Commits · f2f0806f717a7646d1065eea20f6359f7fb4a69b · Roger Ferrer / llvm-epi-0.8

Mar 08, 2011
- fix incorrect comment. · c7b48d40
  Chris Lattner authored Mar 08, 2011
```
llvm-svn: 127252
```
  c7b48d40
- Since last couple of days, argument number is encoded using 8 bits from line... · e1aa67cf
  Devang Patel authored Mar 08, 2011
```
Since last couple of days, argument number is encoded using 8 bits from line number field in argument's debug info.

llvm-svn: 127250
```
  e1aa67cf
- Update DILexicalBlock doc to mention new fields. · fb30b85e
  Devang Patel authored Mar 08, 2011
```
llvm-svn: 127249
```
  fb30b85e
- X86: Fix the (saddo/ssub x, 1) -> incl/decl selection to check the right operand for 1. · 679cfb54
  Benjamin Kramer authored Mar 08, 2011
```
Found by inspection.

llvm-svn: 127247
```
  679cfb54
- PTX: Add intrinsic support for ntid, ctaid, and nctaid registers · 42e9aaa4
  Justin Holewinski authored Mar 08, 2011
```
llvm-svn: 127246
```
  42e9aaa4
- Make a comparator's argument `const'. This fixes the build for MSVC 9. · 6ec5983a
  Oscar Fuentes authored Mar 08, 2011
```
llvm-svn: 127245
```
  6ec5983a
- Remove remnant of my experimentation with CMake. · a693632d
  Oscar Fuentes authored Mar 08, 2011
```
llvm-svn: 127244
```
  a693632d
- Fix PR9331. Simplified version of a patch by Jakub Staszak. · 7dc3d47c
  Duncan Sands authored Mar 08, 2011
```
llvm-svn: 127243
```
  7dc3d47c
- lib/Support/regcomp.c: Fix cygwin warning. · 5918457b
  NAKAMURA Takumi authored Mar 08, 2011
```
llvm-svn: 127241
```
  5918457b
- Use $(ECHOPATH) to make llvm-lit from llvm-lit.in. · 8506e4a1
  NAKAMURA Takumi authored Mar 08, 2011
```
llvm-svn: 127240
```
  8506e4a1
- Reapply r127073(partially): Introduce $(ECHOPATH) to print DOSish path string... · 0b26c3a6
  NAKAMURA Takumi authored Mar 08, 2011
```
Reapply r127073(partially): Introduce $(ECHOPATH) to print DOSish path string on MSYS bash for alternative of $(ECHO).

On mingw and python/w32, lit would not be expected to understand MSYS-style path.

llvm-svn: 127239
```
  0b26c3a6
- Testcase for r127187. · 05334dad
  Bill Wendling authored Mar 08, 2011
```
llvm-svn: 127236
```
  05334dad
- Don't show commands. · 45f470b8
  Bill Wendling authored Mar 08, 2011
```
llvm-svn: 127224
```
  45f470b8
- Reorder comments to put them the right way around. · afc8098c
  Nick Lewycky authored Mar 08, 2011
```
llvm-svn: 127220
```
  afc8098c
- While sinking an instruction, do not lose llvm.dbg.value intrinsic. · 97d0be8e
  Devang Patel authored Mar 08, 2011
```
llvm-svn: 127214
```
  97d0be8e
- Change the next release number to 3.0svn. · 1d5532a0
  Bill Wendling authored Mar 08, 2011
```
llvm-svn: 127213
```
  1d5532a0
- A few more tests for instruction encodings. · a259686d
  Bill Wendling authored Mar 08, 2011
```
llvm-svn: 127209
```
  a259686d
- Turn on list-ilp scheduling by default on x86 and x86-64, fix up · eb19e9e9
  Eric Christopher authored Mar 08, 2011
```
testcases accordingly. Some are currently xfailed and will be filed
as bugs to be fixed or understood.

Performance results:

roughly neutral on SPEC
some micro benchmarks in the llvm suite are up between 100 and 150%, only
a pair of regressions that are due to be investigated

john-the-ripper saw:
10% improvement in traditional DES
8% improvement in BSDI DES
59% improvement in FreeBSD MD5
67% improvement in OpenBSD Blowfish
14% improvement in LM DES

Small compile time impact.

llvm-svn: 127208
```
  eb19e9e9
- Further improvements to pre-RA-sched=list-ilp. · 52b3e38a
  Andrew Trick authored Mar 08, 2011
```
This change uses the MaxReorderWindow for both height and depth, which
tends to limit the negative effects of high register pressure.

