Commits · 74ded57bb85c28969fd528cb1576a5017c4921ee · Roger Ferrer / llvm-epi-0.8

Jan 13, 2011
- Try again enabling LiveDebugVariables. · 74ded57b
  Jakob Stoklund Olesen authored Jan 12, 2011
```
llvm-svn: 123342
```
  74ded57b
- Added clarifying comment · 14558422
  Jason W Kim authored Jan 12, 2011
```
llvm-svn: 123341
```
  14558422
- JimG sez: "The value-kinds look like masks, but they're not consistently used · e9eae0f8
  Jason W Kim authored Jan 12, 2011
```
that way, unfortunately. If you want to change them to work additively instead
of a one-variant-kind-per-symbolref, that's great and I completely agree it's
worth doing, but it really should be a separate patch. Until then, this isn't
correct."

So I am reverting this bit until a more opportune time.

llvm-svn: 123340
```
  e9eae0f8
- Don't emit a DBG_VALUE for a spill slot that the rewriter decided not to use after all. · e63dfeee
  Jakob Stoklund Olesen authored Jan 12, 2011
```
llvm-svn: 123339
```
  e63dfeee
- Fix braino in dominator tree walk. · 2ffee66e
  Jakob Stoklund Olesen authored Jan 12, 2011
```
llvm-svn: 123338
```
  2ffee66e
Jan 12, 2011
- Sometimes, old virtual registers can linger on DBG_VALUE instructions. · 1a3534af
  Jakob Stoklund Olesen authored Jan 12, 2011
```
Make sure we don't crash in that case, but simply turn them into %noreg instead.

llvm-svn: 123335
```
  1a3534af
- Teach VirtRegRewriter to update slot indexes when erasing instructions. · 013c4649
  Jakob Stoklund Olesen authored Jan 12, 2011
```
It was leaving dangling pointers in the slot index maps.

llvm-svn: 123334
```
  013c4649
- Annotate VirtRegRewriter debug output with slot indexes. · 71a38533
  Jakob Stoklund Olesen authored Jan 12, 2011
```
llvm-svn: 123333
```
  71a38533
- Verify slot index ordering. · 58b6f4d8
  Jakob Stoklund Olesen authored Jan 12, 2011
```
The slot indexes must be monotonically increasing through the function.

llvm-svn: 123324
```
  58b6f4d8
- Verify that machine instruction parent pointers are consistent. · b5b4a5d0
  Jakob Stoklund Olesen authored Jan 12, 2011
```
llvm-svn: 123322
```
  b5b4a5d0
- Sort the register list based on the *actual* register numbers rather than the · e6ff05c5
  Bill Wendling authored Jan 12, 2011
```
enum values we give to them. <rdar://problem/8823730>

llvm-svn: 123321
```
  e6ff05c5
- Use SmallVector instead of SmallPtrSet and avoid non-deterministic behavior. · 30f3ebbc
  Devang Patel authored Jan 12, 2011
```
llvm-svn: 123318
```
  30f3ebbc
- Mostly undo r123297, but move the default case in EvaluateAsPCRel to the top · 3077bb64
  Matt Beaumont-Gay authored Jan 12, 2011
```
of the switch block to appease GCC.

llvm-svn: 123317
```
  3077bb64
- Add another note taken from the gcc bugzilla. · 7ecc2fc4
  Nick Lewycky authored Jan 12, 2011
```
llvm-svn: 123315
```
  7ecc2fc4
- Implement RETURNADDR and FRAMEADDR lowering in SPARC backend. · d964580f
  Venkatraman Govindaraju authored Jan 12, 2011
```
llvm-svn: 123310
```
  d964580f
- Remove SPARC backend getpcx instruction's Uses. Also, insert an assert to · ee347f80
  Venkatraman Govindaraju authored Jan 12, 2011
```
ensure %o7 is not assigned as the destination of getpcx instruction.

llvm-svn: 123304
```
  ee347f80
- revert 123144, reenabling the rest of memset formation. · dd5f60b7
  Chris Lattner authored Jan 12, 2011
```
llvm-svn: 123302
```
  dd5f60b7
- Fix SPARC backend call instruction so that arguments passed through registers · 3b71b0ae
  Venkatraman Govindaraju authored Jan 12, 2011
```
are correctly marked as used instead of passing all possible argument registers
as used.  

llvm-svn: 123301
```
  3b71b0ae
- revert r123146 which disabled code that wasn't the root cause · 654098f4
  Chris Lattner authored Jan 12, 2011
```
of the bootstrap miscompare issue.

llvm-svn: 123299
```
  654098f4
- revert r123149, reenabling an improvement to memcpyopt that wasn't · fa7c29d2
  Chris Lattner authored Jan 12, 2011
```
the source of the bootstrap problem.

