08645f1991
Don't hoist things out of a large switch inside a
...
loop, for the reasons in the comments. This is a
major win on 253.perlbmk on ARM Darwin. I expect it
to be a good heuristic in general, but it's possible
some things will regress; I'll be watching.
7940152
2010-07-20 00:50:13 +00:00
d542414945
Teach ProcessImplicitDefs to transform more COPY instructions into IMPLICIT_DEF (and subsequently eliminate them). This allows machine LICM to hoist IMPLICIT_DEF's. PR7620.
...
llvm-svn: 108304
2010-07-14 01:22:19 +00:00
0b7ae20a35
change machinelicm to use MachineInstr::isSafeToMove. No
...
intended functionality change.
The avoidance of hoistiing implicitdef seems wrong though.
llvm-svn: 108109
2010-07-12 00:00:35 +00:00
7929c448fc
Fix MachineLICM to actually visit inner loops.
...
llvm-svn: 108001
2010-07-09 18:49:45 +00:00
3570f81b1e
Move PHIElimination's SplitCriticalEdge for MachineBasicBlocks out
...
into a utility routine, teach it how to update MachineLoopInfo, and
make use of it in MachineLICM to split critical edges on demand.
llvm-svn: 106555
2010-06-22 17:25:57 +00:00
032f3261a2
Doh. Machine LICM is re-initializing the CSE map over and over. Patch by Anna Zaks. rdar://8037934.
...
llvm-svn: 105065
2010-05-29 00:06:36 +00:00
c90f51c00b
Teach MachineLICM and MachineSink how to clear kill flags conservatively
...
when they move instructions.
llvm-svn: 103737
2010-05-13 20:34:42 +00:00
011207a0ae
When MachineLICM is hoisting a physical register after regalloc, make sure the
...
register is not killed in the loop.
This fixes 188.ammp on ARM where the post-ra scheduler would grab a register
that looked available but wasn't.
A testcase would be huge and fragile, sorry.
llvm-svn: 101930
2010-04-20 18:45:47 +00:00
5fdb57cc10
Postra machine licm must add registers defined by loop invariants to *all* of
...
the live-in sets of BBs in the loop. Otherwise later pass may end up using the
registers and override the invariant. rdar://7852937
No reasonablly sized test case possible.
llvm-svn: 101626
2010-04-17 07:07:11 +00:00
87585d72a5
Fast path implicit_def check.
...
llvm-svn: 101183
2010-04-13 22:13:34 +00:00
cce672c172
Avoid variable shadowing.
...
llvm-svn: 101170
2010-04-13 20:25:29 +00:00
89e74792b6
Expand postra machine licm's capability a little more. If an instruction's register operands are all loop invariants, then it's safe to hoist it.
...
llvm-svn: 101167
2010-04-13 20:21:05 +00:00
0a2aff2d12
Teach postra machine licm to hoist more obvious invariants, e.g. instructions with no source operands.
...
llvm-svn: 101154
2010-04-13 18:16:00 +00:00
9d2d053e11
Eliminate MachineBasicBlock::const_livein_iterator and make
...
MachineBasicBlock::livein_iterator a const_iterator, because
clients shouldn't ever be using the iterator interface to
mutate the livein set.
llvm-svn: 101147
2010-04-13 16:57:55 +00:00
d0b5c6cbed
Plug trivial leak.
...
llvm-svn: 101034
2010-04-12 11:38:35 +00:00
87f8207c97
Delete this obsolete comment.
...
llvm-svn: 100858
2010-04-09 14:12:01 +00:00
058b9f04e8
Make post regalloc machine licm functional. It now passes all of MultiSource.
...
llvm-svn: 100742
2010-04-08 01:03:47 +00:00
5ed679282b
Add comments for missed opportunities.
...
llvm-svn: 100610
2010-04-07 06:00:33 +00:00
fcbcc0bd51
Fix typo.
...
llvm-svn: 100609
2010-04-07 05:59:12 +00:00
6ea5949a93
Post regalloc LICM. Work in progress.
...
llvm-svn: 100592
2010-04-07 00:41:17 +00:00
e9c46c25a1
- Change MachineInstr::isIdenticalTo to take a new option that determines whether it should skip checking defs or at least virtual register defs. This subsumes part of the TargetInstrInfo::isIdentical functionality.
