cc77062194
Pre-regalloc tale duplication. Work in progress.
...
llvm-svn: 90759
2009-12-07 10:15:19 +00:00
92c310cb79
If BB is empty, insert PHI before end() instead of front().
...
llvm-svn: 90744
2009-12-07 03:07:01 +00:00
1bcece70bd
Truncate the arguments of llvm.frameaddress / llvm.returnaddress intrinsics from i32 to platform's largest native type
...
llvm-svn: 90741
2009-12-07 02:28:26 +00:00
35f5646ef0
Remove old DBG_LABEL code.
...
llvm-svn: 90669
2009-12-05 17:56:26 +00:00
6e7073b846
Remove the unused DisableLegalizeTypes option and related code.
...
llvm-svn: 90668
2009-12-05 17:51:33 +00:00
f89986235d
Temporarily revert r90502. It was causing the llvm-gcc bootstrap on PPC to fail.
...
llvm-svn: 90653
2009-12-05 07:30:23 +00:00
0891d759b5
Don't print a space before the : between the file name and line number.
...
And separate the directory and file name with a '/'.
llvm-svn: 90641
2009-12-05 02:00:34 +00:00
3a6164e8ab
Print newlines after printing labels for debug info, so that the output
...
isn't cluttered with things like "Llabel47:Llabel48: movq (%rsi), %xmm3"
llvm-svn: 90638
2009-12-05 01:42:34 +00:00
c82272a7b6
Don't blindly set the debug location for PHI node copies.
...
llvm-svn: 90637
2009-12-05 01:29:04 +00:00
18f94469dc
Make TargetSelectInstruction protected and called from FastISel.cpp
...
instead of SelectionDAGISel.cpp.
llvm-svn: 90636
2009-12-05 01:27:58 +00:00
047a767d74
Remove the target hook TargetInstrInfo::BlockHasNoFallThrough in favor of
...
MachineBasicBlock::canFallThrough(), which is target-independent and more
thorough.
llvm-svn: 90634
2009-12-05 00:44:40 +00:00
0b44cb0d78
Simplify this code: don't call AnalyzeBranch before doing simpler checks.
...
llvm-svn: 90633
2009-12-05 00:32:59 +00:00
02578a3805
The debug information for an LLVM Instruction applies to that Instruction
...
and that Instruction only. Implement this by setting the "current debug position"
back to Unknown after processing each instruction.
llvm-svn: 90632
2009-12-05 00:27:08 +00:00
abc77742c8
Fix this code to use DIScope instead of DICompileUnit, as in r90181.
...
Don't print "SrcLine"; just print the filename and line number, which
is obvious enough and more informative.
llvm-svn: 90631
2009-12-05 00:23:29 +00:00
33004b6302
Don't print the debug directory; it's often long and uninteresting. Omit
...
the column number if it is not known. Handle the case of a missing filename
better.
llvm-svn: 90630
2009-12-05 00:20:51 +00:00
e6d5445dc1
Print a space between the comment character and the text.
...
llvm-svn: 90621
2009-12-04 23:19:55 +00:00
8f0460278c
In TAG_subrange_type, uppder bound is zero indexed.
...
llvm-svn: 90617
2009-12-04 23:10:24 +00:00
15f9df5f4b
Use new interfaces to print spill size.
...
llvm-svn: 90611
2009-12-04 22:46:04 +00:00
9e67255748
Handle recursive PHI's.
...
llvm-svn: 90575
2009-12-04 19:09:10 +00:00
6154dbd5ee
Add a pre-regalloc tail duplication pass.
...
llvm-svn: 90567
2009-12-04 09:42:45 +00:00
0f1cc35c65
Don't try to be cute with undef optimization here. Let ProcessImplicitDefs handle it.
...
llvm-svn: 90566
2009-12-04 09:23:37 +00:00
1602277b70
Add note about a subtle bug in this code. Does not effect the main
...
architectures that LLVM targets, because they don't use this code.
llvm-svn: 90564
2009-12-04 08:42:17 +00:00
ca9cf65455
Also attempt trivial coalescing for live intervals that end in a copy.
