positive and negative forms of constants together. This
allows us to compile:
int foo(int x, int y) {
return (x-y) + (x-y) + (x-y);
}
into:
_foo: ## @foo
subl %esi, %edi
leal (%rdi,%rdi,2), %eax
ret
instead of (where the 3 and -3 were not factored):
_foo:
imull $-3, 8(%esp), %ecx
imull $3, 4(%esp), %eax
addl %ecx, %eax
ret
this started out as:
movl 12(%ebp), %ecx
imull $3, 8(%ebp), %eax
subl %ecx, %eax
subl %ecx, %eax
subl %ecx, %eax
ret
This comes from PR5359.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92381 91177308-0d34-0410-b5e6-96231b3b80d8
getMDKindID/getMDKindNames methods to LLVMContext (and add
convenience methods to Module), eliminating MetadataContext.
Move the state that it maintains out to LLVMContext.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92259 91177308-0d34-0410-b5e6-96231b3b80d8
I asked Devang to do back on Sep 27. Instead of going through the
MetadataContext class with methods like getMD() and getMDs(), just
ask the instruction directly for its metadata with getMetadata()
and getAllMetadata().
This includes a variety of other fixes and improvements: previously
all Value*'s were bloated because the HasMetadata bit was thrown into
value, adding a 9th bit to a byte. Now this is properly sunk down to
the Instruction class (the only place where it makes sense) and it
will be folded away somewhere soon.
This also fixes some confusion in getMDs and its clients about
whether the returned list is indexed by the MDID or densely packed.
This is now returned sorted and densely packed and the comments make
this clear.
This introduces a number of fixme's which I'll follow up on.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92235 91177308-0d34-0410-b5e6-96231b3b80d8
non-templated IRBuilderBase class. Move that large CreateGlobalString
out of line, eliminating the need to #include GlobalVariable.h in IRBuilder.h
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92227 91177308-0d34-0410-b5e6-96231b3b80d8
SDISel. This optimization was causing simplifylibcalls to
introduce type-unsafe nastiness. This is the first step, I'll be
expanding the memcmp optimizations shortly, covering things that
we really really wouldn't want simplifylibcalls to do.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92098 91177308-0d34-0410-b5e6-96231b3b80d8
load is needed when we have a small store into a large alloca (at which
point we get a load/insert/store sequence), but when you do a full-sized
store, this load ends up being dead.
This dead load is bad in really large nasty testcases where the load ends
up causing mem2reg to insert large chains of dependent phi nodes which only
ADCE can delete. Instead of doing this, just don't insert the dead load.
This fixes rdar://6864035
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91917 91177308-0d34-0410-b5e6-96231b3b80d8
missing check that an array reference doesn't go past the end of the array,
and remove some redundant checks for in-bound array and vector references
that are no longer needed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91897 91177308-0d34-0410-b5e6-96231b3b80d8
by merging all returns in a function into a single one, but simplifycfg
currently likes to duplicate the return (an unfortunate choice!)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91890 91177308-0d34-0410-b5e6-96231b3b80d8
instead of stored. This reduces memdep memory usage, and also eliminates a bunch of
weakvh's. This speeds up gvn on gcc.c-torture/20001226-1.c from 23.9s to 8.45s (2.8x)
on a different machine than earlier.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91885 91177308-0d34-0410-b5e6-96231b3b80d8
load to avoid even messing around with SSAUpdate at all. In this case (which
is very common, we can just use the input value directly).
This speeds up GVN time on gcc.c-torture/20001226-1.c from 36.4s to 16.3s,
which still isn't great, but substantially better and this is a simple speedup
that applies to lots of different cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91851 91177308-0d34-0410-b5e6-96231b3b80d8
two-element arrays. After restructuring the SROA code, it was not safe to
do this without adding more checking. It is not clear that this special-case
has really been useful, and removing this simplifies the code quite a bit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91828 91177308-0d34-0410-b5e6-96231b3b80d8
'GetValueInMiddleOfBlock' case, instead of inserting
duplicates.
