813384951e
Use Intrinsic::getDeclaration in more places.
...
llvm-svn: 49338
2008-04-07 13:45:04 +00:00
1416ebf1fe
The "stacksave is not nounwind problem" no longer
...
needs to be fixed here - a previous commit made sure
that intrinsics always get the right attributes.
So remove no-longer needed code, and while there use
Intrinsic::getDeclaration rather than getOrInsertFunction.
llvm-svn: 49337
2008-04-07 13:43:58 +00:00
fbc6adcc59
Use Intrinsic::getDeclaration to get hold of
...
intrinsics. Fix up the argument type (should
be i8*, was an array*).
llvm-svn: 49336
2008-04-07 13:41:19 +00:00
a8ff6ca2e3
Make sure that intrinsics automagically get the
...
right parameter attributes no matter how they
are obtained.
llvm-svn: 49335
2008-04-07 13:39:11 +00:00
51f532f92d
Re-commit of the r48822, where the infinite looping problem discovered
...
by Dan Gohman is fixed.
llvm-svn: 49330
2008-04-07 10:06:32 +00:00
0c1e634cbb
Make GVN more memory efficient, particularly on code that contains a large number of
...
allocations, which GVN can't optimize anyways.
llvm-svn: 49329
2008-04-07 09:59:07 +00:00
87e484f08b
Mark calls to llvm.stacksave, llvm.stackrestore as
...
nounwind. When such calls are inlined into something
else that is invoked, they were getting changed to invokes,
which is badness.
llvm-svn: 49299
2008-04-07 00:08:48 +00:00
92f493dc72
regenerate
...
llvm-svn: 49293
2008-04-06 23:07:54 +00:00
982e8502c9
fix warnings with assertions disabled.
...
llvm-svn: 49285
2008-04-06 21:50:58 +00:00
4db1f62d84
Silence warning when no assertions.
...
llvm-svn: 49284
2008-04-06 21:46:45 +00:00
a39cfc5c5b
silence a warning when assertions are disabled.
...
llvm-svn: 49283
2008-04-06 21:44:08 +00:00
dd8b7a0a47
fix a warning
...
llvm-svn: 49282
2008-04-06 21:42:13 +00:00
613d7afe64
Prefer to expand mask for xor to -1, so we have a chance to turn it into a not.
...
If it cannot be expanded, it will keep the old behaviour and try to shrink the constant.
Part of enhancement for PR2191.
llvm-svn: 49280
2008-04-06 21:23:02 +00:00
e9ecc68d8f
API changes for class Use size reduction, wave 1.
...
Specifically, introduction of XXX::Create methods
for Users that have a potentially variable number of
Uses.
llvm-svn: 49277
2008-04-06 20:25:17 +00:00
8519c0ead8
disable this for now.
...
llvm-svn: 49248
2008-04-05 04:12:01 +00:00
b5fdc923d3
1. IMPLICIT_DEF can *re-define* any register.
...
2. Coalescer can now create an interesting situation where a register def can
reaches itself without being killed.
llvm-svn: 49246
2008-04-05 01:27:09 +00:00
f77b5ef3d0
Favors pshufd over shufps when shuffling elements from one vector. pshufd is faster than shufps.
...
llvm-svn: 49244
2008-04-05 00:30:36 +00:00
0ce4a7cc44
Make sure both PendingLoads and PendingExports are flushed
...
before an invoke. Failure to do this causes references in
the landing pad to variables that were not set. Fixes
g++.dg/eh/delayslot1.C
g++.dg/eh/fp-regs.C
g++.old-deja/g++.brendan/eh1.C
llvm-svn: 49243
2008-04-04 23:48:31 +00:00
b20e659770
strdup needs <cstring>. This fixes a build error with g++-4.3.
...
llvm-svn: 49218
2008-04-04 16:08:00 +00:00
318d3ef88e
Provide an initial cut at exposing JIT compiled symbols to performance
...
tools. This is currently only enabled on the mac, but could easily be
supported by other hosts that are interested.
llvm-svn: 49207
2008-04-04 05:51:42 +00:00
ee1817300a
Make ExecutionEngine::updateGlobalMapping return the old mapping.
...
llvm-svn: 49206
2008-04-04 04:47:41 +00:00
14bee50e06
Undo PHI elimination copy placement patch. This causes coalescing (performace) issues.
...
llvm-svn: 49198
2008-04-04 01:20:05 +00:00
823017fdd1
This is done.
...
llvm-svn: 49197
2008-04-04 01:19:03 +00:00
bfb7246fb6
if some functions don't have debug info, we were outputing the same label at the start of each of those functions. This makes assemblers unhappy
...
llvm-svn: 49176
2008-04-03 17:37:43 +00:00
58936a48ee
- Turn copies of implicit_def into implicit_def instructions.
...
- Be smarter about coalescing copies from implicit_def.
llvm-svn: 49168
2008-04-03 16:41:54 +00:00
6d07b625aa
Special handling of zero-sized live intervals.
...
llvm-svn: 49167
2008-04-03 16:40:27 +00:00
20aed56504
- Treat a live range defined by an implicit_def as a zero-sized one.
...
