| 2010-03-05 |
Dale Johannesen | Fix a case where LSR is sensitive to debug info.
|
commit | commitdiff | tree |
| 2010-03-05 |
Dale Johannesen | Fix some more places where dbg_value affected codegen.
|
commit | commitdiff | tree |
| 2010-02-26 |
Dale Johannesen | Move dbg_value generation to target-independent FastISel,
|
commit | commitdiff | tree |
| 2010-02-26 |
Dale Johannesen | Further constify MDNode* references.
|
commit | commitdiff | tree |
| 2010-02-26 |
Dale Johannesen | Add type printing for Metadata pseudo.
|
commit | commitdiff | tree |
| 2010-02-19 |
Dale Johannesen | recommit 96626, evidence that it broke things appears
|
commit | commitdiff | tree |
| 2010-02-19 |
Dale Johannesen | Revert 96626, which causes build failure on ppc Darwin.
|
commit | commitdiff | tree |
| 2010-02-18 |
Dale Johannesen | Generate DBG_VALUE from dbg.value intrinsics. These...
|
commit | commitdiff | tree |
| 2010-02-16 |
Dale Johannesen | Make g5 target explicit; scheduling affects register...
|
commit | commitdiff | tree |
| 2010-02-16 |
Dale Johannesen | Adjust register numbers in tests to compensate for the
|
commit | commitdiff | tree |
| 2010-02-16 |
Dale Johannesen | Really reserve R2 on PPC Darwin. PR 6314.
|
commit | commitdiff | tree |
| 2010-02-16 |
Dale Johannesen | Handle DBG_VALUE mixed with labels when doing PHI
|
commit | commitdiff | tree |
| 2010-02-16 |
Dale Johannesen | Reapply 96294; now that I've gotten around to looking
|
commit | commitdiff | tree |
| 2010-02-16 |
Dale Johannesen | Something broke. Hard to believe it was this patch
|
commit | commitdiff | tree |
| 2010-02-15 |
Dale Johannesen | More handling of DBG_VALUE.
|
commit | commitdiff | tree |
| 2010-02-15 |
Dale Johannesen | Ignore DBG_VALUE in a couple more places.
|
commit | commitdiff | tree |
| 2010-02-12 |
Dale Johannesen | Add the problem I just hacked around in 96015/96020.
|
commit | commitdiff | tree |
| 2010-02-12 |
Dale Johannesen | This should have gone in with 26015, see comments there.
|
commit | commitdiff | tree |
| 2010-02-12 |
Dale Johannesen | When save/restoring CR at prolog/epilog, in a large
|
commit | commitdiff | tree |
| 2010-02-12 |
Dale Johannesen | Rewrite handling of DBG_VALUE; previous algorithm
|
commit | commitdiff | tree |
| 2010-02-11 |
Dale Johannesen | Allow for more than one DBG_VALUE targeting the
|
commit | commitdiff | tree |
| 2010-02-11 |
Dale Johannesen | Don't allow DBG_VALUE to affect codegen.
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | Ignore debug info one more place during coalescing.
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | Allow isDebug inquiry on any MO.
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | Skip debug info in a couple of places.
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | When I rewrote this loop per Chris' preference I
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | Rewrite loop to suit Chris' preference.
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | Skip DBG_VALUE many places in live intervals and
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | more comment updates
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | Add isDebug argument to ChangeToRegister; this prevents
|
commit | commitdiff | tree |
| 2010-02-10 |
Dale Johannesen | Fix comments to reflect renaming elsewhere.
|
commit | commitdiff | tree |
| 2010-02-09 |
Dale Johannesen | Disable unittests/ADT/BitVectorTest on PPC Darwin.
|
commit | commitdiff | tree |
| 2010-02-09 |
Dale Johannesen | Re-disable for Darwin; I was mistaken to think this...
|
commit | commitdiff | tree |
| 2010-02-09 |
Dale Johannesen | Skip DEBUG_VALUE in some places where it was affecting...
|
commit | commitdiff | tree |
| 2010-02-09 |
Dale Johannesen | Debug operands should not be def or kill.