A similar fix is almost certainly needed by the machine-level
SSAUpdate implementation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91820 91177308-0d34-0410-b5e6-96231b3b80d8
implement some optimizations for MIN(MIN()) and MAX(MAX()) and
MIN(MAX()) etc. This substantially improves the code in PR5822 but
doesn't kick in much elsewhere. 2 max's were optimized in
pairlocalalign and one in smg2000.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91814 91177308-0d34-0410-b5e6-96231b3b80d8
Use the presence of NSW/NUW to fold "icmp (x+cst), x" to a constant in
cases where it would otherwise be undefined behavior.
Surprisingly (to me at least), this triggers hundreds of the times in
a few benchmarks: lencode, ldecode, and 466.h264ref seem to *really*
like this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91812 91177308-0d34-0410-b5e6-96231b3b80d8
a bunch in lencode, ldecod, spass, 176.gcc, 252.eon, among others. It is
also the first part of PR5822
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91811 91177308-0d34-0410-b5e6-96231b3b80d8
where instcombine would have to split a critical edge due to a
phi node of an invoke. Since instcombine can't change the CFG,
it has to bail out from doing the transformation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91763 91177308-0d34-0410-b5e6-96231b3b80d8
* change FindElementAndOffset to return a uint64_t instead of unsigned, and
to identify the type to be used for that result in a GEP instruction.
* move "isa<ConstantInt>" to be first in conditional.
* replace some dyn_casts with casts.
* add a comment about handling mem intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91762 91177308-0d34-0410-b5e6-96231b3b80d8
bootstrap. This also replaces the WeakVH references that Chris objected to
with normal Value references.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91711 91177308-0d34-0410-b5e6-96231b3b80d8
contains another loop, or an instruction. The loop form is
substantially more efficient on large loops than the typical
code it replaces.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91654 91177308-0d34-0410-b5e6-96231b3b80d8
of 91296 that caused trouble -- the Processed list needs to be
preserved for the livetime of the pass, as AddUsersIfInteresting
is called from other passes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91641 91177308-0d34-0410-b5e6-96231b3b80d8
to memcpy. (Such a memcpy is technically illegal, but in practice is safe
and is generated by struct self-assignment in C code.)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91621 91177308-0d34-0410-b5e6-96231b3b80d8
problem", this broke llvm-gcc bootstrap for release builds on
x86_64-apple-darwin10.
This reverts commit db22309800b224a9f5f51baf76071d7a93ce59c9.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91534 91177308-0d34-0410-b5e6-96231b3b80d8
found last time. Instead of trying to modify the IR while iterating over it,
I've change it to keep a list of WeakVH references to dead instructions, and
then delete those instructions later. I also added some special case code to
detect and handle the situation when both operands of a memcpy intrinsic are
referencing the same alloca.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91459 91177308-0d34-0410-b5e6-96231b3b80d8
isPodLike type trait. This is a generally useful type trait for
more than just DenseMap, and we really care about whether something
acts like a pod, not whether it really is a pod.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91421 91177308-0d34-0410-b5e6-96231b3b80d8
While scanning through the uses of an alloca, keep track of the current offset
relative to the start of the alloca, and check memory references to see if
the offset & size correspond to a component within the alloca. This has the
nice benefit of unifying much of the code from isSafeUseOfAllocation,
isSafeElementUse, and isSafeUseOfBitCastedAllocation. The code to rewrite
the uses of a promoted alloca, after it is determined to be safe, is
reorganized in the same way.
Also, when rewriting GEP instructions, mark them as "in-bounds" since all the
indices are known to be safe.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91184 91177308-0d34-0410-b5e6-96231b3b80d8
value size. This only manifested when memdep inprecisely returns clobber,
which is do to a caching issue in the PR5744 testcase. We can 'efficiently
emulate' this by using '-no-aa'
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@91004 91177308-0d34-0410-b5e6-96231b3b80d8
clobbers to forward pieces of large stores to small loads, we need to consider
the properly phi translated pointer in the store block.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90978 91177308-0d34-0410-b5e6-96231b3b80d8
phi translation of complex expressions like &A[i+1]. This has the
following benefits:
1. The phi translation logic is all contained in its own class with
a strong interface and verification that it is self consistent.