- Eliminate an implicit_def when it's being spilled.
llvm-svn: 49166
2008-04-03 16:39:43 +00:00
aacf4f15b3
- PHI elimination also eliminates implicit_def that fits into a PHI node rather than copying it.
...
- Be (slightly) smarter about where to place the copies.
llvm-svn: 49165
2008-04-03 16:38:20 +00:00
916802a78e
Start of a series of patches related to implicit_def.
...
There is no point in creating a long live range defined by an implicit_def. Scheduler now duplicates implicit_def instruction for each of its uses. Therefore, if an implicit_def node has multiple uses, it will become a number of very short live ranges, rather than a long one. This will make coalescer's job easier.
llvm-svn: 49164
2008-04-03 16:36:07 +00:00
1d0436cdb0
Re-implemented Path::createDirectoryOnDisk (for Unix).
...
This method allows one to create a directory, and optionally create all parent
directories that do not exist.
The original implementation would require that *all* directories along a path
are writable by the user, including directories that already exist. For example,
suppose we wanted to create the directory "/tmp/foo/bar", and the directory
"/tmp" already exists, but not "/tmp/foo". Since "/tmp" is writable by all
users, the original implementation would work, and create "/tmp/foo", followed
by "/tmp/bar".
A problem occurred, however if one wanted to created the directory
"/Users/myuser/bar" (or equivalently "/home/myuser/bar"), and "/Users/myuser"
already existed and is writable by the current user. The directory
"/User/myuser" is writable by the user, but "/User" is not. The original
implementation of createDirectoryOnDisk would return with failure since "/User"
is not writable, even though "/User/mysuser" is writable.
The new implementation works by recursively creating parents as needed, and thus
doesn't need to check the permissions on every directory in a path.
llvm-svn: 49162
2008-04-03 16:11:31 +00:00
6c66bd368e
Re-enable SSE4.
...
llvm-svn: 49158
2008-04-03 08:53:29 +00:00
6db4b4cc65
Fix x86-64 encoding bug. REX prefix must always follow 0x0F prefix. For example, extractps in 64bit mode: 66 REX 0F 3A 17, not 66 0F 3A REX 17.
...
llvm-svn: 49157
2008-04-03 08:53:17 +00:00
d9129d1de3
Cosmetic
...
llvm-svn: 49156
2008-04-03 07:45:18 +00:00
3063c5546e
Temporarily disabling SSE4 until we fix the encoding issues.
...
llvm-svn: 49129
2008-04-03 04:49:54 +00:00
025cea1126
Backing out 48222 temporarily.
...
llvm-svn: 49124
2008-04-03 03:13:16 +00:00
bd72cea737
Suppress the 128-bit integer typedef on 32-bit targets, because
...
it causes compile errors.
llvm-svn: 49122
2008-04-02 23:52:49 +00:00
491557712a
Make EH work with unnamed functions. Reenable running
...
StripSymbols when EH is on.
llvm-svn: 49110
2008-04-02 20:10:52 +00:00
39d8b26322
Partial CBackend support for 128-bit integers. This is needed
...
now that llvm-gcc is lowering appropriately-sized struct returns
to i128 on x86-64.
llvm-svn: 49109
2008-04-02 19:40:14 +00:00
586740f401
Iterators folloring a SmallVector erased element are invalidated so
...
don't access cached iterators from after the erased element.
Re-apply 49056 with SmallVector support.
llvm-svn: 49106
2008-04-02 18:24:46 +00:00
d8616064d8
Now that I am told MachineRegisterInfo also tracks physical register uses / defs, I can do away with the horribleness I introduced a while back. It's impossible to detect if there is any use of a physical register below an instruction (and before any def of the register) with some cheap book keeping.
...
llvm-svn: 49105
2008-04-02 18:04:08 +00:00
be3d44c3cb
Remove #include<map> from LiveVariables.h. Not referenced.
...
llvm-svn: 49099
2008-04-02 17:23:50 +00:00
8780ecbbac
Cosmetic changes per EH patch review feedback.
...
llvm-svn: 49096
2008-04-02 17:04:45 +00:00
20c9e4cbee
Add new CC lowering rule: provide a list of registers, which can be 'shadowed',
...
when some another register is used for argument passing.
Currently is used on Win64.
llvm-svn: 49079
2008-04-02 05:23:57 +00:00
2412158111
In some situations, we need to check for local interferences between the PHI
...
node and its inputs.
llvm-svn: 49070
2008-04-02 03:00:13 +00:00
edfc2eb558
Correctly mark a valno that was previous defined by a PHI node as having an
...
unknown defining inst after PHI elimination.
llvm-svn: 49069
2008-04-02 02:12:45 +00:00
fd967cf3fa
Recommitting EH patch; this should answer most of the
...
review feedback.
-enable-eh is still accepted but doesn't do anything.
EH intrinsics use Dwarf EH if the target supports that,
and are handled by LowerInvoke otherwise.
The separation of the EH table and frame move data is,
I think, logically figured out, but either one still
causes full EH info to be generated (not sure how to
split the metadata correctly).
MachineModuleInfo::needsFrameInfo is no longer used and
is removed.
llvm-svn: 49064
2008-04-02 00:25:04 +00:00
ac38d444e2
1. Drop default inline threshold back down to 200.
...