llvm-svn: 123298
```
  fa7c29d2
- Prefer llvm_unreachable to assert(0) · ea431722
  Matt Beaumont-Gay authored Jan 12, 2011
```
llvm-svn: 123297
```
  ea431722
- 1. Support ELF pcrel relocations for movw/movt: · 9c5b65d2
  Jason W Kim authored Jan 12, 2011
```
  R_ARM_MOVT_PREL and R_ARM_MOVW_PREL_NC.
2. Fix minor bug in ARMAsmPrinter - treat bitfield flag as a bitfield, not an enum.
3. Add support for 3 new elf section types (no-ops)

llvm-svn: 123294
```
  9c5b65d2
- Workaround for bug 8721. · 1f7bc070
  Jason W Kim authored Jan 11, 2011
```
.s Test added.

llvm-svn: 123292
```
  1f7bc070
- The world is not ready for LiveDebugVariables yet. · 43812bfa
  Jakob Stoklund Olesen authored Jan 11, 2011
```
llvm-svn: 123290
```
  43812bfa
Jan 11, 2011
- Remove the PR8954 workaround. · 12cc296b
  Jakob Stoklund Olesen authored Jan 11, 2011
```
llvm-svn: 123288
```
  12cc296b
- Fix a non-deterministic loop in llvm::MergeBlockIntoPredecessor. · f2407aa9
  Jakob Stoklund Olesen authored Jan 11, 2011
```
DT->changeImmediateDominator() trivially ignores identity updates, so there is
really no need for the uniqueing provided by SmallPtrSet.

I expect this to fix PR8954.

llvm-svn: 123286
```
  f2407aa9
- Enable LiveDebugVariables by default. · 8c98495f
  Jakob Stoklund Olesen authored Jan 11, 2011
```
llvm-svn: 123282
```
  8c98495f
- SPARC backend: correct ICC/FCC uses for ADDX and SELECT_CC · 4d6ade0e
  Venkatraman Govindaraju authored Jan 11, 2011
```
llvm-svn: 123281
```
  4d6ade0e
- Dial back the speculative fix for PR8954 a bit, so that we only recompute dominators · cb9c4f85
  Cameron Zwarich authored Jan 11, 2011
```
once at the beginning of GVN instead of once per iteration.

llvm-svn: 123278
```
  cb9c4f85
- Don't insert DBG_VALUE instructions after the first terminator. · 803f48bc
  Jakob Stoklund Olesen authored Jan 11, 2011
```
For one, MachineBasicBlock::getFirstTerminator() doesn't understand what is
happening, and it also makes sense to have all control flow run through the
DBG_VALUE.

llvm-svn: 123277
```
  803f48bc
- Clean up ARM subtarget code by using Triple ADT. · e45d6858
  Evan Cheng authored Jan 11, 2011
```
llvm-svn: 123276
```
  e45d6858
- Appropriately truncate debug info range in dwarf output. · 447cb38f
  Devang Patel authored Jan 11, 2011
```
This is not yet completely enabled.

llvm-svn: 123274
```
  447cb38f
- Attempt to fix the bootstrap buildbot. Rafael says this works for him on x86-64 Linux. · 51eb4039
  Cameron Zwarich authored Jan 11, 2011
```
llvm-svn: 123270
```
  51eb4039
- McARM: Fill in GetMnemonicAcceptInfo(). · 09264124
  Daniel Dunbar authored Jan 11, 2011
```
llvm-svn: 123253
```
  09264124
- Remove dead variable, const-ref-ize an APInt. · 0022a4b4
  Owen Anderson authored Jan 11, 2011
```
llvm-svn: 123248
```
  0022a4b4
- this pass claims to preserve scev, make sure to tell it about deletions. · d41db8f9
  Chris Lattner authored Jan 11, 2011
```
llvm-svn: 123247
```
  d41db8f9
- Fix a comment: We now have intrinsics for vcvtr. · e5863d66
  Bob Wilson authored Jan 11, 2011
```
llvm-svn: 123246
```
  e5863d66
- some comment improvements. · d30de955
  Chris Lattner authored Jan 11, 2011
```
llvm-svn: 123243
```
  d30de955
- Fix PR8946, a missing reg/reg form of movdqu. · abd2dfd3
  Chris Lattner authored Jan 11, 2011
```
llvm-svn: 123242
```
  abd2dfd3
- McARM: Sketch some logic for determining when to add carry set and predication... · 5a384c86
  Daniel Dunbar authored Jan 11, 2011
```
McARM: Sketch some logic for determining when to add carry set and predication code operands based on the "canonical mnemonic".

llvm-svn: 123239
```
  5a384c86