| Commit message (Expand) | Author | Age | Files | Lines |
* | MC/Mach-O: Fix some differences in symbol flag handling. | Daniel Dunbar | 2010-05-17 | 1 | -2/+15 |
* | Remove debug option. Add comment on spill order determinism. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -7/+4 |
* | Turn on -neon-reg-sequence by default. | Evan Cheng | 2010-05-17 | 1 | -1/+2 |
* | llvm-mc: Support reassignment of variables in one special case, when the | Daniel Dunbar | 2010-05-17 | 1 | -1/+8 |
* | Avoid allocating the same physreg to multiple virtregs in one instruction. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -0/+1 |
* | Minor optimizations. DenseMap::begin() is surprisingly slow on an empty map. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -8/+7 |
* | Extract spill cost calculation to a new method, and use definePhysReg() to clear | Jakob Stoklund Olesen | 2010-05-17 | 1 | -86/+64 |
* | Remove unused member variable. | Zhongxing Xu | 2010-05-17 | 1 | -1/+0 |
* | Only use clairvoyance when defining a register, and then only if it has one use. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -16/+11 |
* | Eliminate a hash table probe when killing virtual registers. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -15/+20 |
* | Execute virtreg kills immediately instead of after processing all uses. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -19/+16 |
* | Sprinkle superregister <imp-def> and <imp-kill> operands when dealing with su... | Jakob Stoklund Olesen | 2010-05-17 | 1 | -10/+28 |
* | Now that we don't keep live registers across calls, there is not reason to go | Jakob Stoklund Olesen | 2010-05-17 | 1 | -7/+12 |
* | Boldly attempt consistent capitalization. Functional changes unintended. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -42/+42 |
* | Assume that we'll handle mangling the symbols earlier and just put the | Eric Christopher | 2010-05-17 | 2 | -11/+6 |
* | Spill and kill all virtual registers across a call. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -56/+28 |
* | Reduce hashtable probes by using DenseMap::insert() for lookup. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -48/+46 |
* | Make MBB a class member instead of passing it around everywhere. | Jakob Stoklund Olesen | 2010-05-17 | 1 | -62/+62 |
* | Yes, if the redef is a copy, update the old val# with the copy. But make sure... | Evan Cheng | 2010-05-17 | 1 | -8/+7 |
* | No reason not to run the NEON domain croassing fix up pass in thumb2 mode. | Evan Cheng | 2010-05-17 | 1 | -1/+1 |
* | Revert 103911; it broke a test that expects bitconvert | Dale Johannesen | 2010-05-16 | 1 | -4/+0 |
* | Make x86-64 64-bit bitconvert work when SSE is not available. | Dale Johannesen | 2010-05-16 | 1 | -0/+4 |
* | Chris said that the comment char should be escaped. Fix all the occurences of... | Anton Korobeynikov | 2010-05-16 | 4 | -28/+28 |
* | Add support for thiscall calling convention. | Anton Korobeynikov | 2010-05-16 | 10 | -3/+38 |
* | Generalize the ARM DAG combiner of mul with constants to all power-of-two cases. | Anton Korobeynikov | 2010-05-16 | 1 | -34/+21 |
* | Model vst lane instructions with REG_SEQUENCE. | Evan Cheng | 2010-05-16 | 1 | -7/+75 |
* | Fix uint64->{float, double} conversion to do rounding correctly in 32-bit. | Dale Johannesen | 2010-05-15 | 2 | -24/+76 |
* | Improve assertion messages. | Dale Johannesen | 2010-05-15 | 1 | -3/+8 |
* | Some cheap DAG combine goodness for multiplication with a particular constant. | Anton Korobeynikov | 2010-05-15 | 1 | -0/+71 |
* | "trap" pseudo-op turned out to be apple-local. | Anton Korobeynikov | 2010-05-15 | 2 | -2/+8 |
* | improve portability to systems that don't have round, patch by | Chris Lattner | 2010-05-15 | 1 | -4/+3 |
* | improve portability to systems that don't have powf/modf (e.g. solaris 9) | Chris Lattner | 2010-05-15 | 4 | -5/+5 |
* | Fix an GCC warning that seems to have actually caught a bug (!!!) in | Chandler Carruth | 2010-05-15 | 1 | -2/+2 |
* | Model 128-bit vld lane with REG_SEQUENCE. | Evan Cheng | 2010-05-15 | 1 | -19/+44 |
* | Calculate liveness on the fly for local registers. | Jakob Stoklund Olesen | 2010-05-15 | 1 | -0/+31 |
* | Teach the always inliner to release its inline cost estimates, like the basic | Nick Lewycky | 2010-05-15 | 1 | -0/+3 |
* | Clean up, no functional change. | Nick Lewycky | 2010-05-15 | 1 | -9/+9 |
* | v4i64 and v8i64 are only synthesizable when NEON is available. | Evan Cheng | 2010-05-15 | 1 | -4/+6 |
* | Allow TargetLowering::getRegClassFor() to be called on illegal types. Also | Evan Cheng | 2010-05-15 | 2 | -7/+17 |
* | Model 64-bit lane vld with REG_SEQUENCE. | Evan Cheng | 2010-05-15 | 1 | -6/+28 |
* | A partial re-def instruction may be a copy. | Evan Cheng | 2010-05-15 | 1 | -1/+8 |
* | Teach two-address pass to do some coalescing while eliminating REG_SEQUENCE | Evan Cheng | 2010-05-14 | 6 | -3/+192 |
* | Model VST*_UPD and VST*oddUPD pair with REG_SEQUENCE. | Evan Cheng | 2010-05-14 | 1 | -25/+68 |
* | Fast ISel trivially coalesces away no-op casts, so check for this when | Dan Gohman | 2010-05-14 | 1 | -4/+16 |
* | Don't bother spilling before a return | Jakob Stoklund Olesen | 2010-05-14 | 1 | -4/+13 |
* | RegAllocLocal can count copies too | Jakob Stoklund Olesen | 2010-05-14 | 1 | -1/+4 |
* | SystemZ really does mean "has calls" and not just "adjusts stack." Go ahead and | Bill Wendling | 2010-05-14 | 1 | -5/+5 |
* | Track allocatable instead of reserved regs, and never take an unallocatable h... | Jakob Stoklund Olesen | 2010-05-14 | 1 | -8/+10 |
* | Don't set kill flags for instructions which the scheduler has cloned. | Dan Gohman | 2010-05-14 | 2 | -20/+34 |
* | BR is a barrier. | Dan Gohman | 2010-05-14 | 1 | -1/+1 |