1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-24 11:42:57 +01:00
Commit Graph

124 Commits

Author SHA1 Message Date
Evan Cheng
b4dc8bdd22 Sink SubtargetFeature and TargetInstrItineraries (renamed MCInstrItineraries) into MC.
llvm-svn: 134049
2011-06-29 01:14:12 +00:00
Evan Cheng
4a169be530 - Rename TargetInstrDesc, TargetOperandInfo to MCInstrDesc and MCOperandInfo and
sink them into MC layer.
- Added MCInstrInfo, which captures the tablegen generated static data. Chang
TargetInstrInfo so it's based off MCInstrInfo.

llvm-svn: 134021
2011-06-28 19:10:37 +00:00
Evan Cheng
2c06c8b3c2 More refactoring. Move getRegClass from TargetOperandInfo to TargetInstrInfo.
llvm-svn: 133944
2011-06-27 21:26:13 +00:00
Evan Cheng
ea0d287a8a Look pass copies when determining whether hoisting would end up inserting more copies. rdar://9266679
llvm-svn: 129297
2011-04-11 21:09:18 +00:00
Cameron Zwarich
a1920d7f51 Move getRegPressureLimit() from TargetLoweringInfo to TargetRegisterInfo.
llvm-svn: 127175
2011-03-07 21:56:36 +00:00
Evan Cheng
6dc21c7358 Sorry, several patches in one.
TargetInstrInfo:
Change produceSameValue() to take MachineRegisterInfo as an optional argument.
When in SSA form, targets can use it to make more aggressive equality analysis.

Machine LICM:
1. Eliminate isLoadFromConstantMemory, use MI.isInvariantLoad instead.
2. Fix a bug which prevent CSE of instructions which are not re-materializable.
3. Use improved form of produceSameValue.

ARM:
1. Teach ARM produceSameValue to look pass some PIC labels.
2. Look for operands from different loads of different constant pool entries
   which have same values.
3. Re-implement PIC GA materialization using movw + movt. Combine the pair with
   a "add pc" or "ldr [pc]" to form pseudo instructions. This makes it possible
   to re-materialize the instruction, allow machine LICM to hoist the set of
   instructions out of the loop and make it possible to CSE them. It's a bit
   hacky, but it significantly improve code quality.
4. Some minor bug fixes as well.

With the fixes, using movw + movt to materialize GAs significantly outperform the
load from constantpool method. 186.crafty and 255.vortex improved > 20%, 254.gap
and 176.gcc ~10%.

llvm-svn: 123905
2011-01-20 08:34:58 +00:00
Jakob Stoklund Olesen
32f1783ca1 Simplify a bunch of isVirtualRegister() and isPhysicalRegister() logic.
These functions not longer assert when passed 0, but simply return false instead.

No functional change intended.

llvm-svn: 123155
2011-01-10 02:58:51 +00:00
Dan Gohman
d88f1d63ac Add a FIXME comment.
llvm-svn: 118803
2010-11-11 18:08:43 +00:00
Jakob Stoklund Olesen
56f67c7d02 MachineLICM should not claim to be preserving the CFG when it can split critical
edges on demand.

llvm-svn: 117982
2010-11-01 23:59:55 +00:00
Evan Cheng
324e678bb7 Use instruction itinerary to determine what instructions are 'cheap'.
llvm-svn: 117348
2010-10-26 02:08:50 +00:00
Evan Cheng
9a06e4c7c8 More accurate estimate / tracking of register pressure.
- Initial register pressure in the loop should be all the live defs into the
  loop. Not just those from loop preheader which is often empty.
- When an instruction is hoisted, update register pressure from loop preheader
  to the original BB.
- Treat only use of a virtual register as kill since the code is still SSA.

llvm-svn: 116956
2010-10-20 22:03:58 +00:00
Dan Gohman
c781a28a1d Make CodeGen TBAA-aware.
llvm-svn: 116890
2010-10-20 00:31:05 +00:00
Evan Cheng
1c8dafd12a Re-enable register pressure aware machine licm with fixes. Hoist() may have
erased the instruction during LICM so UpdateRegPressureAfter() should not
reference it afterwards.

llvm-svn: 116845
2010-10-19 18:58:51 +00:00
Owen Anderson
46990c17f7 Get rid of static constructors for pass registration. Instead, every pass exposes an initializeMyPassFunction(), which
must be called in the pass's constructor.  This function uses static dependency declarations to recursively initialize
the pass's dependencies.

Clients that only create passes through the createFooPass() APIs will require no changes.  Clients that want to use the
CommandLine options for passes will need to manually call the appropriate initialization functions in PassInitialization.h
before parsing commandline arguments.