...
- Eliminate TargetInstrInfo::isIdentical and replace it with produceSameValue. In the default case, produceSameValue just checks whether two machine instructions are identical (except for virtual register defs). But targets may override it to check for unusual cases (e.g. ARM pic loads from constant pools).
llvm-svn: 97628
2010-03-03 01:44:33 +00:00
34021b7445
Don't try to replace physical registers when doing CSE.
...
llvm-svn: 97360
2010-02-28 01:33:43 +00:00
6fb6a59766
Don't unconditionally suppress hoisting of instructions with implicit
...
defs or uses. The regular def and use checking below covers them, and
can be more precise. It's safe to hoist an instruction with a dead
implicit def if the register isn't live into the loop header.
llvm-svn: 97352
2010-02-28 00:08:44 +00:00
b06015aa69
move target-independent opcodes out of TargetInstrInfo
...
into TargetOpcodes.h. #include the new TargetOpcodes.h
into MachineInstr. Add new inline accessors (like isPHI())
to MachineInstr, and start using them throughout the
codebase.
llvm-svn: 95687
2010-02-09 19:54:29 +00:00
55cf95c202
Change errs() to dbgs().
...
llvm-svn: 92547
2010-01-05 00:03:48 +00:00
18fa5686f6
Add Loop contains utility methods for testing whether a loop
...
contains another loop, or an instruction. The loop form is
substantially more efficient on large loops than the typical
code it replaces.
llvm-svn: 91654
2009-12-18 01:24:09 +00:00
73f9a9e2c8
Enable hoisting load from constant memories.
...
llvm-svn: 89510
2009-11-20 23:31:34 +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
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
0229536301
Fix comment.
...
llvm-svn: 89129
2009-11-17 19:19:01 +00:00
7ff831962a
- Add TargetInstrInfo::isIdentical(). It's similar to MachineInstr::isIdentical
...
except it doesn't care if the definitions' virtual registers differ. This is
used by machine LICM and other MI passes to perform CSE.
- Teach Thumb2InstrInfo::isIdentical() to check two t2LDRpci_pic are identical.
Since pc relative constantpool entries are always different, this requires it
it check if the values can actually the same.
llvm-svn: 86328
2009-11-07 03:52:02 +00:00
921152fe9d
Code refactoring.
...
llvm-svn: 86085
2009-11-05 00:51:13 +00:00
f42b5af549
Re-apply 85799. It turns out my code isn't buggy.
...
llvm-svn: 85947
2009-11-03 21:40:02 +00:00
a8a58efc03
Revert 85799 for now. It might be breaking llvm-gcc driver.
...
llvm-svn: 85827
2009-11-02 21:49:14 +00:00
2729543984
Initilize the machine LICM CSE map upon the first time an instruction is hoisted to
...
the loop preheader. Add instructions which are already in the preheader block that
may be common expressions of those that are hoisted out. These does get a few more
instructions CSE'ed.
llvm-svn: 85799
2009-11-02 08:09:49 +00:00
49fa51d936
Fix MachineLICM to use the correct virtual register class when
...
unfolding loads for hoisting. getOpcodeAfterMemoryUnfold returns the
opcode of the original operation without the load, not the load
itself, MachineLICM needs to know the operand index in order to get
the correct register class. Extend getOpcodeAfterMemoryUnfold to
return this information.
llvm-svn: 85622
2009-10-30 22:18:41 +00:00
104f57cae8
Refactor the code for unfolding a load into a separate function.
...
llvm-svn: 85515
2009-10-29 17:47:20 +00:00
57780dfdfc
Simplify this code: if the unfolded load can't be hoisted, just delete
...
the new instructions and leave the old one in place.
llvm-svn: 85393
2009-10-28 15:23:36 +00:00
1b44f10ff4
Teach MachineLICM to unfold loads from constant memory from
...
otherwise unhoistable instructions in order to allow the loads
to be hoisted.
llvm-svn: 85364
2009-10-28 03:21:57 +00:00
974e12b2d3
Remove includes of Support/Compiler.h that are no longer needed after the
...