2. The logic is more correct than before. Previously, if intermediate
expressions got PHI translated, we'd miss the update and scan for
the wrong pointers in predecessor blocks. @phi_trans2 is a testcase
for this.
3. We have a lot less code in memdep.
We can handle phi translation across blocks of things like @phi_trans3,
which is pretty insane :).
This patch should fix the miscompiles of 255.vortex, and I tested it
with a bootstrap of llvm-gcc, llvm-test and dejagnu of course.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90926 91177308-0d34-0410-b5e6-96231b3b80d8
I'm not aware that this does anything significant on its own, but it's
needed for another patch that I'm working on.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90864 91177308-0d34-0410-b5e6-96231b3b80d8
handle cases like this:
void test(int N, double* G) {
long j;
for (j = 1; j < N - 1; j++)
G[j+1] = G[j] + G[j+1];
}
where G[1] isn't live into the loop.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90041 91177308-0d34-0410-b5e6-96231b3b80d8
array indexes. The "complex" case of SRoA still handles them, and correctly.
This fixes a weirdness where we'd correctly avoid transforming A[0][42] if
the 42 was too large, but we'd only do it if it was one gep, not two separate
ones.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@90007 91177308-0d34-0410-b5e6-96231b3b80d8
where it is not available. It's unclear how to get this inserted
computation into GVN's scalar availability sets, Owen, help? :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89997 91177308-0d34-0410-b5e6-96231b3b80d8
generates store to undef and some generates store to null as the idiom
for undefined behavior. Since simplifycfg zaps both, don't remove the
undefined behavior in instcombine.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89971 91177308-0d34-0410-b5e6-96231b3b80d8
ConstantExpr, not just the top-level operator. This allows it to
fold many more constants.
Also, make GlobalOpt call ConstantFoldConstantExpression on
GlobalVariable initializers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89659 91177308-0d34-0410-b5e6-96231b3b80d8
it may be used in contexts where preheader insertion may have failed due
to an indirectbr.
Make LoopSimplify's LoopSimplify::SeparateNestedLoop properly fail in
the case that it would require splitting an indirectbr edge.
These fix PR5502.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89484 91177308-0d34-0410-b5e6-96231b3b80d8
tests/Transforms/InstCombine/shufflemask-undef.ll. If
anyone cares, the use of 2*e here (and the equivalent
all over the place in instcombine) seems wrong, though
harmless: it should really be twice the length of the
input vector. I think shufflevector used to require
that the mask have the same length as the input, but I
don't think that's true any more. I don't care enough
about vectors to do anything about this...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89456 91177308-0d34-0410-b5e6-96231b3b80d8
if it is not ultimately captured. Teach BasicAliasAnalysis that a
local object address which does not escape and is never stored does
not alias with a value resulting from a load.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89398 91177308-0d34-0410-b5e6-96231b3b80d8
they are lowered to instruction sequences more complex than a simple
load, such that CodeGen cannot rematerialize them, a reload from a
spill slot is likely to be cheaper than the complex sequence.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@89374 91177308-0d34-0410-b5e6-96231b3b80d8
The Mask and LHSMask may not be of the same size, so don't do the
transformation if they're different.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@88972 91177308-0d34-0410-b5e6-96231b3b80d8
running IPSCCP early, and we run functionattrs interlaced with the inliner,
we often (particularly for small or noop functions) completely propagate
all of the information about a call to its call site in IPSSCP (making a call
dead) and functionattrs is smart enough to realize that the function is
readonly (because it is interlaced with inliner).
To improve compile time and make the inliner threshold more accurate, realize
that we don't have to inline dead readonly function calls. Instead, just
delete the call. This happens all the time for C++ codes, here are some
counters from opt/llvm-ld counting the number of times calls were deleted vs
inlined on various apps:
Tramp3d opt:
5033 inline - Number of call sites deleted, not inlined
24596 inline - Number of functions inlined
llvm-ld:
667 inline - Number of functions deleted because all callers found
699 inline - Number of functions inlined
483.xalancbmk opt:
8096 inline - Number of call sites deleted, not inlined
62528 inline - Number of functions inlined
llvm-ld:
217 inline - Number of allocas merged together
2158 inline - Number of functions inlined
471.omnetpp:
331 inline - Number of call sites deleted, not inlined
8981 inline - Number of functions inlined
llvm-ld:
171 inline - Number of functions deleted because all callers found
629 inline - Number of functions inlined
Deleting a call is much faster than inlining it, and is insensitive to the
size of the callee. :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86975 91177308-0d34-0410-b5e6-96231b3b80d8
cannot be folded into target cmp instruction.
