108423a94a
Change ConstantArray::get to form a ConstantDataArray when possible,
...
kicking in the big win of ConstantDataArray. As part of this, change
the implementation of GetConstantStringInfo in ValueTracking to work
with ConstantDataArray (and not ConstantArray) making it dramatically,
amazingly, more efficient in the process and renaming it to
getConstantStringInfo.
This keeps around a GetConstantStringInfo entrypoint that (grossly)
forwards to getConstantStringInfo and constructs the std::string
required, but existing clients should move over to
getConstantStringInfo instead.
llvm-svn: 149351
2012-01-31 04:42:22 +00:00
61a1d6cb81
progress making the world safe to ConstantDataVector. While
...
we're at it, allow PatternMatch's "neg" pattern to match integer
vector negations, and enhance ComputeNumSigned bits to handle
shl of vectors.
llvm-svn: 149082
2012-01-26 21:37:55 +00:00
6705883ad8
use Constant::getAggregateElement to simplify a bunch of code.
...
llvm-svn: 148934
2012-01-25 06:48:06 +00:00
9be59599b3
Use the right method to get the # elements in a CDS.
...
llvm-svn: 148897
2012-01-25 01:27:20 +00:00
f7eb543380
teach valuetracking about ConstantDataSequential
...
llvm-svn: 148790
2012-01-24 07:54:10 +00:00
7ac046a261
Generalize isSafeToSpeculativelyExecute to work on arbitrary
...
Values, rather than just Instructions, since it's interesting
for ConstantExprs too.
llvm-svn: 147560
2012-01-04 23:01:09 +00:00
9442cd01f6
PatternMatch: Introduce a matcher for instructions with the "exact" bit. Use it to simplify a few matchers.
...
llvm-svn: 147403
2012-01-01 17:55:30 +00:00
4ee5747fdd
ComputeMaskedBits: Make knownzero computation more aggressive for ctlz with undef zero.
...
unsigned foo(unsigned x) { return 31 - __builtin_clz(x); }
now compiles into a single "bsrl" instruction on x86.
llvm-svn: 147255
2011-12-24 17:31:46 +00:00
b4039f633c
Make some intrinsics safe to speculatively execute.
...
llvm-svn: 147036
2011-12-21 05:52:02 +00:00
75d7d5e988
Move Instruction::isSafeToSpeculativelyExecute out of VMCore and
...
into Analysis as a standalone function, since there's no need for
it to be in VMCore. Also, update it to use isKnownNonZero and
other goodies available in Analysis, making it more precise,
enabling more aggressive optimization.
llvm-svn: 146610
2011-12-14 23:49:11 +00:00
8abf65a130
Probably not a good idea to convert a single vector load into a memcpy. We
...
don't do this now, but add a test case to prevent this from happening in the
future.
Additional test for rdar://9892684
llvm-svn: 145879
2011-12-06 00:19:08 +00:00
3924cb0267
Add support for vectors of pointers.
...
llvm-svn: 145801
2011-12-05 06:29:09 +00:00
ca6f8ddbf8
Fix a theoretical problem (not seen in the wild): if different instances of a
...
weak variable are compiled by different compilers, such as GCC and LLVM, while
LLVM may increase the alignment to the preferred alignment there is no reason to
think that GCC will use anything more than the ABI alignment. Since it is the
GCC version that might end up in the final program (as the linkage is weak), it
is wrong to increase the alignment of loads from the global up to the preferred
alignment as the alignment might only be the ABI alignment.
Increasing alignment up to the ABI alignment might be OK, but I'm not totally
convinced that it is. It seems better to just leave the alignment of weak
globals alone.
llvm-svn: 145413
2011-11-29 18:26:38 +00:00
e7ab1a2f0f
Make SelectionDAG::InferPtrAlignment use llvm::ComputeMaskedBits instead of duplicating the logic for globals. Make llvm::ComputeMaskedBits handle GlobalVariables slightly more aggressively, to match what InferPtrAlignment knew how to do.
...
llvm-svn: 145304
2011-11-28 22:48:22 +00:00
81a2af12d6
Fix a crash in which a multiplication was being reported as being both negative
...
and positive: positive, because it could be directly computed to be positive;
negative, because the nsw flags means it is either negative or undefined (the
multiplication always overflowed).
llvm-svn: 145104
2011-11-23 16:26:47 +00:00
985ba6386d
A shift of a power of two is a power of two or zero.
...