...
The coalescer is supposed to clean these up, but when setting up parameters
for a function call, there may be copies to physregs. If the defining
instruction has been LICM'ed far away, the coalescer won't touch it.
The register allocation hint does not always work - when the register
allocator is backtracking, it clears the hints.
This patch takes care of a few more cases that r90163 missed.
llvm-svn: 90502
2009-12-04 00:16:04 +00:00
e156f611ab
- If the reaching definition is an undef and the use is a PHI, add the implicit_def to the end of the source block.
...
- When reaching value is replaced with another, update the cache as well.
llvm-svn: 90501
2009-12-04 00:09:05 +00:00
3b666fef67
Insert composite type DIE into the map before processing type fields. This allows fields to find their context DIE from the map.
...
llvm-svn: 90498
2009-12-03 23:46:57 +00:00
b2c1529d8f
Handle undef values properly.
...
llvm-svn: 90489
2009-12-03 21:51:55 +00:00
8a19371370
Watch out for PHI instruction with no source operands.
...
llvm-svn: 90488
2009-12-03 21:50:58 +00:00
bbd6b6ddf4
Fix ExpandShiftWithUnknownAmountBit, which was completely bogus.
...
Pointed out by Javier Martinez (who also provided a patch). Since
this logic is not used on (for example) x86, I guess nobody noticed.
Tested by generating SHL, SRL, SRA on various choices of i64 for all
possible shift amounts, and comparing with gcc. Since I did this on
x86-32, I had to force the use of ExpandShiftWithUnknownAmountBit.
What I'm saying here is that I don't have a testcase I can add to the
repository.
llvm-svn: 90482
2009-12-03 21:37:32 +00:00
18c7cbd99b
Clean up some loop logic.
...
llvm-svn: 90481
2009-12-03 20:49:10 +00:00
eb57c59b66
Add support to emit debug info for virtual functions and virtual base classes.
...
llvm-svn: 90474
2009-12-03 19:11:07 +00:00
1bbe6be037
Teach tail duplication to update SSA form. Work in progress.
...
llvm-svn: 90432
2009-12-03 08:43:53 +00:00
9655f84662
Don't pull vector sext through both hands of a logical operation, since doing so prevents the fusion of vector sext and setcc into vsetcc.
...
Add a testcase for the above transformation.
Fix a bogus use of APInt noticed while tracking this down.
llvm-svn: 90423
2009-12-03 07:11:29 +00:00
32042f9475
Don't call getValueType() on a null SDValue
...
llvm-svn: 90415
2009-12-03 05:15:35 +00:00
7145382389
Fill out codegen SSA updater. It's not yet tested.
...
llvm-svn: 90395
2009-12-03 02:31:43 +00:00
8392456f1b
Don't hang on to pointers or references after vector::push_back.
...
The MO reference to a MachineOperand can be invalidated by
MachineInstr::addOperand. Don't even use it for debugging.
llvm-svn: 90381
2009-12-03 01:49:56 +00:00
236526dd18
Emit method definition DIE at module level (even for methods with inlined functino body at soure level) so that the debugger can invoke it. This fixes many test failures in gdb test suite.
...
llvm-svn: 90375
2009-12-03 01:25:38 +00:00
a48f44d9ee
improve portability to avoid conflicting with std::next in c++'0x.
...
Patch by Howard Hinnant!
llvm-svn: 90365
2009-12-03 00:50:42 +00:00
a3b825edd1
Fix CMake makefiles
...
llvm-svn: 90354
2009-12-02 22:19:31 +00:00
20e9d030c9
Skeleton for MachineInstr level SSA updater.
...
llvm-svn: 90353
2009-12-02 22:02:52 +00:00
51663f7165
Add MaxStackAlignment.cpp to CMake
...
llvm-svn: 90337
2009-12-02 19:31:07 +00:00
2c3a6c6589
Factor the stack alignment calculations out into a target independent pass.