|
commit | commitdiff | tree |
| 2010-02-08 |
Dale Johannesen | Apply the 95471 fix to SelectionDAGBuilder as well;
|
commit | commitdiff | tree |
| 2010-02-06 |
Dale Johannesen | Add a Debug bit to MachineOperand, for uses that
|
commit | commitdiff | tree |
| 2010-02-06 |
Dale Johannesen | After Victor's latest commits I am seeing null
|
commit | commitdiff | tree |
| 2010-02-04 |
Dale Johannesen | Rewrite FP constant handling in DEBUG_VALUE yet
|
commit | commitdiff | tree |
| 2010-02-03 |
Dale Johannesen | This test passes now on ppc darwin; if it doesn't pass
|
commit | commitdiff | tree |
| 2010-02-03 |
Dale Johannesen | This test passes now on ppc darwin, so reenable it.
|
commit | commitdiff | tree |
| 2010-02-03 |
Dale Johannesen | Debugging is now reenabled on PPC darwin, so reenable
|
commit | commitdiff | tree |
| 2010-02-03 |
Dale Johannesen | Reapply 95050 with a tweak to check the register class.
|
commit | commitdiff | tree |
| 2010-02-03 |
Dale Johannesen | Print FPImm a less kludgy way; APFloat.toString seems
|
commit | commitdiff | tree |
| 2010-02-02 |
Dale Johannesen | Accept floating point immediates in DEBUG_VALUE.
|
commit | commitdiff | tree |
| 2010-02-02 |
Dale Johannesen | Test revert 95050; there's a good chance it's causing
|
commit | commitdiff | tree |
| 2010-02-02 |
Dale Johannesen | Make local RA smarter about reusing input register...
|
commit | commitdiff | tree |
| 2010-02-01 |
Dale Johannesen | Testcase for 94996 (PR 6157)
|
commit | commitdiff | tree |
| 2010-02-01 |
Dale Johannesen | fix PR 6157. Testcase pending.
|
commit | commitdiff | tree |
| 2010-01-30 |
Dale Johannesen | Fix a case where debug_value could affect codegen.
|
commit | commitdiff | tree |
| 2010-01-29 |
Dale Johannesen | Add assertion to humor the paranoid.
|
commit | commitdiff | tree |
| 2010-01-27 |
Dale Johannesen | If the only use of something is a DEBUG_VALUE, don't
|
commit | commitdiff | tree |
| 2010-01-27 |
Dale Johannesen | Treat MO_REG 0 location as undefined in DEBUG_VALUE,
|
commit | commitdiff | tree |
| 2010-01-26 |
Dale Johannesen | Accept immediate as value of a dbg_value.
|
commit | commitdiff | tree |
| 2010-01-26 |
Dale Johannesen | Generate DEBUG_VALUE comments on x86. The (limited)
|
commit | commitdiff | tree |
| 2010-01-26 |
Dale Johannesen | use findDebugLoc in more places.
|
commit | commitdiff | tree |
| 2010-01-22 |
Dale Johannesen | Ignore DEBUG_VALUE when building live intervals;
|
commit | commitdiff | tree |
| 2010-01-20 |
Dale Johannesen | make findDebugLoc a class method
|
commit | commitdiff | tree |
| 2010-01-20 |
Dale Johannesen | Move findDebugLoc somewhere more central. Fix
|
commit | commitdiff | tree |
| 2010-01-19 |
Dale Johannesen | Fix a case where debug_value was perturbing the
|
commit | commitdiff | tree |
| 2010-01-19 |
Dale Johannesen | Reverting.
|
commit | commitdiff | tree |
| 2010-01-19 |
Dale Johannesen | Revert 93811 per request.
|
commit | commitdiff | tree |
| 2010-01-18 |
Dale Johannesen | Enable code to emit dbg.declare as DEBUG_VALUE
|
commit | commitdiff | tree |
| 2010-01-15 |
Dale Johannesen | Adjust some comments per review.
|
commit | commitdiff | tree |
| 2010-01-15 |
Dale Johannesen | Remove unneeded argument per review.
|
commit | commitdiff | tree |
| 2010-01-15 |
Dale Johannesen | DEBUG_VALUE is now variable sized, as it has a
|
commit | commitdiff | tree |
| 2010-01-15 |
Dale Johannesen | Revert 93499. After discussion with Chris we agreed
|
commit | commitdiff | tree |
| 2010-01-15 |
Dale Johannesen | Lower FrameIndex operand of DEBUG_VALUE (specially...