I have tested this with all standard configurations of clang and llvm-gcc on Darwin.  It is possible that there are problems
with the static dependencies that will only be visible with non-standard options.  If you encounter any crash in pass
registration/creation, please send the testcase to me directly.

llvm-svn: 116820
2010-10-19 17:21:58 +00:00
Daniel Dunbar
6ff550c84d Revert r116781 "- Add a hook for target to determine whether an instruction def
is", which breaks some nightly tests.

llvm-svn: 116816
2010-10-19 17:14:24 +00:00
Andrew Trick
3df2830391 Fix for machine licm assert: RCCost <= RegPressure[RCId]
in MultiSource/Benchmarks/VersaBench/beamformer/beamformer.
SmallSet.insert returns true if the element is inserted.

llvm-svn: 116790
2010-10-19 02:50:50 +00:00
Evan Cheng
9c3f6f486e - Add a hook for target to determine whether an instruction def is
"long latency" enough to hoist even if it may increase spilling. Reloading
  a value from spill slot is often cheaper than performing an expensive
  computation in the loop. For X86, that means machine LICM will hoist
  SQRT, DIV, etc. ARM will be somewhat aggressive with VFP and NEON
  instructions.
- Enable register pressure aware machine LICM by default.

llvm-svn: 116781
2010-10-19 00:55:07 +00:00
Evan Cheng
5bd7971922 More machine LICM work. It now tracks register pressure for path from preheader to current BB and use the information determine whether hoisting is worthwhile.
llvm-svn: 116654
2010-10-16 02:20:26 +00:00
Evan Cheng
e1e152c3fc Register pressure and instruction latency aware machine LICM. Work in progress.
llvm-svn: 116465
2010-10-14 01:16:09 +00:00
Owen Anderson
63f757463c Begin adding static dependence information to passes, which will allow us to
perform initialization without static constructors AND without explicit initialization
by the client.  For the moment, passes are required to initialize both their
(potential) dependencies and any passes they preserve.  I hope to be able to relax
the latter requirement in the future.

llvm-svn: 116334
2010-10-12 19:48:12 +00:00
Evan Cheng
8833658eb0 Don't waste time unfolding simple loads. The unfolded copy won't be hoisted.
llvm-svn: 116081
2010-10-08 18:59:19 +00:00
Owen Anderson
69cbf2e8b7 Now with fewer extraneous semicolons!
llvm-svn: 115996
2010-10-07 22:25:06 +00:00
Owen Anderson
f2fea95f2f Reapply r110396, with fixes to appease the Linux buildbot gods.
llvm-svn: 110460
2010-08-06 18:33:48 +00:00
Owen Anderson
aadd8a89ca Revert r110396 to fix buildbots.
llvm-svn: 110410
2010-08-06 00:23:35 +00:00
Owen Anderson
b9762c07cb Don't use PassInfo* as a type identifier for passes. Instead, use the address of the static
ID member as the sole unique type identifier.  Clean up APIs related to this change.

llvm-svn: 110396
2010-08-05 23:42:04 +00:00
Dale Johannesen
71204e008b Comment typo.
llvm-svn: 109765
2010-07-29 17:45:24 +00:00
Owen Anderson
f8addbb0a1 Fix batch of converting RegisterPass<> to INTIALIZE_PASS().
llvm-svn: 109045
2010-07-21 22:09:45 +00:00
Dale Johannesen
6589cf8e0d Fix test for switch statements and increase
threshold a bit per experimentation.

llvm-svn: 108935
2010-07-20 21:29:12 +00:00
Dale Johannesen
6b9a191b08 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.

llvm-svn: 108792
2010-07-20 00:50:13 +00:00
Evan Cheng
72e40c4e08 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
Chris Lattner
55bd17a388 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
Dan Gohman
0c7d7ee288 Fix MachineLICM to actually visit inner loops.
llvm-svn: 108001
2010-07-09 18:49:45 +00:00
Dan Gohman
2303159bba 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
Evan Cheng
058f778d52 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
Dan Gohman
33e9c347bf 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
Jakob Stoklund Olesen
1ba4c9d2c2 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
Evan Cheng
0004496a98 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
Evan Cheng
aadcfa8ad7 Fast path implicit_def check.
llvm-svn: 101183
2010-04-13 22:13:34 +00:00
Evan Cheng
6e5407402e Avoid variable shadowing.
llvm-svn: 101170
2010-04-13 20:25:29 +00:00
Evan Cheng
bc94b6f1fd 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
Evan Cheng
d83c1b5a0e 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
Dan Gohman
2bca857d78 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
Benjamin Kramer
9904b07b61 Plug trivial leak.
llvm-svn: 101034
2010-04-12 11:38:35 +00:00
Dan Gohman
f9cf2772f6 Delete this obsolete comment.
llvm-svn: 100858
2010-04-09 14:12:01 +00:00
Evan Cheng
2f7cee1730 Make post regalloc machine licm functional. It now passes all of MultiSource.
llvm-svn: 100742
2010-04-08 01:03:47 +00:00
Evan Cheng
0f71b51991 Add comments for missed opportunities.
llvm-svn: 100610
2010-04-07 06:00:33 +00:00
Evan Cheng
eb32efa3b6 Fix typo.
llvm-svn: 100609
2010-04-07 05:59:12 +00:00
Evan Cheng
ac90d47ea3 Post regalloc LICM. Work in progress.
llvm-svn: 100592
2010-04-07 00:41:17 +00:00
Evan Cheng
70b7ecdef9 - 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
Dan Gohman
0799a41c48 Don't try to replace physical registers when doing CSE.
llvm-svn: 97360
2010-02-28 01:33:43 +00:00