...
No functionality change.
llvm-svn: 90336
2009-12-02 19:30:24 +00:00
fffbc0c5d9
Don't count PHI instructions toward the limit for tail duplicating a block.
...
llvm-svn: 90326
2009-12-02 17:15:24 +00:00
8c33959df2
Clarify that DIEString does not keep a copy of the string.
...
llvm-svn: 90318
2009-12-02 15:25:16 +00:00
a605b9124a
Reuse existing subprogram DIE.
...
llvm-svn: 90281
2009-12-01 23:07:59 +00:00
732351f732
Fix PR5391: support early clobber physical register def tied with a use (ewwww)
...
- A valno should be set HasRedefByEC if there is an early clobber def in the middle of its live ranges. It should not be set if the def of the valno is defined by an early clobber.
- If a physical register def is tied to an use and it's an early clobber, it just means the HasRedefByEC is set since it's still one continuous live range.
- Add a couple of missing checks for HasRedefByEC in the coalescer. In general, it should not coalesce a vr with a physical register if the physical register has a early clobber def somewhere. This is overly conservative but that's the price for using such a nasty inline asm "feature".
llvm-svn: 90269
2009-12-01 22:25:00 +00:00
b2ae02979f
Add edge source labels to SelectionDAG graphs, now that the graph printing
...
framework omits differentiated edge sources in the case where the labels
are empty strings.
llvm-svn: 90254
2009-12-01 19:20:00 +00:00
8def6e3daf
Minor cleanups.
...
llvm-svn: 90253
2009-12-01 19:16:15 +00:00
939c828604
Trim an unnecessary #include.
...
llvm-svn: 90252
2009-12-01 19:13:27 +00:00
0a2c0bcb14
Clear function specific containers while processing end of a function, even if DW_TAG_subprogram for current function is not found.
...
llvm-svn: 90247
2009-12-01 18:13:48 +00:00
defc47088a
Move PHIElimination::isLiveOut method to LiveVariables.
...
We want LiveVariables clients to use methods rather than accessing the
getVarInfo data structure directly. That way it will be possible to change the
LiveVariables representation.
llvm-svn: 90240
2009-12-01 17:13:31 +00:00
26667abbd3
Use CFG connectedness as a secondary sort key when deciding the order of copy coalescing.
...
This means that well connected blocks are copy coalesced before the less connected blocks. Connected blocks are more difficult to
coalesce because intervals are more complicated, so handling them first gives a greater chance of success.
llvm-svn: 90194
2009-12-01 03:03:00 +00:00
461b937053
Devang pointed out that this code should use DIScope instead of
...
DICompileUnit. This code now prints debug filenames successfully.
llvm-svn: 90181
2009-12-01 00:45:56 +00:00
1d31fc9123
Fix PR5614: parts of a physical register def may be killed the rest.
...
llvm-svn: 90180
2009-12-01 00:44:45 +00:00
ae466efe2e
If pointer type has a name then do not ignore the name.
...
llvm-svn: 90172
2009-11-30 23:56:56 +00:00
020d8d4c63
New virtual registers created for spill intervals should inherit allocation hints from the original register.
...
This helps us avoid silly copies when rematting values that are copied to a physical register:
leaq _.str44(%rip), %rcx
movq %rcx, %rsi
call _strcmp
becomes:
leaq _.str44(%rip), %rsi
call _strcmp
The coalescer will not touch the movq because that would tie down the physical register.
llvm-svn: 90163
2009-11-30 22:55:54 +00:00
598f8ff9e5
Reprioritize tests for tail duplication to be aggressive about indirect
...
branches even when optimizing for code size. Unless we find evidence to the
contrary in the future, the special treatment for indirect branches does not
have a significant effect on code size, and performance still matters with -Os.
llvm-svn: 90147
2009-11-30 18:56:45 +00:00
505ddaa4dc
Remove isProfitableToDuplicateIndirectBranch target hook. It is profitable
...
for all the processors where I have tried it, and even when it might not help
performance, the cost is quite low. The opportunities for duplicating
indirect branches are limited by other factors so code size does not change
much due to tail duplicating indirect branches aggressively.
llvm-svn: 90144
2009-11-30 18:35:03 +00:00
9caf3801ca
Fix last DOTGraphTraits problems in CompilationGraph.
