Commit Graph

43509 Commits

Author SHA1 Message Date
Chris Lattner
1af2113e10 minor cleanups and comment improvements.
llvm-svn: 61564
2009-01-02 08:05:26 +00:00
Chris Lattner
2950e29626 add a #include to hopefully get the x86-64-linux buildbot building.
llvm-svn: 61563
2009-01-02 07:18:46 +00:00
Chris Lattner
a8158ed0a9 Remove lex/bison support from makefile.rules.
llvm-svn: 61562
2009-01-02 07:16:45 +00:00
Chris Lattner
2d37a9be23 update the cmakefile. This is a "best guess", I haven't tested this.
llvm-svn: 61561
2009-01-02 07:14:23 +00:00
Chris Lattner
59d439dd15 lex and bison aren't needed to build llvm anymore.
llvm-svn: 61560
2009-01-02 07:10:51 +00:00
Chris Lattner
53fa6e3f6b new non-ambiguous syntax.
llvm-svn: 61559
2009-01-02 07:02:56 +00:00
Chris Lattner
f28c74870f Reimplement the old and horrible bison parser for .ll files with a nice
and clean recursive descent parser.

This change has a couple of ramifications:
1. The parser code is about 400 lines shorter (in what we maintain, not
   including what is autogenerated).
2. The code should be significantly faster than the old code because we 
   don't have to work around bison's poor handling of datatypes with 
   ctors/dtors.  This also makes the code much more resistant to memory 
   leaks.
3. We now get caret diagnostics from the .ll parser, woo.
4. The actual diagnostics emited from the parser are completely different
   so a bunch of testcases had to be updated.
5. I now disallow "%ty = type opaque %ty = type i32".  There was no good
   reason to support this, it was just an accident of the old 
   implementation.  I have no reason to think that anyone is actually using
   this.
6. The syntax for sticking a global variable has changed to make it 
   unambiguous.  I don't think anyone is depending on this since only clang
   supports this and it is not solid yet, so I'm not worried about anything
   breaking.
7. This gets rid of the last use of bison, and along with it the .cvs files.
   I'll prune this from the makefiles as a subsequent commit.

There are a few minor cleanups that can be done after this commit (suggestions
welcome!) but this passes dejagnu testing and is ready for its time in the
limelight.

llvm-svn: 61558
2009-01-02 07:01:27 +00:00
Evan Cheng
c52f942d67 Do not isel load folding bt instructions for pentium m, core, core2, and AMD processors. These are significantly slower than a load followed by a bt of a register.
llvm-svn: 61557
2009-01-02 05:35:45 +00:00
Evan Cheng
f460ec040c Fix x86 CPU id detection to identify Penryn (and future processors).
llvm-svn: 61556
2009-01-02 05:29:20 +00:00
Evan Cheng
57115c1887 Use movaps / movd to extract vector element 0 even with sse4.1. It's still cheaper than pextrw especially if the value is in memory.
llvm-svn: 61555
2009-01-02 05:29:08 +00:00
Nick Lewycky
35e441f7f1 Fix typo.
llvm-svn: 61554
2009-01-02 05:01:22 +00:00
Nick Lewycky
0993a85522 Remove the cyclic part of this test, it was passing for the wrong
reason. Two functions which mutually require each other to be nocapture 
are not currently supported.

llvm-svn: 61553
2009-01-02 03:52:27 +00:00
Nick Lewycky
6c53fbb21d Make adding nocapture a bit stronger. FreeInst is nocapture. Also,
functions that don't write can't leak a pointer except through 
the return value, so a void readonly function is implicitly nocapture.