VISIBILITY_HIDDEN removal.
llvm-svn: 85043
2009-10-25 06:57:41 +00:00
02d5f77d26
Remove VISIBILITY_HIDDEN from class/struct found inside anonymous namespaces.
...
Chris claims we should never have visibility_hidden inside any .cpp file but
that's still not true even after this commit.
llvm-svn: 85042
2009-10-25 06:33:48 +00:00
87b02d5bbc
Factor out LiveIntervalAnalysis' code to determine whether an instruction
...
is trivially rematerializable and integrate it into
TargetInstrInfo::isTriviallyReMaterializable. This way, all places that
need to know whether an instruction is rematerializable will get the
same answer.
This enables the useful parts of the aggressive-remat option by
default -- using AliasAnalysis to determine whether a memory location
is invariant, and removes the questionable parts -- rematting operations
with virtual register inputs that may not be live everywhere.
llvm-svn: 83687
2009-10-09 23:27:56 +00:00
e19c1810d7
isTriviallyReMaterializable checks the
...
TargetInstrDesc::isRematerializable flag, so it isn't necessary to do
this check in its callers.
llvm-svn: 83671
2009-10-09 21:02:10 +00:00
7ddb844a2d
Give Dan and my recent changes, machine LICM is now code size neutral.
...
llvm-svn: 83624
2009-10-09 06:31:25 +00:00
ccd4545ecb
Fix a logic error that caused non-rematable loop invariants loads to be licm'ed out of loop.
...
llvm-svn: 83622
2009-10-09 06:21:52 +00:00
be8137b0b4
Replace TargetInstrInfo::isInvariantLoad and its target-specific
...
implementations with a new MachineInstr::isInvariantLoad, which uses
MachineMemOperands and is target-independent. This brings MachineLICM
and other functionality to targets which previously lacked an
isInvariantLoad implementation.
llvm-svn: 83475
2009-10-07 17:38:06 +00:00
2f5bdcb7c2
Don't hoist or sink instructions with physreg uses if the physreg is
...
allocatable. Even if it doesn't appear to have any defs, it may latter
on after register allocation.
llvm-svn: 82834
2009-09-26 02:34:00 +00:00
e30d63f1d8
Unbreak MachineLICM for instructions that reference RIP on x86-64 too.
...
llvm-svn: 82825
2009-09-25 23:58:45 +00:00
8781b35be5
Convert DOUT to DEBUG(errs()...).
...
llvm-svn: 79754
2009-08-22 20:25:44 +00:00
7c50c9bd63
Tidy #includes.
...
llvm-svn: 78677
2009-08-11 16:02:12 +00:00
0dd5e1ed39
More migration to raw_ostream, the water has dried up around the iostream hole.
...
- Some clients which used DOUT have moved to DEBUG. We are deprecating the
"magic" DOUT behavior which avoided calling printing functions when the
statement was disabled. In addition to being unnecessary magic, it had the
downside of leaving code in -Asserts builds, and of hiding potentially
unnecessary computations.
llvm-svn: 77019
2009-07-25 00:23:56 +00:00
8d0b4d4fd6
MachineLICM CSE should match destination register classes; avoid hoisting implicit_def's.
...
llvm-svn: 65592
2009-02-27 00:02:22 +00:00
9d8bfbfdef
Machine LICM increases register pressure and it almost always increase code size. For now, disable it for optimizing for size.
...
llvm-svn: 63856
2009-02-05 08:51:13 +00:00
399660c384
Teach machine licm to CSE hoisted instructions.
...
llvm-svn: 63854
2009-02-05 08:45:46 +00:00
4ded601445
Don't call isInvariantLoad twice.
...
llvm-svn: 63729
2009-02-04 09:21:58 +00:00
1d9f7ac8cc
Don't bother hoisting out a "cheap" instruction if all of its uses are PHIs. LICM "cheap" instructions are not particularly beneficial to start with. This will just end up making the copies harder to coalesce.
...
llvm-svn: 63728
2009-02-04 09:19:56 +00:00
0649abdfe2
For now, only hoist re-materilizable instructions. LICM will increase register pressure. We want to avoid spilling more instructions if it's possible.