...
llvm-svn: 90136
2009-11-30 13:34:51 +00:00
dd7f2e797f
Remove ShortNames from getNodeLabel in DOTGraphTraits
...
llvm-svn: 90134
2009-11-30 12:38:47 +00:00
90d334032a
Instantiate DefaultDOTGraphTraits
...
llvm-svn: 90133
2009-11-30 12:38:13 +00:00
32f8bb9ed4
Added support to allow clients to custom widen. For X86, custom widen vectors for
...
divide/remainder since these operations can trap by unroll them and adding undefs
for the resulting vector.
llvm-svn: 90108
2009-11-30 02:42:02 +00:00
9594db5693
Rename new TailDuplicationPass to avoid name conflict with the old one.
...
llvm-svn: 89968
2009-11-26 21:38:41 +00:00
595a16300c
When all defs of a vr are implicit_def, delete all of the defs.
...
llvm-svn: 89905
2009-11-26 00:32:36 +00:00
2d4ff12da9
Split tail duplication into a separate pass. This is needed to avoid
...
running tail duplication when doing branch folding for if-conversion, and
we also want to be able to run tail duplication earlier to fix some
reg alloc problems. Move the CanFallThrough function from BranchFolding
to MachineBasicBlock so that it can be shared by TailDuplication.
llvm-svn: 89904
2009-11-26 00:32:21 +00:00
44df27e964
ProcessImplicitDefs should watch out for invalidated iterator and extra implicit operands on copies.
...
llvm-svn: 89880
2009-11-25 21:13:39 +00:00
4cd30817d3
Avoid some possibly unsafe uses of StringRef::data().
...
llvm-svn: 89873
2009-11-25 18:26:09 +00:00
2d9caf9fe5
Use StringRef (again) in DebugInfo interface.
...
llvm-svn: 89866
2009-11-25 17:36:49 +00:00
d4d40670e8
Refactor target hook for tail duplication as requested by Chris.
...
Make tail duplication of indirect branches much more aggressive (for targets
that indicate that it is profitable), based on further experience with
this transformation. I compiled 3 large applications with and without
this more aggressive tail duplication and measured minimal changes in code
size. ("size" on Darwin seems to round the text size up to the nearest
page boundary, so I can only say that any code size increase was less than
one 4k page.) Radar 7421267.
llvm-svn: 89814
2009-11-24 23:35:49 +00:00
b3e0168428
Use StringRef instead of std::string in DIEString.
...
llvm-svn: 89793
2009-11-24 19:42:17 +00:00
21c2e100ba
Remove DebugLabelFolder pass. It is not used by dwarf writer anymore.
...
llvm-svn: 89790
2009-11-24 19:37:07 +00:00
c8654eb64e
Swith to pubtypes section before emitting pub types.
...
llvm-svn: 89787
2009-11-24 19:18:41 +00:00
04d2f2d192
Emit pubtypes.
...
llvm-svn: 89725
2009-11-24 01:14:22 +00:00
3f3a8b1c7e
<rdar://problem/6721894>. Allow multiple registers to be renamed together (super and sub) if necessary to break an anti-dependence.
...
llvm-svn: 89722
2009-11-24 00:59:08 +00:00
de5dea869f
Remove ISD::DEBUG_LOC and ISD::DBG_LABEL, which are no longer used.
...
Note that "hasDotLocAndDotFile"-style debug info was already broken;
people wanting this functionality should implement it in the
AsmPrinter/DwarfWriter code.
llvm-svn: 89711
2009-11-23 23:20:51 +00:00
3650f4ed0c
Simplify this code.
