Commits · cf3acae26e3ebc72210a8e8761329ca97d31dd7c · Roger Ferrer / llvm-epi-0.8

May 29, 2008
- For PR1338: rename include/llvm/ADT/ilist and friends to end with ".h" · cf3acae2
  Anton Korobeynikov authored May 29, 2008
```
llvm-svn: 51687
```
  cf3acae2
- Force postdom to be linked into opt and bugpoint, even though it is no longer used by any passes. · b1ba352e
  Owen Anderson authored May 29, 2008
```
llvm-svn: 51686
```
  b1ba352e
- Move these tests into the proper directory. · 50d602cd
  Owen Anderson authored May 29, 2008
```
llvm-svn: 51685
```
  50d602cd
- Add a newline at the end of this file. · 9e064a21
  Duncan Sands authored May 29, 2008
```
llvm-svn: 51680
```
  9e064a21
- Re-enable the newly simplified ADCE. This fixes a regression on · fd24ce0a
  Owen Anderson authored May 29, 2008
```
Dhrystone introduced by its removal.

llvm-svn: 51669
```
  fd24ce0a
- Replace the old ADCE implementation with a new one that more simply solves · 7686b555
  Owen Anderson authored May 29, 2008
```
the one case that ADCE catches that normal DCE doesn't: non-induction variable
loop computations.

This implementation handles this problem without using postdominators.

llvm-svn: 51668
```
  7686b555
- Implement vector shift up / down and insert zero with ps{rl}lq / ps{rl}ldq. · 5e28227d
  Evan Cheng authored May 29, 2008
```
llvm-svn: 51667
```
  5e28227d
- Remove debugging code. · f4aece59
  Owen Anderson authored May 29, 2008
```
llvm-svn: 51666
```
  f4aece59
- Add nounwind. · 6892c550
  Evan Cheng authored May 29, 2008
```
llvm-svn: 51665
```
  6892c550
- XOR?RI instructions aren't as cheap as moves. · 0252be17
  Bill Wendling authored May 29, 2008
```
llvm-svn: 51664
```
  0252be17
- convert more operand loops to iterator formulation · 3a9fba5a
  Gabor Greif authored May 29, 2008
```
llvm-svn: 51663
```
  3a9fba5a
- Implement "AsCheapAsAMove" for some obviously cheap instructions: xor and the · 7a1a8eb6
  Bill Wendling authored May 29, 2008
```
like.

llvm-svn: 51662
```
  7a1a8eb6
- convert another operand loop to iterator formulation · f712664e
  Gabor Greif authored May 29, 2008
```
llvm-svn: 51661
```
  f712664e
- Add a flag to indicate that an instruction is as cheap (or cheaper) than a move · 3f6bb271
  Bill Wendling authored May 28, 2008
```
instruction to execute. This can be used for transformations (like two-address
conversion) to remat an instruction instead of generating a "move"
instruction. The idea is to decrease the live ranges and register pressure and
all that jazz.

llvm-svn: 51660
```
  3f6bb271
- Check the "isSafeToMove" predicate, which has a series of tests to make sure · 5a83b097
  Bill Wendling authored May 28, 2008
```
that it's safe to remat an instruction.

llvm-svn: 51659
```
  5a83b097
- Remap VNInfo data as well when doing renumbering. · 779b4180
  Owen Anderson authored May 28, 2008
```
llvm-svn: 51658
```
  779b4180
May 28, 2008
- Factor the numbering computation into a separate method, and add the slightest... · 4f8e1ad3
  Owen Anderson authored May 28, 2008
```
Factor the numbering computation into a separate method, and add the slightest attempt at some renumbering logic, which is currently unused.

llvm-svn: 51652
```
  4f8e1ad3
- Fix PR2289: vr defined by multiple implicit_def as result of coalescing. · 68079268
  Evan Cheng authored May 28, 2008
```
llvm-svn: 51648
```
  68079268
- Teach local register allocator to deal with landing pad MBB's. · 427412e7
  Evan Cheng authored May 28, 2008
```
llvm-svn: 51647
```
  427412e7
- Update some comments noticed in a recent checkin · e993b80e
  Nate Begeman authored May 28, 2008
```
llvm-svn: 51644
```
  e993b80e
- Update text to point people at the right version of the tutorial for · e438c561
  Chris Lattner authored May 28, 2008
```
their release.