- Avoid a phase ordering issue where early cmp optimization would prevent the
later count-to-zero optimization.
- Add missing checks which could cause LSR to reuse stride that does not have
users.
- Fix a bug in count-to-zero optimization code which failed to find the pre-inc
iv's phi node.
- Remove, tighten, loosen some incorrect checks disable valid transformations.
- Quite a bit of code clean up.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86969 91177308-0d34-0410-b5e6-96231b3b80d8
making the new LVI stuff smart enough to subsume some special
cases in the old code. Disable them when LVI is around, the
testcase still passes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86951 91177308-0d34-0410-b5e6-96231b3b80d8
llvm.invariant.start to be used without necessarily being paired with a call
to llvm.invariant.end. If you run the entire optimization pipeline then such
calls are in fact deleted (adce does it), but that's actually a good thing since
we probably do want them to be zapped late in the game. There should really be
an integration test that checks that the llvm.invariant.start call lasts long
enough that all passes that do interesting things with it get to do their stuff
before it is deleted. But since no passes do anything interesting with it yet
this will have to wait for later.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86840 91177308-0d34-0410-b5e6-96231b3b80d8
start using them in a trivial way when -enable-jump-threading-lvi
is passed. enable-jump-threading-lvi will be my playground for
awhile.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86789 91177308-0d34-0410-b5e6-96231b3b80d8
debug intrinsics, and an unconditional branch when possible. This
reuses the TryToSimplifyUncondBranchFromEmptyBlock function split
out of simplifycfg.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86722 91177308-0d34-0410-b5e6-96231b3b80d8
just one level deep. On the testcase we go from getting this:
F1: ; preds = %T2
%F = and i1 true, %cond ; <i1> [#uses=1]
br i1 %F, label %X, label %Y
to a fully threaded:
F1: ; preds = %T2
br label %Y
This changes gets us to the point where we're forming (too many) switch
instructions on doug's strswitch testcase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86646 91177308-0d34-0410-b5e6-96231b3b80d8
except that the result may not be a constant. Switch jump threading to
use it so that it gets things like (X & 0) -> 0, which occur when phi preds
are deleted and the remaining phi pred was a zero.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86637 91177308-0d34-0410-b5e6-96231b3b80d8
This patch forbids implicit conversion of DenseMap::const_iterator to
DenseMap::iterator which was possible because DenseMapIterator inherited
(publicly) from DenseMapConstIterator. Conversion the other way around is now
allowed as one may expect.
The template DenseMapConstIterator is removed and the template parameter
IsConst which specifies whether the iterator is constant is added to
DenseMapIterator.
Actually IsConst parameter is not necessary since the constness can be
determined from KeyT but this is not relevant to the fix and can be addressed
later.
Patch by Victor Zverovich!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86636 91177308-0d34-0410-b5e6-96231b3b80d8
Simplify[IF]Cmp pieces. Add some predicates to CmpInst to
determine whether a predicate is fp or int.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86624 91177308-0d34-0410-b5e6-96231b3b80d8
simplifies instruction users of PHIs when the phi is eliminated. This
will be moved to transforms/utils after some other refactoring.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86603 91177308-0d34-0410-b5e6-96231b3b80d8
the loop. This is needed because with indirectbr it may not be possible
for LoopSimplify to guarantee that all loop exit predecessors are
inside the loop. This fixes PR5437.
LCCSA no longer actually requires LoopSimplify form, but for now it
must still have the dependency because the PassManager doesn't know
how to schedule LoopSimplify otherwise.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86569 91177308-0d34-0410-b5e6-96231b3b80d8
here:
1) We need to avoid processing sigma nodes as phi nodes for constraint generation.
2) We need to generate constraints for comparisons against constants properly.
