d9d80b1dde
LVI: Recognize the form instcombine canonicalizes range checks into when forming constant ranges.
...
This could probably be made a lot smarter, but this is a common case and doesn't require LVI to scan a lot
of code. With this change CVP can optimize away the "shift == 0" case in Hashing.h that only gets hit when
"shift" is in a range not containing 0.
llvm-svn: 151919
2012-03-02 15:34:43 +00:00
513aaa5691
SwitchInst refactoring.
...
The purpose of refactoring is to hide operand roles from SwitchInst user (programmer). If you want to play with operands directly, probably you will need lower level methods than SwitchInst ones (TerminatorInst or may be User). After this patch we can reorganize SwitchInst operands and successors as we want.
What was done:
1. Changed semantics of index inside the getCaseValue method:
getCaseValue(0) means "get first case", not a condition. Use getCondition() if you want to resolve the condition. I propose don't mix SwitchInst case indexing with low level indexing (TI successors indexing, User's operands indexing), since it may be dangerous.
2. By the same reason findCaseValue(ConstantInt*) returns actual number of case value. 0 means first case, not default. If there is no case with given value, ErrorIndex will returned.
3. Added getCaseSuccessor method. I propose to avoid usage of TerminatorInst::getSuccessor if you want to resolve case successor BB. Use getCaseSuccessor instead, since internal SwitchInst organization of operands/successors is hidden and may be changed in any moment.
4. Added resolveSuccessorIndex and resolveCaseIndex. The main purpose of these methods is to see how case successors are really mapped in TerminatorInst.
4.1 "resolveSuccessorIndex" was created if you need to level down from SwitchInst to TerminatorInst. It returns TerminatorInst's successor index for given case successor.
4.2 "resolveCaseIndex" converts low level successors index to case index that curresponds to the given successor.
Note: There are also related compatability fix patches for dragonegg, klee, llvm-gcc-4.0, llvm-gcc-4.2, safecode, clang.
llvm-svn: 149481
2012-02-01 07:49:51 +00:00
75afc7afe8
Remove dead code.
...
llvm-svn: 148384
2012-01-18 10:10:28 +00:00
58c7569854
A DenseMap of a std::map isn't a very good idea because the "grow()" method will
...
need to make a deep copy of each of the std::maps. Use a std::map of the
std::map instead. This improves the compile time of sqlite3 by ~2%.
llvm-svn: 148003
2012-01-12 01:41:03 +00:00
4ec081a4d2
Revert r147978. A DenseMap's iterators may become invalidated here.
...
llvm-svn: 147980
2012-01-11 23:43:34 +00:00
f0275df9e3
Use a DenseMap.
...
This appears to improve sqlite3's compile time by ~2%.
llvm-svn: 147978
2012-01-11 22:57:32 +00:00
bbf3c60786
Clear the new cache.
...
llvm-svn: 145771
2011-12-03 15:19:55 +00:00
3664708378
Add a "seen blocks" cache to LVI to avoid a linear scan over the whole cache just to remove no blocks from the maps.
...
-15% on ARMDisassembler.cpp (Release build). It's not that great to add another
layer of caching to the caching-heavy LVI but I don't see a better way.
llvm-svn: 145770
2011-12-03 15:16:45 +00:00
43a33066b4
Fix a few more places where TargetData/TargetLibraryInfo is not being passed.
...
Add FIXMEs to places that are non-trivial to fix.
llvm-svn: 145661
2011-12-02 01:26:24 +00:00
229907cd11
land David Blaikie's patch to de-constify Type, with a few tweaks.
...
llvm-svn: 135375
2011-07-18 04:54:35 +00:00
7a5fc693f9
llvm.memcpy.* has two distinct associated address spaces; the source address space, and the destination address space. Fix up the interface on MemIntrinsic and MemTransferInst to make this clear, and fix InstructionDereferencesPointer in LazyValueInfo.cpp to use the interface properly.
...
llvm-svn: 132356
2011-05-31 20:40:16 +00:00
1a180156b6
Remove unused STL header includes.