For completeness - not spotted in the wild.
llvm-svn: 143211
2011-10-28 18:30:05 +00:00
7cb61e5a0e
Reapply commit 143028 with a fix: the problem was casting a ConstantExpr Mul
...
using BinaryOperator (which only works for instructions) when it should have
been a cast to OverflowingBinaryOperator (which also works for constants).
While there, correct a few other dubious looking uses of BinaryOperator.
Thanks to Chad Rosier for the testcase. Original commit message:
My super-optimizer noticed that we weren't folding this expression to
true: (x *nsw x) sgt 0, where x = (y | 1). This occurs in 464.h264ref.
llvm-svn: 143125
2011-10-27 19:16:21 +00:00
1455ce27e4
Revert Duncan's r143028 expression folding which appears to be the culprit
...
behind a compile failure on 483.xalancbmk.
llvm-svn: 143102
2011-10-27 15:47:25 +00:00
ba286d7c73
The maximum power of 2 dividing a power of 2 is itself. This occurs
...
in 403.gcc and was spotted by my super-optimizer.
llvm-svn: 143054
2011-10-26 20:55:21 +00:00
1d2bb9882d
My super-optimizer noticed that we weren't folding this expression to
...
true: (x *nsw x) sgt 0, where x = (y | 1). This occurs in 464.h264ref.
llvm-svn: 143028
2011-10-26 15:31:51 +00:00
462098824f
"@string = constant i8 0" is a value i8* string of length zero. Analyze that
...
correctly in GetStringLength, fixing PR11181!
llvm-svn: 142558
2011-10-20 00:34:35 +00:00
717d7edd3e
Migrate LLVM and Clang to use the new makeArrayRef(...) functions where previously explicit non-default constructors were used.
...
Mostly mechanical with some manual reformatting.
llvm-svn: 135390
2011-07-18 12:00:32 +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
57aa636794
Convert InsertValueInst and ExtractValueInst APIs to use ArrayRef.
...
llvm-svn: 135040
2011-07-13 10:26:04 +00:00
3e334a42d7
Move onlyUsedByLifetimeMarkers to ValueTracking so that it can be used by other
...
passes as well.
llvm-svn: 133904
2011-06-27 04:20:45 +00:00
b362884ca9
Renamed llvm.x86.sse42.crc32 intrinsics; crc64 doesn't exist.
...
crc32.[8|16|32] have been renamed to .crc32.32.[8|16|32] and
crc64.[8|16|32] have been renamed to .crc32.64.[8|64].
llvm-svn: 132163
2011-05-26 23:13:19 +00:00
83791ced7b
Teach valuetracking that byval arguments with a specified alignment are
...
aligned.
Teach memcpyopt to not give up all hope when confonted with an underaligned
memcpy feeding an overaligned byval. If the *source* of the memcpy can be
determined to be adequeately aligned, or if it can be forced to be, we can
eliminate the memcpy.
This addresses PR9794. We now compile the example into:
define i32 @f(%struct.p* nocapture byval align 8 %q) nounwind ssp {
entry:
%call = call i32 @g(%struct.p* byval align 8 %q) nounwind
ret i32 %call
}
in both x86-64 and x86-32 mode. We still don't get a tailcall though,
because tailcalls apparently can't handle byval.
llvm-svn: 131884
2011-05-23 00:03:39 +00:00
2a746bfe36
Teach ValueTracking about x86 crc32 intrinsics.
...
llvm-svn: 131861
2011-05-22 18:25:30 +00:00
0ab5e2cded
Fix a ton of comment typos found by codespell. Patch by
...
Luis Felipe Strano Moraes!
llvm-svn: 129558
2011-04-15 05:18:47 +00:00
8baa2c7ad9
Don't assume something which might be a constant expression is an instruction.
...
Based on PR9429, but no testcase because I can't figure out how to trigger it
anymore given other changes to the relevant code.
llvm-svn: 128781
2011-04-02 22:11:56 +00:00
f0469af63e
Fix INT_MIN gotcha pointed out by Eli Friedman.
...
llvm-svn: 128028
2011-03-21 21:40:32 +00:00
5acc751b6f
Teach ComputeMaskedBits about sub nsw.
...
llvm-svn: 127548
2011-03-12 17:18:11 +00:00
391a946fa9
ComputeMaskedBits: sub falls through to add, and sub doesn't have the same overflow semantics as add.
...
Should fix the selfhost failures that started with r127463.
llvm-svn: 127465
2011-03-11 14:46:49 +00:00
cc79973856
Teach ComputeMaskedBits about nsw on add. I don't think there's anything we can
...
do with nuw here, but sub and mul should be given similar treatment.