...
llvm-svn: 89702
2009-11-23 21:30:55 +00:00
2e3f187cbd
Print the debug info line and column in MachineInstr::print even when there's
...
no filename. This situation is apparently fairly common right now.
llvm-svn: 89701
2009-11-23 21:29:08 +00:00
b5b51598ba
Revert r89487.
...
llvm-svn: 89686
2009-11-23 18:43:37 +00:00
9d72cbf2d5
Move CopyCatchInfo into FunctionLoweringInfo.cpp too, for consistency.
...
llvm-svn: 89683
2009-11-23 18:12:11 +00:00
1a6c47f1cb
Rename SelectionDAGLowering to SelectionDAGBuilder, and rename
...
SelectionDAGBuild.cpp to SelectionDAGBuilder.cpp.
llvm-svn: 89681
2009-11-23 18:04:58 +00:00
91aad4b834
Move RegsForValue to an anonymous namespace, since it is only used
...
in this file.
llvm-svn: 89675
2009-11-23 17:46:23 +00:00
ad97b3dbd0
Move some more code out of SelectionDAGBuild.cpp and into
...
FunctionLoweringInfo.cpp.
llvm-svn: 89674
2009-11-23 17:42:46 +00:00
9b6515794f
Update CMake file.
...
llvm-svn: 89671
2009-11-23 17:26:04 +00:00
a3624b6099
Move the FunctionLoweringInfo class and some related utility functions out
...
of SelectionDAGBuild.h/cpp into its own files, to help separate
general lowering logic from SelectionDAG-specific lowering logic.
llvm-svn: 89667
2009-11-23 17:16:22 +00:00
15a1287c1f
Pull LLVMContext out of PromoteMemToReg.
...
llvm-svn: 89645
2009-11-23 03:50:44 +00:00
bf6d35893f
Add getFrameIndexReference() to TargetRegisterInfo, which allows targets to
...
tell debug info which base register to use to reference a frame index on a
per-index basis. This is useful, for example, in the presence of dynamic
stack realignment when local variables are indexed via the stack pointer and
stack-based arguments via the frame pointer.
llvm-svn: 89620
2009-11-22 20:14:00 +00:00
00e9c6103b
80-column cleanup
...
llvm-svn: 89612
2009-11-22 19:20:36 +00:00
8a3fdae0aa
Teach MachineBasicBlock::updateTerminator() to handle a failing TII->ReverseBranchCondition(Cond) call.
...
This fixes the MallocBench/cfrac test case regression.
llvm-svn: 89608
2009-11-22 18:28:04 +00:00
042483efb1
remove trailing whitespace
...
llvm-svn: 89567
2009-11-21 23:12:12 +00:00
1f4062f194
Allow target to disable if-converting predicable instructions. e.g. NEON instructions under ARM mode.
...
llvm-svn: 89541
2009-11-21 06:20:26 +00:00
930143b0dc
Cosmetic changes, which were long overdue, in DwarfDebug.cpp.
...
llvm-svn: 89537
2009-11-21 02:48:08 +00:00
ed85e12da6
We are not using DBG_STOPPOINT anymore.
...
llvm-svn: 89536
2009-11-21 02:46:55 +00:00
bc630aca39
Be more clever about calculating live variables through new basic blocks.
...
When splitting a critical edge, the registers live through the edge are:
- Used in a PHI instruction, or
- Live out from the predecessor, and
- Live in to the successor.
This allows the coalescer to eliminate even more phi joins.
llvm-svn: 89530
2009-11-21 02:05:21 +00:00
e1bbc33355
Remove dead code.
...
llvm-svn: 89522
2009-11-21 00:54:03 +00:00
b91eba382d
When generating a vector the really slow way, via loads
...
and stores, handle the case where the element size is not
a valid target type correctly (PPC).
llvm-svn: 89521
2009-11-21 00:53:23 +00:00
92e8c655b2
There is no need to use FoldingSet to unique DIEs.
...
DIEs are created from MDNode, which are already uniqued. And DwarfDebug already uses ValueMaps to find and use existing DIE for a given MDNode.
llvm-svn: 89518
2009-11-21 00:31:03 +00:00
5305dc0be1
Restructure code to allow renaming of multiple-register groups for anti-dep breaking.