2. Do not use # of basic blocks as part of the cost computation since it doesn't really figure into function size.
3. More aggressively inline function with vector code.
llvm-svn: 49061
2008-04-01 23:59:29 +00:00
052838c55d
Reverting 49056 due to the build being broken.
...
llvm-svn: 49060
2008-04-01 23:41:44 +00:00
b86595fb0a
ReMat of load from stub in pic mode extends the life of pic base. Currently spiller doesn't do a good job of estimating the impact. Disable for now.
...
llvm-svn: 49059
2008-04-01 23:26:12 +00:00
7f7edc3824
Iterators folloring a SmallVector erased element are invalidated so
...
don't access cached iterators from after the erased element.
llvm-svn: 49056
2008-04-01 22:14:23 +00:00
fc9f25dab0
Fix compilation errors on MSVC. Patch by Argiris Kirtzidis!
...
llvm-svn: 49055
2008-04-01 22:09:20 +00:00
19a6dd9f2a
Remove unnecessary and non-deterministic checking code. Re-enable remat of load from gv stub.
...
llvm-svn: 49054
2008-04-01 21:38:20 +00:00
985a0b51d7
Re-materialization is for uses only.
...
llvm-svn: 49053
2008-04-01 21:37:32 +00:00
cb9f8f6e4e
Don't use __bzero for memset if the second argument isn't zero.
...
llvm-svn: 49050
2008-04-01 20:56:18 +00:00
980d7200c1
Speculatively micro-optimize memory-zeroing calls on Darwin 10.
...
llvm-svn: 49048
2008-04-01 20:38:36 +00:00
5e4e051c2a
Revert 49006 for the moment.
...
llvm-svn: 49046
2008-04-01 20:00:57 +00:00
49dd9f16a9
Don't dereference MBB->end().
...
llvm-svn: 49043
2008-04-01 18:05:08 +00:00
6f2ffdb73f
Change the MemoryBuffer::getFile* methods to take just a pointer to the
...
start of a filename, not a filename+length. All clients can produce a
null terminated name, and the system api's require null terminated
strings anyway.
llvm-svn: 49041
2008-04-01 18:04:03 +00:00
306e3dcff4
Disabling remat of load from gv stub (temporarily) again to fix llvmgcc bootstrap miscompare.
...
llvm-svn: 49037
2008-04-01 07:33:13 +00:00
ba98fca5e5
add some #includes.
...
llvm-svn: 49036
2008-04-01 06:25:23 +00:00
18356d8857
MappedFile is dead, remove it.
...
llvm-svn: 49035
2008-04-01 06:20:44 +00:00
4c5e15f7da
Implement Path::MapInFilePages/UnMapFilePages on unix, which
...
provides fast MappedFile::getFile for large files.
llvm-svn: 49034
2008-04-01 06:16:24 +00:00
a542518315
Reimplement MemoryBuffer::getFile with three enhancements:
...
1) stop using MappedFile.
2) if profitable use the sys::path::MapInFilePages api to
read the file.
3) otherwise fallback to read.
When sys::path::MapInFilePages is implemented, this provides
several benefits:
#1 : this avoids fragmenting memory for small files.
#2 : this avoids extraneous stat calls when the file size is known.
#3 : this only keeps the file descriptor open while reading the
file, not for the duration of the lifetime of the memory
buffer. This fixes a serious clang FD 'leak' problem.
I believe that this will work on a win32 machine, but I don't have
one to test on. I'd appreciate it if someone could check.
llvm-svn: 49031
2008-04-01 06:05:21 +00:00
3089e1d82e
Stub out some sys::Path::MapInFilePages/UnMapFilePages methods.
...
llvm-svn: 49030
2008-04-01 06:00:12 +00:00
d4310a27aa
change the archive stuff to use MemoryBuffer instead of mappedfile.
...
MemoryBuffer is higher level and more closely matches the model
needed.
llvm-svn: 49029
2008-04-01 04:26:46 +00:00
4ac7bb0294
rewrite SourceFile to be in terms of MemoryBuffer, not MappedFile.
...
llvm-svn: 49027
2008-04-01 03:59:34 +00:00
64959dcf03
Make MappedFile::map return a const correct pointer, don't leak address space on Unix platforms.
...
llvm-svn: 49026
2008-04-01 03:49:38 +00:00
bcfa564759
Remove the MappedFile::charBase member, rename base -> getBase() and
...
make getBase() return a const-correct pointer.
llvm-svn: 49025
2008-04-01 03:40:53 +00:00
314a1414d0
Change DiffFilesWithTolerance to be written in terms of MemoryBuffer,
...
not an mmapped file. This more closely matches its requirements and
provides an implicitly null terminated buffer, something this
routine had to emulate itself before.
llvm-svn: 49024
2008-04-01 03:39:49 +00:00
62385ba3a0
Remove MappedFile support for mapping files for write and exec
...
and shared. This complicates the design, is not used, and probably
doesn't even work.
llvm-svn: 49022
2008-04-01 03:10:22 +00:00
3d7127b3e3
add missing #includes
...
llvm-svn: 49020
2008-04-01 02:58:05 +00:00
0bd72c5ccd
More soft fp fixes.
...
llvm-svn: 49016
2008-04-01 02:18:22 +00:00
4cabe4b452
Pasto.
