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
cd139c0373
Rewrite SelectionDAG::isPredecessorOf to be iterative instead of
...
recursive to avoid consuming extraordinary amounts of stack space
when processing tall graphs.
llvm-svn: 85369
2009-10-28 03:44:30 +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
83896a59e1
Add a second ValueType argument to isFPImmLegal.
...
llvm-svn: 85361
2009-10-28 01:43:28 +00:00
4b46cbfc23
Mark dead physregdefs dead immediately. This helps MachineSink and
...
MachineLICM and other things which run before LiveVariables is run.
llvm-svn: 85360
2009-10-28 01:13:53 +00:00
75d6a4afd2
Allow constants of different types to share constant pool entries
...
if they have compatible encodings.
llvm-svn: 85359
2009-10-28 01:12:16 +00:00
d04cb6d0fa
rename indbr -> indirectbr to appease the residents of #llvm.
...
llvm-svn: 85351
2009-10-28 00:19:10 +00:00
9693f9d465
Record CodeGen optimization level in the BranchFolding pass so that we can
...
use it to control tail merging when there is a tradeoff between performance
and code size. When there is only 1 instruction in the common tail, we have
been merging. That can be good for code size but is a definite loss for
performance. Now we will avoid tail merging in that case when the
optimization level is "Aggressive", i.e., "-O3". Radar 7338114.
Since the IfConversion pass invokes BranchFolding, it too needs to know
the optimization level. Note that I removed the RegisterPass instantiation
for IfConversion because it required a default constructor. If someone
wants to keep that for some reason, we can add a default constructor with
a hard-wired optimization level.
llvm-svn: 85346
2009-10-27 23:49:38 +00:00
cd3939ef0e
Fixed a bug in the coalescer where intervals were occasionally merged despite a real interference. This fixes rdar://problem/7157961.
...
llvm-svn: 85338
2009-10-27 23:16:58 +00:00
a5e078b677
Update the MachineBasicBlock CFG for an indirect branch.
...
llvm-svn: 85325
2009-10-27 22:10:34 +00:00
a4374e66f0
Add CodeGen support for indirect branches.
...
llvm-svn: 85323
2009-10-27 21:56:26 +00:00
cfeaa48642
Do not held on to DenseMap slot accross map insertion. The insertion may cause the map to grow rending the slot invalid.
...
Use this opportunity to use ValueMap instead of DenseMap.
llvm-svn: 85298
2009-10-27 20:47:17 +00:00
26076a8f10
don't use stdio
...
llvm-svn: 85296
2009-10-27 20:42:54 +00:00
16993aa30b
Do away with addLegalFPImmediate. Add a target hook isFPImmLegal which returns true if the fp immediate can be natively codegened by target.
...
llvm-svn: 85281
2009-10-27 19:56:55 +00:00
3ed871fe62
add enough support for indirect branch for the feature test to pass
...
(assembler,asmprinter, bc reader+writer) and document it. Codegen
currently aborts on it.
llvm-svn: 85274
2009-10-27 19:13:16 +00:00
0997991252
pseudosourcevalue is also still using getGlobalContext(), so it isn't
...
thread safe either.
llvm-svn: 85253
2009-10-27 17:02:08 +00:00
7a50b280c1
Add objectsize intrinsic and hook it up through codegen. Doesn't
...
do anything than return "I don't know" at the moment.
llvm-svn: 85189
2009-10-27 00:52:25 +00:00
de5ad42aa1
Remove FreeInst.
...
Remove LowerAllocations pass.
Update some more passes to treate free calls just like they were treating FreeInst.
llvm-svn: 85176
2009-10-26 23:43:48 +00:00
e056d1077e
Allow the aggressive anti-dep breaker to process the same region multiple times. This is necessary because new anti-dependencies are exposed when "current" ones are broken.
...
llvm-svn: 85166
2009-10-26 22:31:16 +00:00
661ea989e9
Define virtual destructor in *.cpp file.
...
llvm-svn: 85146
2009-10-26 19:41:00 +00:00
de11f36ab7
Add aggressive anti-dependence breaker. Currently it is not the default for any target. Enable with -break-anti-dependencies=all.
...
llvm-svn: 85145
2009-10-26 19:32:42 +00:00
9aba0d9988
When checking whether a def of an aliased register is dead, ask the
...
machineinstr whether the aliased register is dead, rather than the original
register is dead. This allows it to get the correct answer when examining
an instruction like this:
CALLpcrel32 <ga:foo>, %AL<imp-def>, %EAX<imp-def,dead>
where EAX is dead but a subregister of it is still live. This fixes PR5294.
llvm-svn: 85135
2009-10-26 18:26:18 +00:00
8370485db9
Break anti-dependence breaking out into its own class.
...
llvm-svn: 85127
2009-10-26 16:59:04 +00:00
b81cb2a9a8
- Revert some changes from 85044, 85045, and 85047 that broke x86_64 tests and
...
bootstrapping. It's not safe to leave identity subreg_to_reg and insert_subreg
around.
- Relax register scavenging to allow use of partially "not-live" registers. It's
common for targets to operate on registers where the top bits are undef. e.g.
s0 =
d0 = insert_subreg d0<undef>, s0, 1
...
