1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 20:23:11 +01:00
Commit Graph

74121 Commits

Author SHA1 Message Date
Bill Wendling
9a3aca1f35 Update the comment. This feature is available only on Darwin at the moment. Though it's not Darwin-specific.
llvm-svn: 135951
2011-07-25 20:15:15 +00:00
Jim Grosbach
f45795436a Make assembly parser method names more consistent.
llvm-svn: 135950
2011-07-25 20:14:50 +00:00
Oscar Fuentes
06558951e6 Unbreak the build.
llvm-svn: 135949
2011-07-25 20:13:36 +00:00
Jakub Staszak
171cc8d07d Fix class description.
llvm-svn: 135948
2011-07-25 20:08:58 +00:00
Jakub Staszak
2873980483 Fix #include guard directive.
llvm-svn: 135947
2011-07-25 20:08:00 +00:00
Jim Grosbach
402ca4879f Tidy up formatting.
Remove some inititalizers that are the same as the default, move defs next to
their (singular) uses and generally simplify some formatting of asm operand
definitions.

llvm-svn: 135946
2011-07-25 20:06:30 +00:00
Jim Grosbach
2ca37e058f Tidy up a bit.
llvm-svn: 135945
2011-07-25 20:00:32 +00:00
Evan Cheng
8a23f6dac1 Missed a file.
llvm-svn: 135943
2011-07-25 19:55:33 +00:00
Evan Cheng
43afb1ea2a Refactor PPC target to separate MC routines from Target routines.
llvm-svn: 135942
2011-07-25 19:53:23 +00:00
Evan Cheng
bbacc163bb More refactoring.
llvm-svn: 135939
2011-07-25 19:33:48 +00:00
Jakub Staszak
5c309cbead Rename BlockFrequency to BlockFrequencyInfo and MachineBlockFrequency to
MachineBlockFrequencyInfo.

llvm-svn: 135937
2011-07-25 19:25:40 +00:00
Evan Cheng
6fb04ad32e Refactor X86 target to separate MC code from Target code.
llvm-svn: 135930
2011-07-25 18:43:53 +00:00
Bill Wendling
713310d300 Changed disabled code into a flag.
llvm-svn: 135924
2011-07-25 18:04:49 +00:00
Bill Wendling
b910d857ef Remove dead variable.
llvm-svn: 135923
2011-07-25 18:01:27 +00:00
Bill Wendling
3817554121 After we've modified the prolog to save volatile registers, generate the compact
unwind encoding for that function. This simply crawls through the prolog looking
for machine instrs marked as "frame setup". It can calculate from these what the
compact unwind should look like.

This is currently disabled because of needed linker support. But initial tests
look good.

llvm-svn: 135922
2011-07-25 18:00:28 +00:00
Jim Grosbach
2f728674cf Move some ELF directives into ELF asm parser.
The .local, .hidden, .internal, and .protected are not legal for all supported
file formats (in particular, they're invalid for MachO). Move the parsing for
them into the ELF assembly parser since that's the format they're for.
Similarly, .weak is used by COFF and ELF, but not MachO, so move the parsing
to the COFF and ELF asm parsers. Previously, using any of these directives
on Darwin would result in an assertion failure in the parser; now we get
a diagnostic as we should.

rdar://9827089

llvm-svn: 135921
2011-07-25 17:55:35 +00:00
Oscar Fuentes
a353ff0b83 Builds llvmc and its examples with CMake.
Patch by arrowdodger!

llvm-svn: 135919
2011-07-25 17:25:10 +00:00
Oscar Fuentes
1654c23bb6 Fixes VS support for llvmc mcc16 llvmc example.
Patch by arrowdodger!

llvm-svn: 135918
2011-07-25 17:24:54 +00:00
Jim Grosbach
feb91f5361 Tidy up. 80 columns.
llvm-svn: 135917
2011-07-25 17:11:29 +00:00
Rafael Espindola
c04ccde860 Add uwtable to the langref.
llvm-svn: 135913
2011-07-25 15:27:59 +00:00
Jakob Stoklund Olesen
1b71204b66 Add an RS_Split2 stage used for loop prevention.
This mechanism already exists, but the RS_Split2 stage makes it clearer.