Test these, and add a test that verifies that f1 calling f2 with an 
otherwise dead pointer gets both of them marked nocapture.

llvm-svn: 61552
2009-01-02 03:46:56 +00:00
Bill Wendling
46c9d8127d Don't error out if ranlib fails.
llvm-svn: 61551
2009-01-02 02:53:24 +00:00
Chris Lattner
2d3e57c337 rename a file to follow naming conventions.
llvm-svn: 61550
2009-01-02 01:52:35 +00:00
Duncan Sands
253f6a5dce Add tests for two types of traps that escape analysis
might one day fall into.

llvm-svn: 61549
2009-01-02 00:55:51 +00:00
Duncan Sands
e4fd98d306 Mention that this pass does escape analysis in the
leading comments.

llvm-svn: 61548
2009-01-01 20:45:19 +00:00
Duncan Sands
0fca32114b Factorize (and generalize) the code promoting SELECT
and BRCOND conditions.  Reorder a few methods while
there.

llvm-svn: 61547
2009-01-01 20:36:20 +00:00
Misha Brukman
361188d5bc * Quoted the executable 'runtest' to emphasize the binary needed;
otherwise, some unlucky souls start looking for a 'dejagnu' binary...
* Properly capitalized LLVM.

llvm-svn: 61546
2009-01-01 20:26:05 +00:00
Duncan Sands
07002edaca Remove trailing spaces.
llvm-svn: 61545
2009-01-01 19:56:02 +00:00
Duncan Sands
190d6bc636 Fix PR3274: when promoting the condition of a BRCOND node,
promote from i1 all the way up to the canonical SetCC type.
In order to discover an appropriate type to use, pass
MVT::Other to getSetCCResultType.  In order to be able to
do this, change getSetCCResultType to take a type as an
argument, not a value (this is also more logical).

llvm-svn: 61542
2009-01-01 15:52:00 +00:00
Misha Brukman
3b9228c5ce Original patch by Talin.
* Added the first LLVM unittest -- DenseMap.
* Updated mkpatch utility to include llvm/unittests dir
* Added top-level target "unittests" to run all unittests

llvm-svn: 61541
2009-01-01 02:24:48 +00:00
Misha Brukman
7ac554bb07 * Removed gtest-all.cc; .cc files including other .cc files is weird
* Removed gtest_main.cc: we have our own main() elsewhere
* Simplified the Makefile as we don't need SOURCES

* Moved the internal header to gtest/internal/
* Simplified the Makefile to remove -I param to CPP.Flags

* Updated README.LLVM with all the steps I took to massage GTest to
  work in LLVM so far

llvm-svn: 61540
2009-01-01 02:05:43 +00:00
Misha Brukman
fb14f57734 Moved Google Test code up one directory so that we can use a standard LLVM
Makefile with it, without resorting to the use of VPATH.

Also added Makefiles at every level of the directory tree to properly recurse
to Google Test and build it as a library (original Makefiles by Talin).

llvm-svn: 61539
2009-01-01 01:29:44 +00:00
Bill Wendling
779f2e1702 Fix comment.
llvm-svn: 61538
2009-01-01 01:19:59 +00:00
Bill Wendling
efbe8b808c Add transformation:
xor (or (icmp, icmp), true) -> and(icmp, icmp)

This is possible because of De Morgan's law.

llvm-svn: 61537
2009-01-01 01:18:23 +00:00
Bill Wendling
bf9874fd09 Some compilers are picky about accessing the first element of a std::vector if
there's nothing in the vector. Pacify them.

llvm-svn: 61536
2009-01-01 01:14:31 +00:00
Duncan Sands
e112cf52cb Look through phi nodes and select instructions when
calculating nocapture attributes.