...
llvm-svn: 63725
2009-02-04 07:17:49 +00:00
79618d1de8
Simplify the MachineLICM pass by having it only traverse outer
...
loops, hoisting instructions all the way out in one step rather
than hoisting them one nest level at a time. Also, make a few
other code simplifications. This speeds up MachineLICM
by several fold.
llvm-svn: 62283
2009-01-15 22:01:38 +00:00
906152a20f
Tidy up #includes, deleting a bunch of unnecessary #includes.
...
llvm-svn: 61715
2009-01-05 17:59:02 +00:00
072e52f170
Use isTerminator() instead of isBranch()||isReturn() in
...
several places. isTerminator() returns true for a superset
of cases, and includes things like FP_REG_KILL, which are
nither return or branch but aren't safe to move/remat/etc.
llvm-svn: 61373
2008-12-23 17:28:50 +00:00
83682a9441
Give MachineLICM a name, for -time-passes etc.
...
llvm-svn: 61184
2008-12-18 01:37:56 +00:00
0d1e9a8e04
Switch the MachineOperand accessors back to the short names like
...
isReg, etc., from isRegister, etc.
llvm-svn: 57006
2008-10-03 15:45:36 +00:00
a79db30d28
Tidy up several unbeseeming casts from pointer to intptr_t.
...
llvm-svn: 55779
2008-09-04 17:05:41 +00:00
58bb4f1bf0
Cosmetic changes to Machine LICM. No functionality change.
...
llvm-svn: 55578
2008-08-31 02:30:23 +00:00
cd01e89852
Don't hoist instructions that define a physical register.
...
llvm-svn: 55074
2008-08-20 20:32:05 +00:00
cf596eb12f
Don't hoist instructions that have implicit uses or defines. Before, it was
...
hoisting out some "test" and "cmp" instructions. This was obvious badness.
llvm-svn: 54908
2008-08-18 00:33:49 +00:00
ecf34435f4
Prevent option name conflict.
...
llvm-svn: 53166
2008-07-07 05:42:27 +00:00
d78c400b5b
Clean up the use of static and anonymous namespaces. This turned up
...
several things that were neither in an anonymous namespace nor static
but not intended to be global.
llvm-svn: 51017
2008-05-13 00:00:25 +00:00
2930845065
Revert the previous commit. Go ahead and hoist rematerializable instructions.
...
llvm-svn: 50990
2008-05-12 19:47:18 +00:00
70613b84da
One real change - don't hoist something that's trivially rematerializable. It's
...
possible for it to produce worse code than before.
The rest of this patch is code cleanup.
llvm-svn: 50987
2008-05-12 19:38:32 +00:00
2823eaebe8
Minor cleanup. No functionality change.
...
llvm-svn: 48142
2008-03-10 08:13:01 +00:00
d7a258d325
Rename PrintableName to Name.
...
llvm-svn: 47629
2008-02-26 21:47:57 +00:00
c24ea4fb41
Change "Name" to "AsmName" in the target register info. Gee, a refactoring tool
...
would have been a Godsend here!
llvm-svn: 47625
2008-02-26 21:11:01 +00:00
3a4be0fdef
Rename MRegisterInfo to TargetRegisterInfo.
...
llvm-svn: 46930
2008-02-10 18:45:23 +00:00
c8226f32e9
Simplify the side effect stuff a bit more and make licm/sinking
...
both work right according to the new flags.
This removes the TII::isReallySideEffectFree predicate, and adds
TII::isInvariantLoad.
It removes NeverHasSideEffects+MayHaveSideEffects and adds
UnmodeledSideEffects as machine instr flags. Now the clients
can decide everything they need.
I think isRematerializable can be implemented in terms of the
flags we have now, though I will let others tackle that.
llvm-svn: 45843
2008-01-10 23:08:24 +00:00
03ad885039
rename TargetInstrDescriptor -> TargetInstrDesc.
...
Make MachineInstr::getDesc return a reference instead
of a pointer, since it can never be null.
llvm-svn: 45695
2008-01-07 07:27:27 +00:00
a98c679de0
Rename MachineInstr::getInstrDescriptor -> getDesc(), which reflects
...
that it is cheap and efficient to get.