= d0
When the insert_subreg is eliminated by the coalescer, the scavenger used to
complain. The previous fix was to keep to insert_subreg around. But that's
brittle and it's overly conservative when we want to use the scavenger to
allocate registers. It's actually legal and desirable for other instructions
to use the "undef" part of d0. e.g.
s0 =
d0 = insert_subreg d0<undef>, s0, 1
...
s1 =
= s1
= d0
We probably need add a "partial-undef" marker on machine operand so the
machine verifier would not complain.
llvm-svn: 85091
2009-10-26 04:56:07 +00:00
56869f22c4
Move DataTypes.h to include/llvm/System, update all users. This breaks the last
...
direct inclusion edge from System to Support.
llvm-svn: 85086
2009-10-26 01:35:46 +00:00
4c799e65d7
Use isIdentityCopy. Fix a bozo bug (flipped condition) in InvalidateRegDef.
...
llvm-svn: 85047
2009-10-25 07:51:47 +00:00
5d2245b210
Code clean up.
...
llvm-svn: 85046
2009-10-25 07:49:57 +00:00
19348415c4
Do not delete identity insert_subreg even if dest is virtual. Let later passes delete them. This avoids register scavenger complain.
...
llvm-svn: 85045
2009-10-25 07:48:51 +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
5027d359e7
When the scavenger is looking for a good candidate location to restore from a
...
spill, it should avoid doing so inside the live range of a virtual register.
llvm-svn: 85026
2009-10-25 00:45:07 +00:00
8f4078ba39
Rename isLoopExit to isLoopExiting, for consistency with the wording
...
used elsewhere - an exit block is a block outside the loop branched to
from within the loop. An exiting block is a block inside the loop that
branches out.
llvm-svn: 85019
2009-10-24 23:34:26 +00:00
b4a1347be2
Add some asserts to catch copyRegToReg() fails early
...
llvm-svn: 84983
2009-10-24 00:27:00 +00:00
d638bc310a
Identity copies should not contribute to spill weight.
...
llvm-svn: 84978
2009-10-23 23:09:19 +00:00
4ef112be62
APInt-ify the gep scaling code, so that it correctly handles the case where
...
the scale overflows pointer-sized arithmetic. This fixes PR5281.
llvm-svn: 84954
2009-10-23 17:57:43 +00:00
6aea1a2a5a
Neuter stack protectors by only checking character arrays. This is what GCC
...
does.
llvm-svn: 84916
2009-10-23 00:01:05 +00:00
02ad4cb32e
Allow the target to select the level of anti-dependence breaking that should be performed by the post-RA scheduler. The default is none.
...
llvm-svn: 84911
2009-10-22 23:19:17 +00:00
4f0b8d7861
Tidying up some code and comments. No functionality change.
...
llvm-svn: 84894
2009-10-22 20:48:59 +00:00
8626367e38
Fix null pointer dereference.
...
llvm-svn: 84806
2009-10-22 00:15:17 +00:00
ff97acd8f1
Revert the main portion of r31856. It was causing BranchFolding
...
to break up CFG diamonds by banishing one of the blocks to the end of
the function, which is bad for code density and branch size.
This does pessimize MultiSource/Benchmarks/Ptrdist/yacr2, the
benchmark cited as the reason for the change, however I've examined
the code and it looks more like a case of gaming a particular
branch than of being generally applicable.
llvm-svn: 84803
2009-10-22 00:03:58 +00:00
c16c8c707b
The spill restore needs to be resolved to the SP/FP just like the spill
...
llvm-svn: 84792
2009-10-21 22:59:56 +00:00
bc30143451
Cleanup of frame index scavenging. Better code flow and more accurately
...
handles T2 and ARM use cases.
llvm-svn: 84761
2009-10-21 15:26:21 +00:00
f781bd8947
Need a comma after imp-use.
...
llvm-svn: 84749
2009-10-21 07:56:02 +00:00
682a2d154a
Revert r84658 and r84691. They were causing llvm-gcc bootstrap to fail.
...
llvm-svn: 84727
2009-10-21 01:44:44 +00:00
4388beb884
Respect src register allocation requirements when breaking anti-dependencies. Remove some dead code.
...
llvm-svn: 84691
2009-10-20 22:50:43 +00:00
a6faf60831
Fix invalid for vector types fneg(bitconvert(x)) => bitconvert(x ^ sign)
...
transform.
llvm-svn: 84683
2009-10-20 21:37:45 +00:00
16edcd724a
Oops. Backing out 84681 - needs to wait for the indexing patch.
...
llvm-svn: 84682
2009-10-20 21:28:22 +00:00
d8938d1ca8
Added some debugging output to pre-alloc splitting.
...
llvm-svn: 84681
2009-10-20 21:25:13 +00:00
baf6dd26ea
Checkpoint more aggressive anti-dependency breaking for post-ra scheduler.
...
llvm-svn: 84658
2009-10-20 19:54:44 +00:00
d2ba321930
Better handle instructions that re-def a scratch register
...
llvm-svn: 84657
2009-10-20 19:52:35 +00:00
Dan Gohman