This includes our first working ABCD test!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86498 91177308-0d34-0410-b5e6-96231b3b80d8
graphs being produced. The cause was that we were incorrectly marking sigma instructions as
processed after handling the sigma-specific constraints for them, potentially neglecting to
process them as normal instructions as well.
Unfortunately, the testcase that inspired this still doesn't work because of a bug in the solver,
which is next on the list to debug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86486 91177308-0d34-0410-b5e6-96231b3b80d8
when both the source and dest are illegal types, since it would cause
the phi to grow (for example, we shouldn't transform test14b's phi to
a phi on i320). This fixes an infinite loop on i686 bootstrap with
phi slicing turned on, so turn it back on.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86483 91177308-0d34-0410-b5e6-96231b3b80d8
not turn a PHI in a legal type into a PHI of an illegal type, and
add a new optimization that breaks up insane integer PHI nodes into
small pieces (PR3451).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86443 91177308-0d34-0410-b5e6-96231b3b80d8
(eliminating some extends) if the new type of the
computation is legal or if both the source and dest
are illegal. This prevents instcombine from changing big
chains of computation into i64 on 32-bit targets for
example.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86398 91177308-0d34-0410-b5e6-96231b3b80d8
(making pred factoring only happen if threading is guaranteed
to be successful).
This now survives an X86-64 bootstrap of llvm-gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86355 91177308-0d34-0410-b5e6-96231b3b80d8
Here is the original commit message:
This commit updates malloc optimizations to operate on malloc calls that have constant int size arguments.
Update CreateMalloc so that its callers specify the size to allocate:
MallocInst-autoupgrade users use non-TargetData-computed allocation sizes.
Optimization uses use TargetData to compute the allocation size.
Now that malloc calls can have constant sizes, update isArrayMallocHelper() to use TargetData to determine the size of the malloced type and the size of malloced arrays.
Extend getMallocType() to support malloc calls that have non-bitcast uses.
Update OptimizeGlobalAddressOfMalloc() to optimize malloc calls that have non-bitcast uses. The bitcast use of a malloc call has to be treated specially here because the uses of the bitcast need to be replaced and the bitcast needs to be erased (just like the malloc call) for OptimizeGlobalAddressOfMalloc() to work correctly.
Update PerformHeapAllocSRoA() to optimize malloc calls that have non-bitcast uses. The bitcast use of the malloc is not handled specially here because ReplaceUsesOfMallocWithGlobal replaces through the bitcast use.
Update OptimizeOnceStoredGlobal() to not care about the malloc calls' bitcast use.
Update all globalopt malloc tests to not rely on autoupgraded-MallocInsts, but instead use explicit malloc calls with correct allocation sizes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86311 91177308-0d34-0410-b5e6-96231b3b80d8
predicates. This allows us to jump thread things like:
_ZN12StringSwitchI5ColorE4CaseILj7EEERS1_RAT__KcRKS0_.exit119:
%tmp1.i24166 = phi i8 [ 1, %bb5.i117 ], [ %tmp1.i24165, %_Z....exit ], [ %tmp1.i24165, %bb4.i114 ]
%toBoolnot.i87 = icmp eq i8 %tmp1.i24166, 0 ; <i1> [#uses=1]
%tmp4.i90 = icmp eq i32 %tmp2.i, 6 ; <i1> [#uses=1]
%or.cond173 = and i1 %toBoolnot.i87, %tmp4.i90 ; <i1> [#uses=1]
br i1 %or.cond173, label %bb4.i96, label %_ZN12...
Where it is "obvious" that when coming from %bb5.i117 that the 'and' is always
false. This triggers a surprisingly high number of times in the testsuite,
and gets us closer to generating good code for doug's strswitch testcase.
This also make a bunch of other code in jump threading redundant, I'll rip
out in the next patch. This survived an enable-checking llvm-gcc bootstrap.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86264 91177308-0d34-0410-b5e6-96231b3b80d8
unsplittable critical edges, which means the introduction of
loops which cannot be transformed to LoopSimplify form. Fix
LoopSimplify to avoid transforming such loops into invalid
code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86176 91177308-0d34-0410-b5e6-96231b3b80d8
makes several optimization passes abort in cases where they're currently
silently miscompiling code.