...
llvm-svn: 49014
2008-04-01 02:00:09 +00:00
611abc03ed
Add comment.
...
llvm-svn: 49013
2008-04-01 01:51:26 +00:00
86e476b7cb
Unbreak ARM / Thumb soft FP support.
...
llvm-svn: 49012
2008-04-01 01:50:16 +00:00
efa81a6979
Accept 'y' constraint (MMX) in inline asm.
...
llvm-svn: 49011
2008-04-01 00:57:48 +00:00
efecfbc113
cleanup the MappedFile API and comments. This removes and updates
...
tons of out of date comments (really nothing throws here!) and fixes
some other fairly glaring issues: "size" used to return the size of
the file *and* change it, depending on how you called it.
llvm-svn: 49009
2008-04-01 00:53:25 +00:00
1fa4dca692
remove DEFINING_FILE_FOR for MappedFile.h
...
llvm-svn: 49008
2008-04-01 00:35:55 +00:00
7d02cf3c9c
Emit exception handling info for functions which are
...
not marked nounwind, or for all functions when -enable-eh
is set, provided the target supports Dwarf EH.
llvm-gcc generates nounwind in the right places; other FEs
will need to do so also. Given such a FE, -enable-eh should
no longer be needed.
llvm-svn: 49006
2008-03-31 23:40:23 +00:00
e4f77c69ac
It's not safe to fold a load from GV stub or constantpool into a two-address use.
...
llvm-svn: 49002
2008-03-31 23:19:51 +00:00
f8be3836e3
Actually disable crash reporting on Mac OS X, returning bugpoint to speedy
...
crash miscompilations.
llvm-svn: 49000
2008-03-31 22:19:25 +00:00
3717cdaf22
Set blockBegin to point to the beginning of the block,
...
not the end.
llvm-svn: 48999
2008-03-31 22:08:00 +00:00
ed6e34fe41
Move reMaterialize() from TargetRegisterInfo to TargetInstrInfo.
...
llvm-svn: 48995
2008-03-31 20:40:39 +00:00
f549b26254
Fix a DAGCombiner optimization to respect volatile qualification.
...
llvm-svn: 48994
2008-03-31 20:32:52 +00:00
4a0da98825
Expose Function::viewCFG and Function::viewCFGOnly to bindings.
...
llvm-svn: 48982
2008-03-31 16:22:09 +00:00
1973a46cd3
Re-apply 48911.
...
llvm-svn: 48977
2008-03-31 07:54:19 +00:00
73d7c3bfba
The support for remat of instructions with a register operand is hackish, to say the least. Since the register operand guaranteed to be PIC base and that it is already live at all uses, we are making sure it will not be spilled after its uses are rematerialized for both performance and correctness reasons.
...
llvm-svn: 48976
2008-03-31 07:53:30 +00:00
f28fc71c93
Fix a major bug in the DFS calculation. Thanks for Christopher Lamb for pointing this out.
...
llvm-svn: 48973
2008-03-31 01:39:20 +00:00
f2b0b0eb17
Don't eliminate bitcast instructions that change the type of a pointer
...
llvm-svn: 48971
2008-03-31 00:22:16 +00:00
7d6c8ae488
Fix comment typo.
...
llvm-svn: 48967
2008-03-30 19:38:55 +00:00
9fb8908457
Moved from PR1570.
...
llvm-svn: 48965
2008-03-30 19:07:11 +00:00
0f760dfe09
Fix "Control reaches the end of non-void function" warnings,
...
patch by David Chisnall.
llvm-svn: 48963
2008-03-30 18:22:13 +00:00
519a51a76f
minor code cleanups, allow constant folding sinf/cosf.
...
llvm-svn: 48961
2008-03-30 18:02:00 +00:00
16d72072df
Cosmetic changes.
...
llvm-svn: 48947
2008-03-29 18:34:22 +00:00
4311ad2dae
change iterator invalidation avoidance to just move the iterator backward
...
when something changes, instead of moving forward. This allows us to
simplify memset lowering, inserting the memset at the end of the range of
stuff we're touching instead of at the start.
This, in turn, allows us to make use of the addressing instructions already
used in the function instead of inserting our own. For example, we now
codegen:
%tmp41 = getelementptr [8 x i8]* %ref_idx, i32 0, i32 0 ; <i8*> [#uses=2]
call void @llvm.memset.i64( i8* %tmp41, i8 -1, i64 8, i32 1 )
instead of:
%tmp20 = getelementptr [8 x i8]* %ref_idx, i32 0, i32 7 ; <i8*> [#uses=1]
%ptroffset = getelementptr i8* %tmp20, i64 -7 ; <i8*> [#uses=1]
call void @llvm.memset.i64( i8* %ptroffset, i8 -1, i64 8, i32 1 )
llvm-svn: 48940
2008-03-29 05:15:47 +00:00
ac95515741
make the common case of a single store (which clearly shouldn't be turned
...
into a memset!) faster by avoiding an allocation of an std::list node.
llvm-svn: 48939
2008-03-29 04:52:12 +00:00
d528b21a65
give form-memset a significantly more sane heuristic, enable it by default.