|
commit | commitdiff | tree |
| 2010-01-15 |
Dale Johannesen | Remove DEBUG_DECLARE, looks like we don't need it.
|
commit | commitdiff | tree |
| 2010-01-14 |
Dale Johannesen | Fix a comment.
|
commit | commitdiff | tree |
| 2010-01-13 |
Dale Johannesen | Fix a comment.
|
commit | commitdiff | tree |
| 2010-01-13 |
Dale Johannesen | Further progration of metadata operands. The
|
commit | commitdiff | tree |
| 2010-01-12 |
Dale Johannesen | Add MO_Metadata as an operand kind. Not used yet.
|
commit | commitdiff | tree |
| 2010-01-09 |
Dale Johannesen | Add DEBUG_DECLARE. Not used yet.
|
commit | commitdiff | tree |
| 2010-01-08 |
Dale Johannesen | Add DEBUG_VALUE. Not used yet.
|
commit | commitdiff | tree |
| 2010-01-06 |
Dale Johannesen | Reenable debug info on PPC. Works well enough to
|
commit | commitdiff | tree |
| 2010-01-06 |
Dale Johannesen | Make sure debug info hook gets called when emitting
|
commit | commitdiff | tree |
| 2009-12-24 |
Dale Johannesen | Testcase for llvm-gcc checkin 92108.
|
commit | commitdiff | tree |
| 2009-12-23 |
Dale Johannesen | Use more sensible type for flags in asms. PR 5570.
|
commit | commitdiff | tree |
| 2009-12-16 |
Dale Johannesen | Do better with physical reg operands (typically, from...
|
commit | commitdiff | tree |
| 2009-11-25 |
Dale Johannesen | Test for llvm-gcc checkin 89898.
|
commit | commitdiff | tree |
| 2009-11-25 |
Dale Johannesen | Fix compiler warnings.
|
commit | commitdiff | tree |
| 2009-11-24 |
Dale Johannesen | Do not store R31 into the caller's link area on PPC.
|
commit | commitdiff | tree |
| 2009-11-24 |
Dale Johannesen | Make capitalization of names starting "is" more consistent.
|
commit | commitdiff | tree |
| 2009-11-21 |
Dale Johannesen | When generating a vector the really slow way, via loads
|
commit | commitdiff | tree |
| 2009-11-20 |
Dale Johannesen | Remove an incorrect overaggressive optimization
|
commit | commitdiff | tree |
| 2009-11-13 |
Dale Johannesen | Adjust isConstantSplat to allow for big-endian targets.
|
commit | commitdiff | tree |
| 2009-11-10 |
Dale Johannesen | Emit correct code when making a ConstantPool entry...
|
commit | commitdiff | tree |
| 2009-10-29 |
Dale Johannesen | When there is a 2-instruction spill sequence, record
|
commit | commitdiff | tree |
| 2009-10-28 |
Dale Johannesen | When we generate spill code, then decide we don't need
|
commit | commitdiff | tree |
| 2009-10-27 |
Dale Johannesen | Add radar number.
|
commit | commitdiff | tree |
| 2009-10-27 |
Dale Johannesen | Testcase for llvm-gcc patch 85284.
|
commit | commitdiff | tree |
| 2009-10-23 |
Dale Johannesen | Commit fixes for half precision I noted in review, so
|
commit | commitdiff | tree |
| 2009-10-21 |
Dale Johannesen | Adjust testcases for msasm -> alignstack.
|
commit | commitdiff | tree |
| 2009-10-21 |
Dale Johannesen | Rename msasm to alignstack per review.
|
commit | commitdiff | tree |
| 2009-10-14 |
Dale Johannesen | Use llvmgxx for C++ test.
|
commit | commitdiff | tree |
| 2009-10-14 |
Dale Johannesen | Testcases for msasm bit (llvm-gcc 84062).
|
commit | commitdiff | tree |
| 2009-10-13 |
Dale Johannesen | Documentation for the new msasm flag, which is no
|
commit | commitdiff | tree |
| 2009-10-13 |
Dale Johannesen | Add an "msasm" flag to inline asm as suggested in PR...
|
commit | commitdiff | tree |
| 2009-10-12 |
Dale Johannesen | Revert the kludge in 76703. I got a clean
|
commit | commitdiff | tree |
| next |
projects / oota-llvm.git / search