...
llvm-svn: 89511
2009-11-20 23:33:54 +00:00
73f9a9e2c8
Enable hoisting load from constant memories.
...
llvm-svn: 89510
2009-11-20 23:31:34 +00:00
7a6611793f
Target-independent support for TargetFlags on BlockAddress operands,
...
and support for blockaddresses in x86-32 PIC mode.
llvm-svn: 89506
2009-11-20 23:18:13 +00:00
e064ad4741
Do not hold on to a map slot while new entries may be inserted into the map.
...
Use ValueMap, instead of std::map.
llvm-svn: 89490
2009-11-20 21:37:22 +00:00
f46c37c0de
Cleanups.
...
Make things a little more efficient as suggested by Evan.
llvm-svn: 89489
2009-11-20 21:13:27 +00:00
85121a3bda
There is no need to emit source location info for DW_TAG_pointer_type.
...
llvm-svn: 89487
2009-11-20 21:05:37 +00:00
b39a9fd3ed
Add option -licm-const-load to hoist all loads from constant memory.
...
llvm-svn: 89477
2009-11-20 19:55:37 +00:00
80a03cc0b1
Remove some old experimental code that is no longer needed. Remove additional, speculative scheduling pass as its cost did not translate into significant performance improvement. Minor tweaks.
...
llvm-svn: 89471
2009-11-20 19:32:48 +00:00
b4c14ebfbb
More consistent labelling of basic blocks in debug output
...
llvm-svn: 89470
2009-11-20 18:54:59 +00:00
cc0a0cb4b7
Fix PR5558, which was caused by a wrong fix for PR3393 (see commit 63048),
...
which was an expensive checks failure due to a bug in the checking. This
patch in essence reverts the original fix for PR3393, and refixes it by a
tweak to the way expensive checking is done.
llvm-svn: 89454
2009-11-20 10:45:10 +00:00
20c8ab655e
Fix fast-isel to avoid selecting the return instruction if a
...
tail call has been encountered.
llvm-svn: 89444
2009-11-20 02:51:26 +00:00
f4fe57ad90
Add an experimental option to run gep-splitting and no-load GVN
...
just before codegen.
llvm-svn: 89439
2009-11-20 02:03:44 +00:00
2bbeaa8774
Add MachineBasicBlock::getName, and use it in place of getBasicBlock()->getName.
...
Fix debug code that assumes getBasicBlock never returns NULL.
llvm-svn: 89428
2009-11-20 01:17:03 +00:00
c6e434573c
Removed references to LiveStacks from Spiller.* . They're no longer needed.
...
llvm-svn: 89422
2009-11-20 00:53:30 +00:00
dd1c6198d4
Fix a couple of problems with maintaining liveness information for antidep breaking.
...
llvm-svn: 89404
2009-11-19 23:12:37 +00:00
331e6880a5
Place new basic blocks immediately after their predecessor when splitting
...
critical edges in PHIElimination.
This has a huge impact on regalloc performance, and we recover almost all of
the 10% compile time regression that edge splitting introduced.
llvm-svn: 89381
2009-11-19 19:42:16 +00:00
31c74dbb10
Reverting the EH table patches.
...
$ svn merge -c -89279 https://llvm.org/svn/llvm-project/llvm/trunk
--- Reverse-merging r89279 into '.':
U lib/CodeGen/AsmPrinter/DwarfException.cpp
U lib/Target/TargetLoweringObjectFile.cpp
$ svn merge -c -89270 https://llvm.org/svn/llvm-project/llvm/trunk
--- Reverse-merging r89270 into '.':
G lib/CodeGen/AsmPrinter/DwarfException.cpp
G lib/Target/TargetLoweringObjectFile.cpp
llvm-svn: 89379
2009-11-19 19:21:09 +00:00
50909c0595
Fix a small bug.
...
