be broken.
|
commit | commitdiff | tree |
| 2008-07-15 |
Owen Anderson | Enable local PRE by default.
|
commit | commitdiff | tree |
| 2008-07-15 |
Owen Anderson | Have GVN do a pre-pass over the CFG that folds away...
|
commit | commitdiff | tree |
| 2008-07-11 |
Owen Anderson | Don't call lookupNumber more than we have to.
|
commit | commitdiff | tree |
| 2008-07-10 |
Owen Anderson | Use DenseMap instead of std::map in local register...
|
commit | commitdiff | tree |
| 2008-07-10 |
Owen Anderson | Fix 403.gcc. Finally got the check for two-address...
|
commit | commitdiff | tree |
| 2008-07-10 |
Owen Anderson | Fix the build by adding a #include.
|
commit | commitdiff | tree |
| 2008-07-09 |
Owen Anderson | Revert r53367, which was breaking things.
|
commit | commitdiff | tree |
| 2008-07-09 |
Owen Anderson | Loosen our check here. Local regalloc only cares that...
|
commit | commitdiff | tree |
| 2008-07-09 |
Owen Anderson | Don't use an expensive check for two-address-ness when...
|
commit | commitdiff | tree |
| 2008-07-09 |
Owen Anderson | Factor local liveness computation out into its own...
|
commit | commitdiff | tree |
| 2008-07-08 |
Owen Anderson | Fix the build. Apparently MachineInstr& is no longer...
|
commit | commitdiff | tree |
| 2008-07-08 |
Owen Anderson | Make the local register allocator compute (purely local...
|
commit | commitdiff | tree |
| 2008-07-08 |
Owen Anderson | Global variables beginning with \01 have special meaning...
|
commit | commitdiff | tree |
| 2008-07-03 |
Owen Anderson | Use information already present in the ValueTable to...
|
commit | commitdiff | tree |
| 2008-07-03 |
Owen Anderson | Remove the ability for ADCE to remove unreachable blocks...
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | Make LiveVariables even more optional, by making it...
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | TwoAddressInstructionPass doesn't really require LiveVariabl...
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | Use ADCE instead of just DCE. ADCE will better clean...
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | Use df_ext_iterator to capture the reachable set without...
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | Avoid a redundant call.
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | Add support to ADCE for pruning unreachable blocks...
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | Use DenseSet rather than SmallPtrSet for the alive...
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | Have DenseSet::insert return a bool indicating whether...
|
commit | commitdiff | tree |
| 2008-07-02 |
Owen Anderson | A better fix for PR2503 that doesn't pessimize GVN...
|
commit | commitdiff | tree |
| 2008-07-01 |
Owen Anderson | Add a version of AddString that takes a const char...
|
commit | commitdiff | tree |
| 2008-07-01 |
Owen Anderson | No need to use std::distance. We can just count the...
|
commit | commitdiff | tree |
| 2008-07-01 |
Owen Anderson | Add a version of AsmPrinter::EOL that takes a const...
|
commit | commitdiff | tree |
| 2008-07-01 |
Owen Anderson | Make the subregister hashtable output more readable...
|
commit | commitdiff | tree |
| 2008-07-01 |
Owen Anderson | Use delete[] instead of free on an array created with...
|
commit | commitdiff | tree |
| 2008-07-01 |
Owen Anderson | Implement suggestions from Chris:
|
commit | commitdiff | tree |
| 2008-07-01 |
Owen Anderson | Properly handle cases where a predecessor of the block...
|
commit | commitdiff | tree |
| 2008-07-01 |
Owen Anderson | Replace the dynamically computed std::set lookup method...
|
commit | commitdiff | tree |
| 2008-06-30 |
Owen Anderson | Add an isReachableFromEntry method.
|
commit | commitdiff | tree |
| 2008-06-27 |
Owen Anderson | Add a NextPowerOf2 function to calculate the next power...
|
commit | commitdiff | tree |
| 2008-06-27 |
Owen Anderson | Use a SmallSet when we can to reduce memory allocations.
|
commit | commitdiff | tree |
| 2008-06-27 |
Owen Anderson | Cache subregister relationships in a set in TargetRegisterIn...
|
commit | commitdiff | tree |
| 2008-06-27 |
Owen Anderson | Don't perform expensive queries checking for super...
|
commit | commitdiff | tree |
| 2008-06-26 |
Owen Anderson | Use a DenseMap instead of an std::map for mangled name...
|
commit | commitdiff | tree |
| 2008-06-26 |
Owen Anderson | Don't create a whole new string just to copy the elements...
|
commit | commitdiff | tree |
| 2008-06-26 |
Owen Anderson | Use the -enable-pre flag so this test doesn't fail.
|
commit | commitdiff | tree |
| 2008-06-26 |
Owen Anderson | Reserve the size we'll need in advance.
|
commit | commitdiff | tree |
| 2008-06-25 |
Owen Anderson | Remember which MachineOperand we were processing, so...
|
commit | commitdiff | tree |
| 2008-06-25 |
Owen Anderson | Use push_back rather than operator[], which is incorrect...
|
commit | commitdiff | tree |
| 2008-06-24 |
Owen Anderson | In ConstantArray::getAsString(), we know the size of...
|
commit | commitdiff | tree |
| 2008-06-24 |
Owen Anderson | Use SmallVector instead of std::vector for a minor...
|
commit | commitdiff | tree |
| 2008-06-24 |
Owen Anderson | Remove an incorrect comment.
|
commit | commitdiff | tree |
| 2008-06-24 |
Owen Anderson | Use InstrSlots::NUM rather than pre-dividing by four...
|
commit | commitdiff | tree |
| next |
projects / oota-llvm.git / search