...
llvm-svn: 48937
2008-03-29 04:36:18 +00:00
8b22873bdd
Remove some unneeded code for LiveInterval joining, and fix a bug in the Phi elimination algorithm where we were accidentally reasoning about
...
the source rather than the destination.
llvm-svn: 48936
2008-03-29 01:58:47 +00:00
fd2eb00cc2
Fix a tokenfactor node to use the load chain rather than the
...
load value. This fixes PR2177.
llvm-svn: 48932
2008-03-28 23:45:16 +00:00
b8654202dd
Backing out 48911 for now. It's breaking stuff.
...
llvm-svn: 48922
2008-03-28 17:49:06 +00:00
a148acdc82
ifdef out a dead function. Should this be removed?
...
llvm-svn: 48916
2008-03-28 15:36:27 +00:00
35c7cdac07
Rename getAnyLoad to getLoad is suggested by Evan.
...
llvm-svn: 48914
2008-03-28 09:45:24 +00:00
81e0c9a32c
New entry.
...
llvm-svn: 48912
2008-03-28 07:07:06 +00:00
9ae4d7b719
Load from stub is already re-materializable.
...
llvm-svn: 48911
2008-03-28 06:49:25 +00:00
f88c84f690
Update example to new syntax.
...
llvm-svn: 48910
2008-03-28 06:46:51 +00:00
d62964a7d8
make memset inference significantly more powerful: it can now handle
...
memsets that initialize "structs of arrays" and other store sequences
that are not sequential. This is still only enabled if you pass
-form-memset-from-stores. The flag is not heavily tested and I haven't
analyzed the perf regressions when -form-memset-from-stores is passed
either, but this causes no make check regressions.
llvm-svn: 48909
2008-03-28 06:45:13 +00:00
87bac50d7b
New entry.
...
llvm-svn: 48908
2008-03-28 06:34:23 +00:00
f740509e58
Implement LegalizeTypes support for softfloat LOAD.
...
In order to handle indexed nodes I had to introduce
a new constructor, and since I was there I factorized
the code in the various load constructors.
llvm-svn: 48894
2008-03-27 20:23:40 +00:00
cad51cb671
Avoid creating chain dependencies from CopyToReg nodes to load and store
...
nodes. This doesn't currently have much impact the generated code, but it
does produce simpler-looking SelectionDAGs, and consequently
simpler-looking ScheduleDAGs, because there are fewer spurious
dependencies.
In particular, CopyValueToVirtualRegister now uses the entry node as the
input chain dependency for new CopyToReg nodes instead of calling getRoot
and depending on the most recent memory reference.
Also, rename UnorderedChains to PendingExports and pull it up from being
a local variable in SelectionDAGISel::BuildSelectionDAG to being a
member variable of SelectionDAGISel, so that it doesn't have to be
passed around to all the places that need it.
llvm-svn: 48893
2008-03-27 19:56:19 +00:00
eb1e3fcbe0
PHI->removeIncomingValue may remove PHInode.
...
Increment iterator in advance.
llvm-svn: 48890
2008-03-27 17:32:46 +00:00
30d09518b5
Fix spelling. Thanks, Duncan! :-)
...
llvm-svn: 48873
2008-03-27 09:44:37 +00:00
bc674501ba
Speed-up the SumOfUnscheduledPredsOfSuccs by introducing a new function
...
called LimitedSumOfUnscheduledPredsOfSuccs. It terminates the computation
after a given treshold is reached. This new function is always faster, but
brings real wins only on bigger test-cases.
The old function SumOfUnscheduledPredsOfSuccs is left in-place for now and therefore a warning about an unused static function is produced.
llvm-svn: 48872
2008-03-27 09:14:57 +00:00
308e564693
Code clean up.
...
llvm-svn: 48856
2008-03-27 01:45:11 +00:00
29e62a59f3
Allow certain lea instructions to be rematerialized.
...
llvm-svn: 48855
2008-03-27 01:41:09 +00:00
4fb07c6500
Remove an unused command line option.
...
llvm-svn: 48854
2008-03-27 01:30:24 +00:00
5832410d77
Fix a memory bug: increment an iterator of a deleted machine instr.
...
llvm-svn: 48853
2008-03-27 01:27:25 +00:00
8ac07c2834
Expose ExecutionEngine::getTargetData() to c and ocaml bindings.
...
llvm-svn: 48851
2008-03-27 00:27:14 +00:00
87c6ada5de
Fix a bug in Darwin EH: FDE->CIE pointer must
...
be relocatable. Describe why .set is needed better.
llvm-svn: 48848
2008-03-26 23:31:39 +00:00
db390694ff
One more coalescer fix wrt deadness propagation.
...
llvm-svn: 48837
2008-03-26 20:15:49 +00:00
289ba4f335
Avoid commuting a def MI in order to coalesce a copy instruction away if any use of the same val# is a copy instruction that has already been coalesced.
...
llvm-svn: 48833
2008-03-26 19:03:01 +00:00
358e04a185
Use a linked data structure for the uses lists of an SDNode, just like
...
LLVM Value/Use does and MachineRegisterInfo/MachineOperand does.
This allows constant time for all uses list maintenance operations.
The idea was suggested by Chris. Reviewed by Evan and Dan.
Patch is tested and approved by Dan.