...
llvm-svn: 130068
2011-04-23 19:53:52 +00:00
2b1ba48f8d
Mark some functions as used which are used within debug-only code. This
...
silences Clang's -Wunused-function when building in release mode.
llvm-svn: 129709
2011-04-18 18:49:44 +00:00
367f98f000
Teach LazyValueInfo that allocas aren't NULL. Over all of llvm-test, this saves
...
half a million non-local queries, each of which would otherwise have triggered a
linear scan over a basic block.
Also fix a fixme for memory intrinsics which dereference pointers. With this,
we prove that a pointer is non-null because it was dereferenced by an intrinsic
112 times in llvm-test.
llvm-svn: 123533
2011-01-15 09:16:12 +00:00
6f060afbbd
Reorder, rename, and document some members to make this easier to follow.
...
llvm-svn: 122929
2011-01-05 23:26:22 +00:00
e86dacf449
When computing the value on an edge, in certain cases LVI would fail to compute the value range
...
in the predecessor block, leading to an incorrect conclusion for the edge value. Found by inspection.
llvm-svn: 122908
2011-01-05 21:37:18 +00:00
118ac80c81
Re-convert several of LazyValueInfo's internal maps to Dense{Map|Set}, and fix the issue in
...
hasBlockValue() that was causing iterator invalidations. Many thanks to Dimitry Andric for
tracking down those invalidations!
llvm-svn: 122906
2011-01-05 21:15:29 +00:00
c6beda80ff
Speculatively revert the use of DenseMap in LazyValueInfo, which may be causing Linux self-host failures.
...
llvm-svn: 122291
2010-12-20 23:53:19 +00:00
9be3ec6264
Attempt to appease the DragonEgg buildbots.
...
llvm-svn: 122288
2010-12-20 23:23:18 +00:00
813a2c45a8
Convert one of LVI's primary maps to a DenseMap, now that we know are more assured of iterator stability.
...
llvm-svn: 122273
2010-12-20 21:30:54 +00:00
d83f98a51e
More LVI cleanups, including trying to simplify the process of maintaining the OverDefinedCache.
...
llvm-svn: 122256
2010-12-20 19:33:41 +00:00
64c2c5798a
Reuse the reference into the LVI cache throughout the solver subsystem. This is much easier to
...
verify as being safe thanks its recent de-recursivization.
llvm-svn: 122254
2010-12-20 18:18:16 +00:00
55a700b0cf
Make LazyValueInfo non-recursive.
...
llvm-svn: 122120
2010-12-18 01:00:40 +00:00
a4fcd2418d
Move Value::getUnderlyingObject to be a standalone
...
function so that it can live in Analysis instead of
VMCore.
llvm-svn: 121885
2010-12-15 20:02:24 +00:00
11678bd299
Clean up some of LVI:
...
* mergeIn now uses constant folding for constants that are provably not-equal.
* sink some sanity checks from the get*() methods into the mark*() methods, to ensure that we never have a constant/notconstant ConstantInt
* some textual cleanups, whitespace changes, removing "else" after return, that sort of thing.
llvm-svn: 121877
2010-12-15 18:57:18 +00:00
c7ed4dc932
Take the first step towards making LVI non-recursive: get rid of the LVIQuery abstraction.
...
llvm-svn: 121357
2010-12-09 06:14:58 +00:00
df7a4f2515
Now with fewer extraneous semicolons!
...
llvm-svn: 115996
2010-10-07 22:25:06 +00:00
140296f5c0
It is possible, under specific circumstances involving ptrtoint ConstantExpr's, for LVI to end up trying to merge
...
a Constant into a ConstantRange. Handle this conservatively for now, rather than asserting. The testcase is
more complex that I would like, but the manifestation of the problem is sensitive to iteration orders and the state of the
LVI cache, and I have not been able to reproduce it with manually constructed or simplified cases.