Fixes PR9343 #15 !
llvm-svn: 127463
2011-03-11 09:00:19 +00:00
7dc3d47c34
Fix PR9331. Simplified version of a patch by Jakub Staszak.
...
llvm-svn: 127243
2011-03-08 12:39:03 +00:00
e467979d0a
Add more analysis of the sign bit of an srem instruction. If the LHS is negative
...
then the result could go either way. If it's provably positive then so is the
srem. Fixes PR9343 #7 !
llvm-svn: 127146
2011-03-07 01:50:10 +00:00
afe4a3062d
Fix comment.
...
llvm-svn: 126645
2011-02-28 09:18:11 +00:00
66f4f22f7b
srem doesn't actually have the same resulting sign as its numerator, you could
...
also have a zero when numerator = denominator. Reverts parts of r126635 and
r126637.
llvm-svn: 126644
2011-02-28 09:17:39 +00:00
c9aab8567b
Teach value tracking to make use of flags in more situations.
...
llvm-svn: 126642
2011-02-28 08:02:21 +00:00
29dbbd12c1
Teach ValueTracking to look at the dividend when determining the sign bit of an
...
srem instruction.
llvm-svn: 126637
2011-02-28 06:52:12 +00:00
acf6b0776a
Stores of null pointers should turn into memset, we weren't recognizing
...
them as splat values.
llvm-svn: 126041
2011-02-19 19:35:49 +00:00
ac0b62c277
Tolerate degenerate phi nodes that can occur in the middle of optimization
...
passes. Fixes PR9112. Patch by Jakub Staszak!
llvm-svn: 125319
2011-02-10 23:54:10 +00:00
4b397fcdc2
Add a m_SignBit pattern for convenience.
...
llvm-svn: 124656
2011-02-01 08:50:33 +00:00
2e9e4f1be3
Fix typo: should have been testing that X was odd, not V.
...
llvm-svn: 124533
2011-01-29 13:27:00 +00:00
8a33733228
APInt has a method for determining whether a number is a power of 2
...
which is more efficient than countPopulation - use it.
llvm-svn: 124283
2011-01-26 08:44:16 +00:00
9e9d5b25e2
In which I discover that zero+zero is zero, d'oh!
...
llvm-svn: 124188
2011-01-25 15:14:15 +00:00
fced7620f5
See if this fixes llvm-gcc bootstrap.
...
llvm-svn: 124184
2011-01-25 12:15:09 +00:00
d395108394
According to my auto-simplifier the most common missed simplifications in
...
optimized code are:
(non-negative number)+(power-of-two) != 0 -> true
and
(x | 1) != 0 -> true
Instcombine knows about the second one of course, but only does it if X|1
has only one use. These fire thousands of times in the testsuite.
llvm-svn: 124183
2011-01-25 09:38:29 +00:00
0f124e1987
Give GetUnderlyingObject a TargetData, to keep it in sync
...
with BasicAA's DecomposeGEPExpression, which recently began
using a TargetData. This fixes PR8968, though the testcase
is awkward to reduce.
Also, update several off GetUnderlyingObject's users
which happen to have a TargetData handy to pass it in.
llvm-svn: 124134
2011-01-24 18:53:32 +00:00
c86e67e110
fix an off-by-one bug that caused a crash analyzing
...
ashr's with huge shift amounts, PR8896
llvm-svn: 122814
2011-01-04 18:19:15 +00:00
9cb1035f94
move isBytewiseValue out to ValueTracking.h/cpp
...
llvm-svn: 122565
2010-12-26 20:15:01 +00:00
7aa18bf46a
Add vector versions of some existing scalar transforms to aid codegen in matching psign & pblend operations to the IR produced by clang/gcc for their C idioms.
...
llvm-svn: 122105
2010-12-17 23:12:19 +00:00
05b18f143f
Reapply r121886, and also update DecomposeGEPExpression to keep
...
it in sync.
llvm-svn: 121895
2010-12-15 20:49:55 +00:00
d02b65982e
Revert r121886. DecomposeGEPExpression needs to be kept
...
in sync.
llvm-svn: 121892
2010-12-15 20:39:25 +00:00
949ab7889c
Strengthen GetUnderlyingObject using InstructionSimplify.
...