Move a variety of predicates from TargetInstrInfo into
TargetInstrDescriptor, which makes it much easier to query a predicate
when you don't have TII around. Now you can use MI->getDesc()->isBranch()
instead of going through TII, and this is much more efficient anyway. Not
all of the predicates have been moved over yet.
Update old code that used MI->getInstrDescriptor()->Flags to use the
new predicates in many places.
llvm-svn: 45674
2008-01-07 01:56:04 +00:00
3bf5603ce4
Add that this preserves some analyses.
...
llvm-svn: 45573
2008-01-04 08:48:49 +00:00
66470d02c3
Move option to enable machine LICM into LLVMTargetMachine.cpp.
...
llvm-svn: 45572
2008-01-04 08:11:03 +00:00
d865697016
Call the parent's getAnalysisUsage.
...
llvm-svn: 45571
2008-01-04 07:50:05 +00:00
b5c1d9b7da
remove dead #includes and reorder the rest.
...
llvm-svn: 45569
2008-01-04 06:41:45 +00:00
0ba4184404
Use the correct MachineRegisterInfo object.
...
llvm-svn: 45499
2008-01-02 21:10:54 +00:00
f0b37780ca
Remove dead code.
...
llvm-svn: 45496
2008-01-02 20:47:37 +00:00
5da1945cdd
Use the new architecture to get the containing machine basic block for a machine
...
instruction. Also, use "splice" to move the new instruction instead of
remove/insert (where it was leaking memory anyway).
llvm-svn: 45492
2008-01-02 19:32:43 +00:00
f3ebc3f3d2
Remove attribution from file headers, per discussion on llvmdev.
...
llvm-svn: 45418
2007-12-29 20:36:04 +00:00
65c001e6bc
Updated comments to reflect what "side effects" means in this situation.
...
llvm-svn: 45245
2007-12-20 01:08:10 +00:00
166f746246
Add debugging info. Use the newly created "hasUnmodelledSideEffects" method.
...
llvm-svn: 45178
2007-12-18 21:38:04 +00:00
38236ef6cb
Need to grow the indexed map. Added debug statements.
...
llvm-svn: 44892
2007-12-11 23:27:51 +00:00
642e15a7cb
Simplify slightly.
...
llvm-svn: 44881
2007-12-11 22:22:22 +00:00
b678ae7c38
Blark! How in the world did this work without this?!
...
llvm-svn: 44874
2007-12-11 19:40:06 +00:00
7717a8a37d
- Update the virtual reg to machine instruction map when hoisting.
...
- Fix subtle bug when creating initially creating this map.
llvm-svn: 44873
2007-12-11 19:17:04 +00:00
5143d898c8
Checking for "zero operands" during the "CanHoistInst()" method isn't necessary
...
because those with side effects will be caught by other checks in here.
Also, simplify the check for a BB in a sub loop.
llvm-svn: 44871
2007-12-11 18:45:11 +00:00
3f19dfe794
Reverting 44702. It wasn't correct to rename them.
...
llvm-svn: 44727
2007-12-08 23:58:46 +00:00
2b07d8c5a0
Renaming:
...
isTriviallyReMaterializable -> hasNoSideEffects
isReallyTriviallyReMaterializable -> isTriviallyReMaterializable
llvm-svn: 44702
2007-12-08 07:17:56 +00:00
4375173ba0
Incorporated comments from Evan and Chris:
...
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20071203/056043.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20071203/056048.html
llvm-svn: 44696
2007-12-08 01:47:01 +00:00
fb706bc52b
Initial commit of the machine code LICM pass. It successfully hoists this:
...
_foo:
li r2, 0
LBB1_1: ; bb
li r5, 0
stw r5, 0(r3)
addi r2, r2, 1
addi r3, r3, 4
cmplw cr0, r2, r4
bne cr0, LBB1_1 ; bb
LBB1_2: ; return
blr
to:
_foo:
li r2, 0
li r5, 0
LBB1_1: ; bb
stw r5, 0(r3)
addi r2, r2, 1
addi r3, r3, 4
cmplw cr0, r2, r4
bne cr0, LBB1_1 ; bb
LBB1_2: ; return
blr
ZOMG!! :-)
Moar to come...
llvm-svn: 44687
2007-12-07 21:42:31 +00:00
Dale Johannesen