Fix one case we missed to make sure we reserve registers from
allocation.
llvm-svn: 89376
2009-11-19 19:09:39 +00:00
f37756d609
Add support for spreading register allocation.
...
Add a -linearscan-skip-count argument (default to 0) that tells the
allocator to remember the last N registers it allocated and skip them
when looking for a register candidate. This tends to spread out
register usage and free up post-allocation scheduling at the cost of
slightly more register pressure. The primary benefit is the ability
to backschedule reloads.
This is turned off by default.
llvm-svn: 89356
2009-11-19 15:55:49 +00:00
6912be4ad6
Added a new Spiller implementation which wraps LiveIntervals::addIntervalsForSpills.
...
All spiller calls in RegAllocLinearScan now go through the new Spiller interface.
The "-new-spill-framework" command line option has been removed. To use the trivial in-place spiller you should now pass "-spiller=trivial -rewriter=trivial".
(Note the trivial spiller/rewriter are only meant to serve as examples of the new in-place modification work. Enabling them will yield terrible, though hopefully functional, code).
llvm-svn: 89311
2009-11-19 04:15:33 +00:00
4b384b0294
The "ReadOnlyWithRel" enum seems to apply more to what Darwin does with the EH
...
exception table than DataRel.
llvm-svn: 89279
2009-11-19 00:09:14 +00:00
b3c5aa20ea
There should be no need to keep renumbering blocks during tail duplication.
...
llvm-svn: 89275
2009-11-18 23:48:57 +00:00
07e6e2c6f8
Attempt #2 :
...
Place the EH table in the __TEXT section on MachO. It saves space.
llvm-svn: 89270
2009-11-18 23:18:46 +00:00
108aadf972
Tail duplication still needs to iterate. Duplicating new instructions onto
...
the tail of a block may make that block a new candidate for duplication.
llvm-svn: 89264
2009-11-18 22:52:37 +00:00
5d87f7d5f8
Add another statistic to measure code size due to tail duplication.
...
llvm-svn: 89254
2009-11-18 22:12:31 +00:00
575c3f3d72
Fix PR5300.
...
When TwoAddressInstructionPass deletes a dead instruction, make sure that all
register kills are accounted for. The 2-addr register does not get special
treatment.
llvm-svn: 89246
2009-11-18 21:33:35 +00:00
9cbffd2155
Allow the machine verifier to be run outside the PassManager.
...
Verify LiveVariables information when present.
llvm-svn: 89241
2009-11-18 20:36:57 +00:00
520039fcb6
Remove the -early-coalescing option
...
llvm-svn: 89240
2009-11-18 20:36:47 +00:00
e9e0cd4e27
Fixed the in-place spiller and trivial rewriter, which had been broken by the recent SlotIndexes work.
...
llvm-svn: 89238
2009-11-18 20:31:20 +00:00
8984e6ec8c
Add statistics for tail duplication.
...
llvm-svn: 89225
2009-11-18 19:29:37 +00:00
736888f7fa
Don't require LiveVariables for PHIElimination. Enable critical edge splitting
...
when LiveVariables is available.
The -split-phi-edges is now gone, and so is the hack to disable it when using
the local register allocator. The PHIElimination pass no longer has
LiveVariables as a prerequisite - that is what broke the local allocator.
Instead we do critical edge splitting when possible - that is when
LiveVariables is available.
llvm-svn: 89213
2009-11-18 18:01:35 +00:00
290e9a47a9
Add a target hook to allow changing the tail duplication limit based on the
...
contents of the block to be duplicated. Use this for ARM Cortex A8/9 to
be more aggressive tail duplicating indirect branches, since it makes it
much more likely that they will be predicted in the branch target buffer.
Testcase coming soon.
llvm-svn: 89187
2009-11-18 03:34:27 +00:00
4797e58d6b
Fix inverted test and add testcase from failing self-host.
...
llvm-svn: 89167
2009-11-18 00:02:18 +00:00
31e24b2d0c
Disable -split-phi-edges to unbreak the buildbots
...
llvm-svn: 89142
2009-11-17 21:23:49 +00:00
f992022e54
Never call UpdateTerminator() when AnalyzeBranch would fail.