llvm-svn: 51638
```
  e438c561
- Implement PR2370: memmove(x,x,size) -> noop. · ecdefb5d
  Chris Lattner authored May 28, 2008
```
llvm-svn: 51636
```
  ecdefb5d
- Add chain inputs for loads. · f8910ab6
  Chris Lattner authored May 28, 2008
```
llvm-svn: 51635
```
  f8910ab6
- Fix CodeGen/Generic/2005-10-21-longlonggtu.ll on ia64. · 633cd594
  Chris Lattner authored May 28, 2008
```
llvm-svn: 51634
```
  633cd594
- loads should get chains. THis helps but does not solve CodeGen/Generic/2003-05-27-phifcmpd.ll · 72489562
  Chris Lattner authored May 28, 2008
```
on ia64.

llvm-svn: 51633
```
  72489562
- Fix 2006-04-28-Sign-extend-bool.ll for ia64. · d2c3e86c
  Chris Lattner authored May 28, 2008
```
llvm-svn: 51632
```
  d2c3e86c
- reindent. · c2fb8d7e
  Chris Lattner authored May 28, 2008
```
llvm-svn: 51631
```
  c2fb8d7e
- Fix the encoding for two more "rm" instructions that were using MRMSrcReg. · 68bddb89
  Dan Gohman authored May 28, 2008
```
llvm-svn: 51630
```
  68bddb89
- Fixed X86 encoding error CVTPS2PD and CVTPD2PS when the source operand · 5e3faf23
  Mon P Wang authored May 28, 2008
```
is a memory location

llvm-svn: 51626
```
  5e3faf23
- Don't attempt to create VZEXT_LOAD out of an extload. This an issue where the · f1e18c7c
  Nate Begeman authored May 28, 2008
```
code generator would do something like this:

f64 = load f32 <anyext>, f32mem
v2f64 = insertelt undef, %0, 0
v2f64 = insertelt %1, 0.0, 1

into 

v2f64 = vzext_load f32mem

which on x86 is movsd, when you really wanted a cvtss2sd/movsd pair.

llvm-svn: 51624
```
  f1e18c7c
- use space insted of tabs · 1c4627c2
  Nick Kledzik authored May 28, 2008
```
llvm-svn: 51623
```
  1c4627c2
- fix infinite recursion if a global's initializer references the global · ce956e06
  Nick Kledzik authored May 27, 2008
```
llvm-svn: 51617
```
  ce956e06
May 27, 2008
- Make Use::init() private, it breaks invariants, clients should assign or use set(). · 29e0ad2f
  Gabor Greif authored May 27, 2008
```
llvm-svn: 51615
```
  29e0ad2f
- Use Function::getEntryBlock instead of Function::begin, for clarity. · a61379ee
  Dan Gohman authored May 27, 2008
```
llvm-svn: 51613
```
  a61379ee
- Print debug output when any edge becomes executable, including · 3ec55202
  Dan Gohman authored May 27, 2008
```
the first visited edge.

llvm-svn: 51612
```
  3ec55202
- Add instcombine after global optimizations. · 8b16aeb8
  Devang Patel authored May 27, 2008
```
llvm-svn: 51611
```
  8b16aeb8
- Fix a word-o and add an explicit keyword. · 4ec1b4e3
  Dan Gohman authored May 27, 2008
```
llvm-svn: 51610
```
  4ec1b4e3
- Incorporated feedback: Check that the implicitly defined operands aren't used · 2e44ec7c
  Bill Wendling authored May 27, 2008
```
before deleting the instruction.

llvm-svn: 51609
```
  2e44ec7c
- Tidy up whitespace in ConstantRange::print output. · e3a6a183
  Dan Gohman authored May 27, 2008
```
llvm-svn: 51606
```
  e3a6a183
- Use IPSCCPPass instead of IPConstantPropagationPass. · d38b712e
  Devang Patel authored May 27, 2008
```
llvm-svn: 51605
```
  d38b712e