On normal use-cases compilation speed is not affected. On very big basic
blocks there are compilation speedups in the range of 15-20% or even better.
llvm-svn: 48822
2008-03-26 12:39:26 +00:00
733a4d6e85
Fixed some spelling errors. Thanks, Duncan!
...
llvm-svn: 48819
2008-03-26 11:23:38 +00:00
7e71b4baaf
Some improvements related to the computation of isReachable.
...
This fixes Bugzilla #1835 (http://llvm.org/bugs/show_bug.cgi?id=1835 ).
This patched is reviewed by Tanya and Dan. Dan tested and approved it.
The reason for the bad performance of the old algorithm is that it is very naive and scans every
time all nodes of the DAG in the worst case.
This patch introduces a new algorithm based on the paper "Online algorithms
for maintaining the topological order of a directed acyclic graph" by
David J.Pearce and Paul H.J.Kelly. This is the MNR algorithm. It has a
linear time worst-case and performs much better in most situations.
The paper can be found here:
http://fano.ics.uci.edu/cites/Document/Online-algorithms-for-maintaining-the-topological-order-of-a-directed-acyclic-graph.html
The main idea of the new algorithm is to compute the topological ordering of the SNodes in the
DAG and to maintain it even after DAG modifications. The topological ordering allows for very fast
node reachability checks.
Tests on very big input files with tens of thousands of instructions in a BB indicate huge
speed-ups (up to 10x compilation time improvement) compared to the old version.
llvm-svn: 48817
2008-03-26 09:18:09 +00:00
292063603e
Fix some SSE4.1 instruction encoding bugs.
...
llvm-svn: 48815
2008-03-26 08:11:49 +00:00
5d2d1776e0
Dead PHI instructions need to be handled specially.
...
llvm-svn: 48811
2008-03-26 03:03:23 +00:00
ad6c23d5e9
Use ## for comment delimiter on darwin x86-32, so
...
llvm's output .s files will go through gcc -std=c99
without triggering preprocesser errors. Approach
suggested by Daveed Vandevoorde.
llvm-svn: 48808
2008-03-25 23:29:30 +00:00
ddc58ff92a
Smaller function alignment when optimizing for size.
...
llvm-svn: 48805
2008-03-25 22:29:46 +00:00
88c44ef91f
Rename option -optimizefor-size to -optimize-size.
...
llvm-svn: 48804
2008-03-25 22:28:39 +00:00
9f129318dc
Remove some debugging code.
...
llvm-svn: 48803
2008-03-25 22:26:43 +00:00
1d46d45e35
StrongPHIElimination doesn't support swapping live intervals like the coalescer does.
...
llvm-svn: 48802
2008-03-25 22:25:27 +00:00
c60c67fc37
Add explicit keywords.
...
llvm-svn: 48801
2008-03-25 22:06:05 +00:00
bdc24adaaf
A quick nm audit turned up several fixed tables and objects that were
...
marked read-write. Use const so that they can be allocated in a
read-only segment.
llvm-svn: 48800
2008-03-25 21:45:14 +00:00
a7ba51f6ec
Avoid outputing spaces at the ends of lines.
...
llvm-svn: 48797
2008-03-25 21:38:12 +00:00
72cfe84f05
Do not align loops if optimizing for size.
...
llvm-svn: 48794
2008-03-25 21:03:02 +00:00
246a52740b
Add optimize-for-size knob.
...
llvm-svn: 48793
2008-03-25 21:02:35 +00:00
df1690dc7c
Handle a special case xor undef, undef -> 0. Technically this should be transformed to undef. But this is such a common idiom (misuse) we are going to handle it.
...
llvm-svn: 48792
2008-03-25 20:08:07 +00:00
2b72c05992
Handle a special case xor undef, undef -> 0. Technically this should be transformed to undef. But this is such a common idiom (misuse) we are going to handle it.
...
llvm-svn: 48791
2008-03-25 20:07:13 +00:00
fd227e9c3a
Fix typos.
...
llvm-svn: 48779
2008-03-25 17:10:29 +00:00
883cbfd0ba
Add CMP32mr and friends to the load-unfolding table. Among
...
other things, this allows the scheduler to unfold a load operand
in the 2008-01-08-SchedulerCrash.ll testcase, so it now successfully
clones the comparison to avoid a pushf+popf.
llvm-svn: 48777
2008-03-25 16:53:19 +00:00
0ee1bd9588
Fix PR2062: Don't build Intrinsics.gen in the source directory. Do it in the
...
object directory.
llvm-svn: 48766
2008-03-25 09:12:48 +00:00
7d564c3b4a
lastRegisterUse() should ignore identity copies. Those will be erased.
...
llvm-svn: 48759
2008-03-25 02:02:19 +00:00
fe7610f37f
Remove an unneeded test.
...
llvm-svn: 48755
2008-03-24 23:55:16 +00:00
50b536eef9
Add \t after .set. Fix by Jay Freeman.
...
llvm-svn: 48753
2008-03-24 23:36:49 +00:00
69a3f9c417
If the coalescer commuted a def MI to allow coalescing, it can changed a previously coalesced copy into an non-identity copy.
...
llvm-svn: 48752
2008-03-24 23:31:21 +00:00
6e225173c5
Add an assertion to catch register of illegal class.