Fixes PR8162.
llvm-svn: 114103
2010-09-16 18:28:33 +00:00
a74fa15f32
Clean up some of the PassRegistry implementation, and pImpl-ize it to reduce #include clutter
...
and exposing internal details.
llvm-svn: 113252
2010-09-07 19:16:25 +00:00
ad48e01eef
Add completely hokey binary-and and binary-or operations to ConstantRange and
...
teach LazyValueInfo to use them.
llvm-svn: 113196
2010-09-07 05:39:02 +00:00
65b48b5dfc
zap dead code.
...
llvm-svn: 113073
2010-09-04 18:12:00 +00:00
c725462245
Add support for simplifying a load from a computed value to a load from a global when it
...
is provable that they're equivalent. This fixes PR4855.
llvm-svn: 112994
2010-09-03 19:08:37 +00:00
2912df072d
Remove incorrect and poorly tested code for trying to reason about values on default edges of
...
switches. Just return the conservatively correct answer.
llvm-svn: 112876
2010-09-02 22:16:52 +00:00
a8c896b704
Fix a bug in LazyValueInfo that CorrelatedValuePropagation exposed: In the LVI lattice, undef and the full set ConstantRange should not
...
be treated as equivalent.
llvm-svn: 112843
2010-09-02 18:23:58 +00:00
9517943d11
It is possible to try to merge a not-constant with a constantrage, when dealing with ptrtoint ConstantExpr's.
...
Unfortunately, the only testcase I have for this is huge and doesn't reduce well because the error is
sensitive to iteration-order issues, since the problem only occurs when merging values in a particular order.
llvm-svn: 112489
2010-08-30 17:03:45 +00:00
38f6b7fe3b
Improve the precision of getConstant().
...
llvm-svn: 112323
2010-08-27 23:29:38 +00:00
6ebbd92380
Use LVI to eliminate conditional branches where we've tested a related condition previously. Update tests for this change.
...
This fixes PR5652.
llvm-svn: 112270
2010-08-27 17:12:29 +00:00
4afea9e3c6
In the default address space, any GEP off of null results in a trap value if you try to load it. Thus,
...
any load in the default address space that completes implies that the base value that it GEP'd from
was not null.
llvm-svn: 112015
2010-08-25 01:16:47 +00:00
a10000006e
NULL loads are only invalid in the default address space.
...
llvm-svn: 111972
2010-08-24 22:00:55 +00:00
b695c83de9
Add support for inferring values for the default cases of switches.
...
llvm-svn: 111971
2010-08-24 21:59:42 +00:00
da34de1599
Add support for inferring that a load from a pointer implies that it is not null.
...
llvm-svn: 111959
2010-08-24 20:47:29 +00:00
c62f704576
Don't assume that all constants with integer types are ConstantInts.
...
llvm-svn: 111906
2010-08-24 07:55:44 +00:00
80d19f0905
Use ConstantRange to propagate information through value definitions.
...
llvm-svn: 111425
2010-08-18 21:11:37 +00:00
208636fa33
Inform LazyValueInfo whenever a block is deleted, to avoid dangling pointer issues.
...
llvm-svn: 111382
2010-08-18 18:39:01 +00:00
fa7d44687f
Fix another iterator invalidation that caused a *really* nasty miscompilation in 403.gcc.
...
llvm-svn: 111210
2010-08-16 23:42:33 +00:00
7b974a45db
Fix a subtle use-after-free issue.
...
llvm-svn: 110863
2010-08-11 22:36:04 +00:00
0bd61240e9
Improve indentation.
...
llvm-svn: 110778
2010-08-11 04:24:25 +00:00
5f1dd0967d
Now that we're using ConstantRange to represent potential values, make use of that represenation to
...
create constraints from comparisons other than eq/neq.
llvm-svn: 110742
2010-08-10 23:20:01 +00:00
185fe00633
Switch over to using ConstantRange to track integral values.
...
llvm-svn: 110714
2010-08-10 20:03:09 +00:00
8afac043fb
Add ConstantRange information to the debugging output.
...
llvm-svn: 110598
2010-08-09 20:50:46 +00:00
0f306a45ad
Add the beginnings of infrastructure for range tracking.
...
llvm-svn: 110388
2010-08-05 22:59:19 +00:00
c3a1413ea1
Split the tag and value members of LVILatticeVal in preparation for expanding the lattice to something that won't fit in two bits.