While LLVM's main design is that analysis code shouldn't
go out of its way to understand code which hasn't been
InstCombined, analysis utility routines like this can
find themselves being called in the middle of transform
passes when instcombine hasn't had a chance to run.
llvm-svn: 121886
2010-12-15 20:10:26 +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
583abbc4df
PR5207: Change APInt methods trunc(), sext(), zext(), sextOrTrunc() and
...
zextOrTrunc(), and APSInt methods extend(), extOrTrunc() and new method
trunc(), to be const and to return a new value instead of modifying the
object in place.
llvm-svn: 121120
2010-12-07 08:25:19 +00:00
25a5e4ca1f
PR5207: Rename overloaded APInt methods set(), clear(), flip() to
...
setAllBits(), setBit(unsigned), etc.
llvm-svn: 120564
2010-12-01 08:53:58 +00:00
e28618de59
move GetPointerBaseWithConstantOffset out of GVN into ValueTracking.h
...
llvm-svn: 120476
2010-11-30 22:25:26 +00:00
15084f085d
PR5207: Make APInt::set(), APInt::clear() and APInt::flip() return void.
...
llvm-svn: 120413
2010-11-30 09:02:01 +00:00
72d283c826
fix PR8063, a crash in globalopt in the malloc analysis code.
...
llvm-svn: 113109
2010-09-05 17:20:46 +00:00
9f7500f57b
move gep decomposition out of ValueTracking into BasicAA. The form of
...
decomposition that it is doing is very basicaa specific and is only used
by basicaa.
Now with less tree breakingness.
llvm-svn: 111433
2010-08-18 22:07:29 +00:00
fbeeb130d8
Revert r111375, "move gep decomposition out of ValueTracking into BasicAA. The
...
form of", it doesn't pass tests.
llvm-svn: 111385
2010-08-18 18:43:08 +00:00
54fe883203
move gep decomposition out of ValueTracking into BasicAA. The form of
...
decomposition that it is doing is very basicaa specific and is only used
by basicaa.
llvm-svn: 111375
2010-08-18 18:22:17 +00:00
a33edcb56c
fix PR7589: In brief:
...
gep P, (zext x) != gep P, (sext x)
DecomposeGEPExpression was getting this wrong, confusing
basicaa.
llvm-svn: 111352
2010-08-18 04:28:19 +00:00
1abbde3103
use ArgOperand accessors
...
llvm-svn: 106697
2010-06-23 23:38:07 +00:00
7258dcd77f
Revert 101465, it broke internal OpenGL testing.
...
Probably the best way to know that all getOperand() calls have been handled
is to replace that API instead of updating.
llvm-svn: 101579
2010-04-16 23:37:20 +00:00
f375520f7b
reapply r101434
...
with a fix for self-hosting
rotate CallInst operands, i.e. move callee to the back
of the operand array
the motivation for this patch are laid out in my mail to llvm-commits:
more efficient access to operands and callee, faster callgraph-construction,
smaller compiler binary
llvm-svn: 101465
2010-04-16 15:33:14 +00:00
403e9694f9
back out r101423 and r101397, they break llvm-gcc self-host on darwin10
...
llvm-svn: 101434
2010-04-16 01:16:20 +00:00
33ae80bff7
reapply r101364, which has been backed out in r101368
...
with a fix
rotate CallInst operands, i.e. move callee to the back
of the operand array
the motivation for this patch are laid out in my mail to llvm-commits:
more efficient access to operands and callee, faster callgraph-construction,
smaller compiler binary
llvm-svn: 101397
2010-04-15 20:51:13 +00:00
9fd00c7d25
back out r101364, as it trips the linux nightlybot on some clang C++ tests
...
llvm-svn: 101368
2010-04-15 12:46:56 +00:00
aafd209632
rotate CallInst operands, i.e. move callee to the back
...
of the operand array
the motivation for this patch are laid out in my mail to llvm-commits:
more efficient access to operands and callee, faster callgraph-construction,
smaller compiler binary
llvm-svn: 101364
2010-04-15 10:49:53 +00:00
0b4df0425f
Constify GetConstantStringInfo.
...
llvm-svn: 101298
2010-04-14 22:20:45 +00:00
2a65429671
Fix a typo in ValueTracking that's causing instcombine to delete needed shift instructions.
...
llvm-svn: 98416
2010-03-13 02:20:29 +00:00
4899cbc77d
Move GetStringLength and helper from SimplifyLibCalls to ValueTracking.
...
No functionality change.
llvm-svn: 97793
2010-03-05 06:58:57 +00:00
19d0b47b1f
There are two ways of checking for a given type, for example isa<PointerType>(T)
...
and T->isPointerTy(). Convert most instances of the first form to the second form.