...
llvm-svn: 48751
2008-03-24 23:28:21 +00:00
6306183df3
Use the bit size of the operand instead of the hard-coded 32 to generate the
...
mask.
llvm-svn: 48750
2008-03-24 23:16:37 +00:00
615488ab45
- SSE4.1 extractfps extracts a f32 into a gr32 register. Very useful! Not. Fix the instruction specification and teaches lowering code to use it only when the only use is a store instruction.
...
llvm-svn: 48746
2008-03-24 21:52:23 +00:00
c905fe9407
Revert r48676. I had plans for using it, but now it's just dead code.
...
llvm-svn: 48743
2008-03-24 21:29:58 +00:00
e2707768a4
Remove #include<iostream>, which I was using for debugging.
...
llvm-svn: 48739
2008-03-24 20:36:47 +00:00
a38f58aa5c
Add incoming value from header only if phi node has any use inside the loop.
...
llvm-svn: 48738
2008-03-24 20:16:14 +00:00
aa583d75b1
Shrink the size of AllocationInst by using its SubclassData
...
field to store the alignment value instead of haing a
separate field.
llvm-svn: 48727
2008-03-24 16:55:58 +00:00
d8ea040c31
APIntify SelectionDAG's EXTRACT_ELEMENT code.
...
llvm-svn: 48726
2008-03-24 16:38:05 +00:00
3471ae8c5d
Increasing the inline limit from (overly conservative) 200 to 300. Given each BB costs 20 and each instruction costs 5, 200 means a 4 BB function + 24 instructions (actually less because caller's size also contributes to it).
...
Furthermore, double the limit when more than 10% of the callee instructions are vector instructions. Multimedia kernels tend to love inlining.
llvm-svn: 48725
2008-03-24 06:37:48 +00:00
21a8e3d260
Temporarily disabling memset forming optimization. Add an option.
...
llvm-svn: 48720
2008-03-24 05:28:38 +00:00
200e57840e
Be sure to remove intervals after we've joined them. Also, remove some duplicated code.
...
With this pass, StrongPHIElim can compile very simple testcases correctly. There's still a ways
to go before it's ready for prime time, though.
llvm-svn: 48719
2008-03-24 04:11:27 +00:00
c3cf9f872a
Transform (zext (or (icmp), (icmp))) to (or (zext (cimp), (zext icmp))) if at least one of the (zext icmp) can be transformed to eliminate an icmp.
...
llvm-svn: 48715
2008-03-24 00:21:34 +00:00
58db865d6e
Remove duplicated entries.
...
llvm-svn: 48714
2008-03-23 22:56:07 +00:00
07a45f4edb
Objective Caml bindings for basic block, function, global, and arg iterators.
...
llvm-svn: 48711
2008-03-23 22:21:29 +00:00
1fdd5e9133
Minor typo fixes. Also add another FIXME.
...
llvm-svn: 48710
2008-03-23 20:32:06 +00:00
d38b3fb127
Preserve calling convention during function cloning
...
llvm-svn: 48708
2008-03-23 16:03:00 +00:00
17fb491469
Add license header
...
llvm-svn: 48707
2008-03-23 14:53:18 +00:00
9f0e820fa3
Add Win64 compilation callback. This allows easy examples to be JITed on Win64!
...
llvm-svn: 48706
2008-03-23 14:44:32 +00:00
a347663762
Provide a JIT selector on win64
...
llvm-svn: 48704
2008-03-23 13:43:47 +00:00
7574ead985
Hack out the PIC mode on Win64 targets. This needs to be investigated later.
...
llvm-svn: 48703
2008-03-23 13:41:18 +00:00
4733e72a25
Code cleanup. Provide generic way of selecting JIT pointer bitwidth regardless
...
of compiler used.
llvm-svn: 48702
2008-03-23 13:40:45 +00:00
bd47269f13
Remove old-standing obsolete code.
...
llvm-svn: 48701
2008-03-23 12:32:54 +00:00
cec773d8e7
Honour built-in defines on win64 targets for automatically subtarget recognize.
...
Force stack alignment to 16 bytes on win targets.
llvm-svn: 48695
2008-03-22 21:18:22 +00:00
07a789d2b5
Recognize "windows" in target triple, not only "win32"
...
llvm-svn: 48694
2008-03-22 21:12:53 +00:00
b86e0936f1
Add information about callee-saved registers on Win64
...
llvm-svn: 48692
2008-03-22 21:04:01 +00:00
7f125b2ba5
Add convenient helper for win64 check. Simplify things slightly.
...
llvm-svn: 48691
2008-03-22 20:57:27 +00:00
7b4f4e1a86
Initial support for Win64 calling conventions. Still in early state.
...
llvm-svn: 48690
2008-03-22 20:37:30 +00:00
7ebdd255b5
Register EH frames emitted in JIT when using gcc unwinding runtime
...
llvm-svn: 48688
2008-03-22 08:53:09 +00:00
dd8c695746
Fix merge error
...
llvm-svn: 48687
2008-03-22 08:50:08 +00:00
3f7fab913d
Add testcase for prev. commit. Minor fixes
...
llvm-svn: 48686
2008-03-22 08:37:05 +00:00
25b2e8203a
Aliasing chains cleanups: update langref, add check into verifier
...
llvm-svn: 48685
2008-03-22 08:36:14 +00:00
72d5d42dbc
Support chained aliases for LLVM IR printing. This fixes PR2145
...
llvm-svn: 48684
2008-03-22 08:17:17 +00:00
2fa75184f3
Another comments fixing
...
llvm-svn: 48683
2008-03-22 07:53:40 +00:00
2fb3897f1b
Slightly increase default set size. It's cheap and won't hurt.