llvm-svn: 61535
2008-12-31 20:21:34 +00:00
Bill Wendling
e288a29970 This is not failing on Darwin for some reason. XFAIL for other platforms.
llvm-svn: 61533
2008-12-31 19:26:09 +00:00
Duncan Sands
03192120cc Don't analyze arguments already marked 'nocapture'.
llvm-svn: 61532
2008-12-31 18:08:59 +00:00
Misha Brukman
4951288453 Fixed grammar and capitalization.
llvm-svn: 61531
2008-12-31 17:44:36 +00:00
Misha Brukman
3eb537b893 Add spacing between type and variable name.
llvm-svn: 61530
2008-12-31 17:41:49 +00:00
Misha Brukman
ba68ff9a5d Comment headers should extend to 80 chars.
llvm-svn: 61529
2008-12-31 17:40:52 +00:00
Misha Brukman
6c8b57ddfa Fix spacing to be uniform for parameters.
llvm-svn: 61528
2008-12-31 17:39:58 +00:00
Misha Brukman
d7b9e41922 Removed extra spaces.
llvm-svn: 61527
2008-12-31 17:38:27 +00:00
Misha Brukman
1aecaaf274 Import of Google Test 1.2.1, with the non-essential bits removed.
Added a README.LLVM file to indicate which files and directories
were removed from the original source tarball.

llvm-svn: 61526
2008-12-31 17:34:06 +00:00
Duncan Sands
36db5853cb Rename AddReadAttrs to FunctionAttrs, and teach it how
to work out (in a very simplistic way) which function
arguments (pointer arguments only) are only dereferenced
and so do not escape.  Mark such arguments 'nocapture'.

llvm-svn: 61525
2008-12-31 16:14:43 +00:00
Oscar Fuentes
4954ccb97b CMake: Added some VC++ specific info to CMake.html.
llvm-svn: 61521
2008-12-31 14:36:41 +00:00
Chris Lattner
5fcb753ac1 Add a new Attribute::getAlignmentFromAttrs method.
llvm-svn: 61517
2008-12-31 08:41:38 +00:00
Bill Wendling
c73286fc81 XFAIL test caused by r61493. Apparently, this is expected?
llvm-svn: 61516
2008-12-31 08:26:55 +00:00
Owen Anderson
ccde9db05a Get live interval reconstruction several steps closer to working.
llvm-svn: 61514
2008-12-31 02:00:25 +00:00
Chris Lattner
1cfa9f47db add a note
llvm-svn: 61513
2008-12-31 00:54:13 +00:00
Scott Michel
c163bf5042 XFAIL this for now until I can figure out what's going on.
llvm-svn: 61512
2008-12-31 00:08:25 +00:00
Scott Michel
12a5f7cfb9 Fix test erratum (which is wierd: works locally for me?)
llvm-svn: 61511
2008-12-30 23:52:05 +00:00
Scott Michel
cdcae67887 - Start moving target-dependent nodes that could be represented by an
instruction sequence and cannot ordinarily be simplified by DAGcombine
  into the various target description files or SPUDAGToDAGISel.cpp.

  This makes some 64-bit operations legal.

- Eliminate target-dependent ISD enums.

- Update tests.

llvm-svn: 61508
2008-12-30 23:28:25 +00:00
Bill Wendling
067c48f7a6 Linux wants the FDE initial location and address range to be forced to 32-bit.
Darwin doesn't. Make this optional for platforms.

llvm-svn: 61484
2008-12-29 22:12:11 +00:00
Bill Wendling
4749654506 The FDE initial location and address range data should be free to be 64-bit
(quad) on a 64-bit platform. This fixes a problem with EH frames on Darwin.

llvm-svn: 61483
2008-12-29 21:51:42 +00:00
Misha Brukman
abfa50fc15 * Updated TOC and fixed named anchors
* Simplified section header marking 

llvm-svn: 61482
2008-12-29 21:33:30 +00:00
Misha Brukman
4d4b7394f9 Minimize the width of the notes region: it will end where the text ends, if the
line of text is short enough, instead of stretching all the way to the right
margin.

llvm-svn: 61481
2008-12-29 21:25:09 +00:00
Misha Brukman
5b00db2c8f Simplified marking code regions with a single <pre> rather than
a (<div>, <pre>) combo.

llvm-svn: 61480
2008-12-29 21:20:51 +00:00