llvm-svn: 127203
```
  52b3e38a
- Fix a compiler crash where a Glue value had multiple uses. Radar 9049552. · 45acbd03
  Bob Wilson authored Mar 08, 2011
```
llvm-svn: 127198
```
  45acbd03
- Fix comment typos. · 70bd3635
  Bob Wilson authored Mar 08, 2011
```
llvm-svn: 127197
```
  70bd3635
- Rename the narrow shift right immediate operands to "shr_imm*" operands. Also · 77ad1dc5
  Bill Wendling authored Mar 07, 2011
```
expand the testing of the narrowing shift right instructions.

No functionality change.

llvm-svn: 127193
```
  77ad1dc5
- Let shrinkToUses optionally return a list of now dead machine instructions. · 71c380f6
  Jakob Stoklund Olesen authored Mar 07, 2011
```
llvm-svn: 127192
```
  71c380f6
Mar 07, 2011
- Typos. · ac546dc6
  Eric Christopher authored Mar 07, 2011
```
llvm-svn: 127186
```
  ac546dc6
- Preserve line no. info. · d00c628f
  Devang Patel authored Mar 07, 2011
```
Radar 9097659

llvm-svn: 127182
```
  d00c628f
- Make the UselessRegs argument optional in the LiveRangeEdit constructor. · 27f942fa
  Jakob Stoklund Olesen authored Mar 07, 2011
```
llvm-svn: 127181
```
  27f942fa
- Move getRegPressureLimit() from TargetLoweringInfo to TargetRegisterInfo. · df616944
  Cameron Zwarich authored Mar 07, 2011
```
llvm-svn: 127175
```
  df616944
- Add test for r127138. · 15a29867
  Rafael Espindola authored Mar 07, 2011
```
llvm-svn: 127172
```
  15a29867
- Often GCC can see that NumBuckets is zero here, resulting in a warning · e5d7369e
  Duncan Sands authored Mar 07, 2011
```
about possibly swapped memset parameters.  Avoid the warning.

llvm-svn: 127170
```
  e5d7369e
- Tidy up. · e9fdc06f
  Jim Grosbach authored Mar 07, 2011
```
llvm-svn: 127169
```
  e9fdc06f
- Handle the special case of registers begin redefined by early-clobber defs. · ac32d8a6
  Jakob Stoklund Olesen authored Mar 07, 2011
```
In this case, the value need to be available at the load index instead of the
normal use index.

llvm-svn: 127167
```
  ac32d8a6
- Use the correct LHS type when determining the legalization of a shift's RHS type. · cd526fa1
  Owen Anderson authored Mar 07, 2011
```
llvm-svn: 127163
```
  cd526fa1
- ptx: Fix calling convention printing in AsmWriter.cpp · 0cfa737f
  Justin Holewinski authored Mar 07, 2011
```
This allows LLVM IR using ptx_kernel or ptx_device calling
conventions to be properly printed when emitted in text form.

llvm-svn: 127157
```
  0cfa737f
- Clarify that the result of an srem is only guaranteed to have the same sign as the · 2769c6e0
  Duncan Sands authored Mar 07, 2011
```
left-hand-side if the result is non-zero.

llvm-svn: 127156
```
  2769c6e0
- Don't keep the log files around. Just pipe to a log file instead. · e51e464c
  Bill Wendling authored Mar 07, 2011
```
llvm-svn: 127155
```
  e51e464c
- Fix tagging name. · 4785885f
  Bill Wendling authored Mar 07, 2011
```
llvm-svn: 127154
```
  4785885f
- Try fixing mingw build. · 35fd6dec
  Argyrios Kyrtzidis authored Mar 07, 2011
```
llvm-svn: 127153
```
  35fd6dec
- Prepend LD.Flags before LDFLAGS to avoid overriding necessary paths. · bb8599ad
  Eric Christopher authored Mar 07, 2011
```
Patch by Jack Howarth.

llvm-svn: 127152
```
  bb8599ad
- Tweak this test. We can analyze what happens and show that we still do the · ac55c79d
  Nick Lewycky authored Mar 07, 2011
```
right thing, instead of merely being unable to analyze and the transform
doesn't occur.

llvm-svn: 127149
```
  ac55c79d
- Add more analysis of the sign bit of an srem instruction. If the LHS is negative · e467979d
  Nick Lewycky authored Mar 07, 2011
```
then the result could go either way. If it's provably positive then so is the
srem. Fixes PR9343 #7!

llvm-svn: 127146
```
  e467979d