...
llvm-svn: 110383
2010-08-05 22:10:46 +00:00
c1561b8400
Add an initial implementation of PHI translation for LazyValueInfo. This involves rolling back some
...
of my earlier data structure improvements until I can ensure that there are no iterator invalidation problems.
llvm-svn: 109935
2010-07-30 23:59:40 +00:00
e4a0ab69d2
Revert my last two patches to LVI, which recent changes have exposed a miscompilation in.
...
llvm-svn: 109889
2010-07-30 20:56:07 +00:00
a44f49f189
Pass the queried value by argument rather than in a member, in preparation for supporting PHI translation.
...
llvm-svn: 109701
2010-07-28 23:50:08 +00:00
6982dd4e1f
Get rid of LVIQuery as a distinct data structure, so that we don't have to initialize a new set of maps on every query.
...
llvm-svn: 109679
2010-07-28 22:07:25 +00:00
aac5a72139
Rearrange several datastructures in LazyValueInfo to improve compile time.
...
This is still not perfect, but better than it was before.
llvm-svn: 109563
2010-07-27 23:58:11 +00:00
aa7f66ba67
Add an initial implementation of LazyValueInfo updating for JumpThreading. Disabled for now.
...
llvm-svn: 109424
2010-07-26 18:48:03 +00:00
a57b97e7e7
Fix batch of converting RegisterPass<> to INTIALIZE_PASS().
...
llvm-svn: 109045
2010-07-21 22:09:45 +00:00
37e9809294
Convert debug messages to use dbgs(). Generally this means
...
s/errs/dbgs/g except for certain special cases.
llvm-svn: 92040
2009-12-23 20:43:58 +00:00
e58c05780e
typo spotted by duncan.
...
llvm-svn: 88884
2009-11-16 03:51:42 +00:00
7735878591
teach LVI to infer edge information from switch instructions.
...
This allows JT to eliminate a ton of infeasible edges when
handling code like the templates in PatternMatch.h
llvm-svn: 88869
2009-11-15 20:02:12 +00:00
b0c0a0df3e
fix a logic error that would cause LVI-JT to miscompile
...
some conditionals
llvm-svn: 88868
2009-11-15 20:01:24 +00:00
2c708562f5
implement the first stab at caching queries. This isn't correct
...
(because the invalidation logic is missing) but LVI isn't enabled
by default anyway.
llvm-svn: 88867
2009-11-15 20:00:52 +00:00
af025d3f6e
refactor a bunch of code forming the new LazyValueInfoCache
...
and LVIQuery classes, no functionality change.
llvm-svn: 88866
2009-11-15 19:59:49 +00:00
22db4b5e0c
various fixes to the lattice transfer functions.
...
llvm-svn: 86952
2009-11-12 04:57:13 +00:00
565ee2f1bd
Add a new getPredicateOnEdge method which returns more rich information for
...
constant constraints. Improve the LVI lattice to include inequality
constraints.
llvm-svn: 86950
2009-11-12 04:36:58 +00:00
d5e25436a1
expose edge information and switch j-t to use it.
...
llvm-svn: 86920
2009-11-12 01:29:10 +00:00
b584d1e456
move some stuff into DEBUG's and turn on lazy-value-info for
...
the basic.ll testcase.
llvm-svn: 86918
2009-11-12 01:22:16 +00:00
19019eaf00
make LazyValueInfo actually to some stuff. This isn't very tested but improves
...
strswitch.
llvm-svn: 86889
2009-11-11 22:48:44 +00:00
fde1f8d0d8
stub out some LazyValueInfo interfaces, and have JumpThreading
...
start using them in a trivial way when -enable-jump-threading-lvi
is passed. enable-jump-threading-lvi will be my playground for
awhile.
llvm-svn: 86789
2009-11-11 02:08:33 +00:00
741c94c719
Stub out a new lazy value info pass, which will eventually
...
vend value constraint information to the optimizer.
llvm-svn: 86767
2009-11-11 00:22:30 +00:00
Benjamin Kramer