When live range splitting creates ranges that may not be making
progress, they are marked RS_Split2 instead of RS_New. These ranges may
be split again, but only in a way that can be proven to make progress.

For local ranges, that means they must be split into ranges used by
strictly fewer instructions.

For global ranges, region splitting is bypassed and the RS_Split2
ranges go straight to per-block splitting.

llvm-svn: 135912
2011-07-25 15:25:43 +00:00
Jakob Stoklund Olesen
e295d32875 Rename live range stages to better reflect how they are used.
The stage is used to control where a live range is going, not where it
is coming from. Live ranges created by splitting will usually be marked
RS_New, but some are marked RS_Spill to avoid wasting time trying to
split them again.

The old RS_Global and RS_Local stages are merged - they are really the
same thing for local and global live ranges.

llvm-svn: 135911
2011-07-25 15:25:41 +00:00
Frits van Bommel
775ac35cf1 Shorten some expressions by using ArrayRef::slice().
llvm-svn: 135910
2011-07-25 15:13:01 +00:00
Oscar Fuentes
920738256c CMake: generalize the system that creates custom targets for
tablegenning to all libraries and executables.

llvm-svn: 135908
2011-07-25 14:11:55 +00:00
Jay Foad
278089b983 Remove uses of std::vector from TypeBuilder.
llvm-svn: 135906
2011-07-25 10:32:27 +00:00
Jay Foad
4f279f9a6f Use ArrayRef in the (protected) constructors of ConstantArray, ConstantStruct and ConstantVector.
llvm-svn: 135905
2011-07-25 10:14:44 +00:00
Jay Foad
6513dac6e2 Convert GetElementPtrInst to use ArrayRef.
llvm-svn: 135904
2011-07-25 09:48:08 +00:00
Chris Lattner
489601c923 switch Triple to take twines instead of stringrefs.
llvm-svn: 135889
2011-07-24 20:45:08 +00:00
Chris Lattner
0e3472660b Add Twine support for characters, and switch twine to use a union internally
to eliminate some casting.

llvm-svn: 135888
2011-07-24 20:44:30 +00:00
Jakob Stoklund Olesen
e00fa0544e Never extend live ranges for <undef> uses.
llvm-svn: 135886
2011-07-24 20:33:23 +00:00
Jakob Stoklund Olesen
0e4f7f92a2 Correctly handle <undef> tied uses when rewriting after a split.
This fixes PR10463. A two-address instruction with an <undef> use
operand was incorrectly rewritten so the def and use no longer used the
same register, violating the tie constraint.

Fix this by always rewriting <undef> operands with the register a def
operand would use.

llvm-svn: 135885
2011-07-24 20:23:50 +00:00
Frits van Bommel
e238f52543 Omit explicit length here, now that I've had a chance to test this with gcc.
llvm-svn: 135867
2011-07-24 09:53:46 +00:00
Roman Divacky
a07d8900f5 Set PPCII::MO_DARWIN_STUB only on MacOSX < 10.5.
llvm-svn: 135866
2011-07-24 08:22:56 +00:00
Chris Lattner
ec448a31de clarify that opaque is actually a struct type, PR10430
llvm-svn: 135861
2011-07-23 19:59:08 +00:00
Chris Lattner
418c6d4374 how about that, StringRef doesn't allow any mutation, thanks to
Frits for straightening me out.

llvm-svn: 135856
2011-07-23 17:18:57 +00:00
Jakob Stoklund Olesen
e5b06d7a17 Add RAGreedy::calcCompactRegion.
This method computes the edge bundles that should be live when splitting
around a compact region. This is independent of interference.

The function returns false if the live range was already a compact
region, or the compact region doesn't have any live bundles - it would
be the same as splitting around basic blocks.