...
llvm-svn: 48682
2008-03-22 07:48:40 +00:00
53ccb62712
implement an initial hack at a straight-line store -> memset optimization.
...
This fires dozens of times across spec and multisource, but I don't know
if it actually speeds stuff up. Hopefully the testers will show something
nice :)
llvm-svn: 48680
2008-03-22 05:37:16 +00:00
168be766a8
implement the logic for memset insertion and store deletion.
...
llvm-svn: 48679
2008-03-22 04:13:49 +00:00
5b993fd70d
Add a comment, and fix a bug where AllocateRW recurred to AllocateRWX instead of itself.
...
llvm-svn: 48677
2008-03-22 02:59:54 +00:00
c1e4e3e9a1
Add an AllocateRW to match AllocateRWX.
...
llvm-svn: 48676
2008-03-22 02:33:53 +00:00
31604a62f6
Teach DAG combiner to commute commutable binary nodes in order to achieve sdisel CSE.
...
llvm-svn: 48673
2008-03-22 01:55:50 +00:00
f5d41c67af
This is a partially implemented and currently disabled start of a store
...
merging optimization. Nothing to see here, hopefully more later :)
llvm-svn: 48670
2008-03-22 00:31:52 +00:00
9988569af8
Don't include <map> in Pass.h, which doesn't need it. This requires
...
adding <map> to many files that actually do need it.
llvm-svn: 48667
2008-03-21 23:51:57 +00:00
026eed69e7
Avoid calling array_endof in a static initializer, as it incurs
...
dynamic initialization.
llvm-svn: 48666
2008-03-21 23:44:21 +00:00
5a6ac79738
Specialize FORCE_DEFINING_FILE_TO_BE_LINKED using a GCC trick
...
to avoid using constructor calls for static objects. This reduces
the number of objects requiring static constructors in a typical
LLVM build by around 20%.
llvm-svn: 48665
2008-03-21 23:38:23 +00:00
30e44a4b40
Fix -view-sunit-dags to support cross-rc-copy nodes.
...
llvm-svn: 48664
2008-03-21 22:51:06 +00:00
804209d17c
the size of a smallvector shouldn't be part of the interface to these methods.
...
llvm-svn: 48662
2008-03-21 22:01:16 +00:00
beb216da0a
make gvn marginally faster by reallocating the lastSeenLoad map for
...
each basic block.
llvm-svn: 48660
2008-03-21 21:33:23 +00:00
2876a645c3
Minor cleanups and shrinkification.
...
llvm-svn: 48658
2008-03-21 21:14:38 +00:00
a25dde6fee
Handle getresult instructions in different basic blocks
...
from their aggregate operands by moving the getresult
instructions.
llvm-svn: 48657
2008-03-21 21:01:32 +00:00
c55b444a8f
Restore this assert now that the livevar bug is fixed.
...
This verifies kill info for "ret" fp operands is right.
llvm-svn: 48656
2008-03-21 20:41:27 +00:00
8c19af1b7e
A couple of kill marker maintainence bug.
...
llvm-svn: 48653
2008-03-21 19:09:30 +00:00
74d154ce57
FunctionExtractorPass has been superceded by GVExtractorPass
...
llvm-svn: 48648
2008-03-21 16:46:53 +00:00
dec270a6db
Make it possible to get an empty struct using
...
the new StructType::get method. The second NULL
is to pacify the gcc warning mechanism. This
patch compiles but is otherwise untested.
llvm-svn: 48645
2008-03-21 15:53:17 +00:00
d97eea372a
Introduce a new node for holding call argument
...
flags. This is needed by the new legalize types
infrastructure which wants to expand the 64 bit
constants previously used to hold the flags on
32 bit machines. There are two functional changes:
(1) in LowerArguments, if a parameter has the zext
attribute set then that is marked in the flags;
before it was being ignored; (2) PPC had some bogus
code for handling two word arguments when using the
ELF 32 ABI, which was hard to convert because of
the bogusness. As suggested by the original author
(Nicolas Geoffray), I've disabled it for the moment.
Tested with "make check" and the Ada ACATS testsuite.
llvm-svn: 48640
2008-03-21 09:14:45 +00:00
c9e09a0588
Fix the build for gcc-4.2.
...
llvm-svn: 48639
2008-03-21 08:32:17 +00:00
68b11e14bc
remove Evan's "ugly hack" that sorta attempted to get
...
x86-64 return conventions correct, but was never enabled.
We can now do the "right thing" with multiple return values.
llvm-svn: 48635
2008-03-21 06:50:21 +00:00
5abbe6cef5
Add support for calls that return two FP values in
...
ST(0)/ST(1).
llvm-svn: 48634
2008-03-21 06:38:26 +00:00
Duncan Sands