...
llvm-svn: 89139
2009-11-17 20:46:00 +00:00
57be2f2c40
Add a WriteAsOperand for MachineBasicBlock so MachineLoopInfo dump looks sane.
...
llvm-svn: 89130
2009-11-17 19:19:59 +00:00
0229536301
Fix comment.
...
llvm-svn: 89129
2009-11-17 19:19:01 +00:00
fffff88a3c
Enable -split-phi-edges by default, except when -regalloc=local.
...
The local register allocator doesn't like it when LiveVariables is run.
We should also disable edge splitting under -O0, but that has to wait a bit.
llvm-svn: 89125
2009-11-17 19:15:50 +00:00
eeac7ce769
Remove a special case for tail merging that seems to be both broken and
...
unnecessary. It is broken because the "isIdenticalTo" check should be
negated. If that is fixed, this code causes the CodeGen/X86/tail-opts.ll
test to fail, in the dont_merge_oddly function. And, I confirmed that the
regression is real -- the generated code is worse. As far as I can tell,
that tail-opts.ll test is checking for what this code is supposed to handle
and we're doing the right thing anyway.
llvm-svn: 89121
2009-11-17 18:30:09 +00:00
2650723d81
Set MadeChange instead of MadeChangeThisIteration.
...
llvm-svn: 89114
2009-11-17 18:04:15 +00:00
724d8a45c1
Update a comment, now that tail duplication happens after other branch
...
folding optimizations.
llvm-svn: 89109
2009-11-17 17:40:31 +00:00
bd22f1965d
Perform tail duplication only once, after tail merging is complete.
...
It was too difficult to keep the heuristics for merging and duplication
consistent.
llvm-svn: 89105
2009-11-17 17:06:18 +00:00
84efacfaad
Revert 89021. It's miscompiling llvm-gcc driver driver at -O0.
...
llvm-svn: 89082
2009-11-17 09:55:52 +00:00
a1c09d674e
Remove VISIBILITY_HIDDEN from the classes in this directory. Fixes bug 5507.
...
llvm-svn: 89075
2009-11-17 09:17:08 +00:00
b7993d61f2
Revert r88939.
...
llvm-svn: 89066
2009-11-17 08:11:44 +00:00
bc0020174b
Refactor the code that creates the "dot-label" difference. This may be used in
...
more than one place. No intended functionality change.
llvm-svn: 89024
2009-11-17 01:23:53 +00:00
9f0d55d8d8
Enable -split-phi-edges by default
...
llvm-svn: 89021
2009-11-17 01:07:22 +00:00
82e80019a5
Remove the optimizations that convert BRCOND and BR_CC into
...
unconditional branches or fallthroghes. Instcombine/SimplifyCFG
should be simplifying branches with known conditions.
This fixes some problems caused by these transformations not
updating the MachineBasicBlock CFG.
llvm-svn: 89017
2009-11-17 00:47:23 +00:00
9b5eea30f7
Initialize the new AsmPrinterFlags field to 0, fixing uses of
...
uninitialized memory.
llvm-svn: 88985
2009-11-16 22:49:38 +00:00
f3d7c08e34
Revert r88939.
...
llvm-svn: 88973
2009-11-16 21:53:40 +00:00
93f529d514
Sink a #include <map> to where it's actually needed.
...
llvm-svn: 88956
2009-11-16 20:40:47 +00:00
6b3f32e6d7
Fix a typo in a comment.
...
llvm-svn: 88953
2009-11-16 20:35:59 +00:00
44724c4122
Fix a comment.
...
llvm-svn: 88940
2009-11-16 19:33:27 +00:00
2ffd9935a1
Add VISIBILITY_HIDDEN marker.
...
llvm-svn: 88939
2009-11-16 19:20:48 +00:00
1a8ea98b76
Fix some comments.
...
llvm-svn: 88932
2009-11-16 18:54:08 +00:00
Evan Cheng