Compact regions are computed using the normal spill placement code. We
pretend there is interference in all live-through blocks that don't use
the live range. This removes all edges from the Hopfield network used
for spill placement, so it converges instantly.

llvm-svn: 135847
2011-07-23 03:41:57 +00:00
Jakob Stoklund Olesen
53bc83b5d0 Fix bug in SplitEditor::splitLiveThroughBlock when switching registers.
If there is no interference and no last split point, we cannot
enterIntvBefore(Stop) - that function needs a real instruction.

Use enterIntvAtEnd instead for that very easy case.

This code doesn't currently run, it is needed by multi-way splitting.

llvm-svn: 135846
2011-07-23 03:32:26 +00:00
Jakob Stoklund Olesen
ec1337a6e8 Prepare RAGreedy::growRegion for compact regions.
A split candidate can have a null PhysReg which means that it doesn't
map to a real interference pattern. Instead, pretend that all through
blocks have interference.

This makes it possible to generate compact regions where the live range
doesn't go through blocks that don't use it. The live range will still
be live between directly connected blocks with uses.

Splitting around a compact region tends to produce a live range with a
high spill weight, so it may evict a less dense live range.

llvm-svn: 135845
2011-07-23 03:22:33 +00:00
Jakob Stoklund Olesen
72ba470ad6 Add a simple method for marking blocks with interference in and out.
This method matches addLinks - All the listed blocks are considered to
have interference, so they add a negative bias to their bundles.

This could also be done by addConstraints, but that requires building a
separate BlockConstraint array.

llvm-svn: 135844
2011-07-23 03:10:19 +00:00
Jakob Stoklund Olesen
4b2cc68d11 Allow null interference cursors to be queried.
They always report 'no interference'.

llvm-svn: 135843
2011-07-23 03:10:17 +00:00
Benjamin Kramer
40ed4c3e06 Fix a silly off by one.
llvm-svn: 135842
2011-07-23 03:04:46 +00:00
Benjamin Kramer
cbeec9972e We always bounds check the bit set, there is no need to emit zero bytes at the end.
llvm-svn: 135841
2011-07-23 02:49:37 +00:00
Benjamin Kramer
35d7f76497 Add more constness.
llvm-svn: 135838
2011-07-23 01:40:15 +00:00
NAKAMURA Takumi
54f168f5a5 ARMMCTargetDesc.h: Fixup to add DataTypes.h, or uint32_t would be unavailable.
llvm-svn: 135837
2011-07-23 01:16:22 +00:00
Benjamin Kramer
a1e84a1998 Turn the DenseSet in MCRegisterClass into a tblgenerated bit field. This should be faster and smaller.
Goodbye static ctors and dtors!

llvm-svn: 135836
2011-07-23 00:47:49 +00:00
Benjamin Kramer
0fc2a68e8f Give TargetRegisterClass a pointer to the MCRegisterClass and use it to access its data.
This makes TargetRegisterClass slightly slower. Next step will be making contains faster.
Eventually TargetRegisterClass will be killed entirely.

llvm-svn: 135835
2011-07-23 00:47:46 +00:00
Evan Cheng
13d54fc7c9 Move TargetAsmParser.h TargetAsmBackend.h and TargetAsmLexer.h to MC where they belong.
llvm-svn: 135833
2011-07-23 00:45:41 +00:00
NAKAMURA Takumi
0912faa08f CMake: Fix LLVM_NATIVE_TARGETMC in config.h.cmake.
llvm-svn: 135832
2011-07-23 00:45:23 +00:00
NAKAMURA Takumi
e8c8f7fe0e config.h.cmake: Reorder along config.h.in.
llvm-svn: 135831
2011-07-23 00:45:16 +00:00
Andrew Trick
53ed0491c0 Move trip count discovery outside of the generic LoopUnroll helper. This
removes its dependence on canonical induction variables.

llvm-svn: 135829
2011-07-23 00:33:05 +00:00