Requested by Chris.
llvm-svn: 96344
2010-02-16 11:11:14 +00:00
9dff9bec31
Uniformize the names of type predicates: rather than having isFloatTy and
...
isInteger, we now have isFloatTy and isIntegerTy. Requested by Chris!
llvm-svn: 96223
2010-02-15 16:12:20 +00:00
26cd6bd0b0
It looks like the changes to the SRem logic of SimplifyDemandedUseBits
...
(fix for PR6165) are needed here too.
llvm-svn: 94801
2010-01-29 06:18:37 +00:00
35d3b9dcd0
teach ComputeNumSignBits to look through PHI nodes.
...
llvm-svn: 92964
2010-01-07 23:44:37 +00:00
d2564e3afb
Move remaining stuff to the isInteger predicate.
...
llvm-svn: 92771
2010-01-05 21:05:54 +00:00
a3aef788ec
Fix GetConstantStringInfo to not look into MDString (it works on
...
real data, not metadata) and fix DbgInfoPrinter to not abuse
GetConstantStringInfo.
llvm-svn: 91444
2009-12-15 19:34:20 +00:00
bb3eac9e7a
Fixed an assertion failure for tracking sext of a vector of integers
...
llvm-svn: 90290
2009-12-02 04:59:58 +00:00
c7bc66dfc6
implement a FIXME: limit the depth that DecomposeGEPExpression goes the same
...
way that getUnderlyingObject does it.
This fixes the 'DecomposeGEPExpression and getUnderlyingObject disagree!'
assertion on sqlite3.
llvm-svn: 90038
2009-11-28 15:12:41 +00:00
8e62d0a93d
limit the recursion depth of GetLinearExpression. This
...
fixes a crash analyzing consumer-lame, which had an "%X = add %X, 1"
in unreachable code.
llvm-svn: 90000
2009-11-27 08:32:52 +00:00
1bf7ff704a
Implement PR1143 (at -m64) by making basicaa look through extensions. We
...
previously already handled it at -m32 because there were no i32->i64
extensions for addressing.
llvm-svn: 89959
2009-11-26 18:53:33 +00:00
d7cabea8ca
fix two transposed lines duncan caught and add an explanatory comment.
...
llvm-svn: 89958
2009-11-26 18:35:46 +00:00
9aa846956e
this todo is resolved.
...
llvm-svn: 89957
2009-11-26 17:14:10 +00:00
29f82a1465
move DecomposeGEPExpression out into ValueTracking.cpp
...
llvm-svn: 89956
2009-11-26 17:12:50 +00:00
4f0b47d9e7
remove some redundant braces
...
llvm-svn: 89912
2009-11-26 01:50:12 +00:00
39dbfd3c58
Remove unused LLVMContext.
...
llvm-svn: 89642
2009-11-23 03:29:18 +00:00
6a976bbcb7
Simplify ComputeMultiple so that it doesn't depend on TargetData.
...
llvm-svn: 89175
2009-11-18 00:58:27 +00:00
4744488e8d
Add ComputeMultiple() analysis function that recursively determines if a Value V is a multiple of unsigned Base
...
llvm-svn: 86675
2009-11-10 08:28:35 +00:00
8acf2956b8
Remove AllocationInst. Since MallocInst went away, AllocaInst is the only subclass of AllocationInst, so it no longer is necessary.
...
llvm-svn: 84969
2009-10-23 21:09:37 +00:00
a3aaf85e23
Remove MallocInst from LLVM Instructions.
...
llvm-svn: 84299
2009-10-17 01:18:07 +00:00
f6a987b784
Handle sqrt in CannotBeNegativeZero. absf and absl
...
appear to be misspellings, removed in favor of fabs*.
llvm-svn: 82796
2009-09-25 20:54:50 +00:00
94262db4d4
Teach ValueTracking how to look through GlobalAliases. GlobalAliases are
...
not folded in the constant folder because the constant folder doesn't
simplify ConstantExpr operands.
llvm-svn: 81864
2009-09-15 16:14:44 +00:00
0cdbc7a2ca
fix ComputeMaskedBits handling of zext/sext/trunc to work with vectors.
...
This fixes PR4905
llvm-svn: 81174
2009-09-08 00:13:52 +00:00
4bc2825d0b
add some comments to describe the invariants.
...
llvm-svn: 81173
2009-09-08 00:06:16 +00:00
18e03dd70c
Fix DbgStopPointInst->getFileName/getDirectory, broken by the MDNodification in
...
r80406, and readd a -print-dbginfo test.
llvm-svn: 80778
2009-09-02 11:13:56 +00:00
05f11351ab
Handle TargetData with const.
...
llvm-svn: 80262
2009-08-27 17:51:25 +00:00
Chris Lattner