Remove the indirectbr assertion from SplitEdge. Indirectbr is only
a problem for critical edges, and SplitEdge defers to SplitCriticalEdge
to handle those, and SplitCriticalEdge has its own assertion for
indirectbr.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86147 91177308-0d34-0410-b5e6-96231b3b80d8
MallocInst-autoupgrade users use non-TargetData-computed allocation sizes.
Optimization uses use TargetData to compute the allocation size.
Now that malloc calls can have constant sizes, update isArrayMallocHelper() to use TargetData to determine the size of the malloced type and the size of malloced arrays.
Extend getMallocType() to support malloc calls that have non-bitcast uses.
Update OptimizeGlobalAddressOfMalloc() to optimize malloc calls that have non-bitcast uses. The bitcast use of a malloc call has to be treated specially here because the uses of the bitcast need to be replaced and the bitcast needs to be erased (just like the malloc call) for OptimizeGlobalAddressOfMalloc() to work correctly.
Update PerformHeapAllocSRoA() to optimize malloc calls that have non-bitcast uses. The bitcast use of the malloc is not handled specially here because ReplaceUsesOfMallocWithGlobal replaces through the bitcast use.
Update OptimizeOnceStoredGlobal() to not care about the malloc calls' bitcast use.
Update all globalopt malloc tests to not rely on autoupgraded-MallocInsts, but instead use explicit malloc calls with correct allocation sizes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86077 91177308-0d34-0410-b5e6-96231b3b80d8
to EmitGEPOffset.
Implement some new transforms for optimizing
subtracts of two pointer to ints into the same vector. This happens
for C++ iterator idioms for example, stringmap takes a const char*
that points to the start and end of a string. Once inlined, we want
the pointer difference to turn back into a length.
This is rdar://7362831.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@86021 91177308-0d34-0410-b5e6-96231b3b80d8
more aggressive an correct. This survives building llvm in 64-bit
mode with optimizations and the built llvm passes make check.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85973 91177308-0d34-0410-b5e6-96231b3b80d8
functions that don't have local linkage. Basically, we need to be more
careful about propagating argument information to functions whose results
we aren't tracking. This fixes a miscompilation of
LLVMCConfigurationEmitter.cpp when built with an llvm-gcc that has ipsccp
enabled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85923 91177308-0d34-0410-b5e6-96231b3b80d8
function to calls of that function, regardless of whether it has local
linkage or has its address taken. Not escaping should only affect
whether we make an aggressive assumption about the arguments to a
function, not whether we can track the result of it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85795 91177308-0d34-0410-b5e6-96231b3b80d8
a DenseMap. Doing this required being aware of subtle iterator
invalidation issues, but it provides a big speedup. In a
release-asserts build, this sped up optimizing 403.gcc from
1.34s -> 0.79s (IPSCCP) and 1.11s -> 0.44s (SCCP).
This commit also conflates in a bunch of general cleanups, sorry.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85788 91177308-0d34-0410-b5e6-96231b3b80d8
not the max. This didn't matter until the previous patch because
instcombine would refuse to sink loads with differenting alignments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85738 91177308-0d34-0410-b5e6-96231b3b80d8
phis, it didn't preserve the alignment of the load. This is a missed
optimization of the alignment is high and a miscompilation when the
alignment is low.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85736 91177308-0d34-0410-b5e6-96231b3b80d8
in a way that should prevent ip constprop. This allows clang/test/CodeGen/indirect-goto.c
to pass with the new indirect goto lowering.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85709 91177308-0d34-0410-b5e6-96231b3b80d8
when BB2 has its address taken. Since it ends up doing BB2->rauw(BB1),
this can cause the address of the entry block to be taken. Since it is
generally undesirable to nuke blocks whose address is taken, even when
we can, just unconditionally stop this xform.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85708 91177308-0d34-0410-b5e6-96231b3b80d8
ended in an uncond branch because the pass requires BreakCriticalEdges.
However, BCE doesn't eliminate critical adges from indbrs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85707 91177308-0d34-0410-b5e6-96231b3b80d8
MergeBlockIntoPredecessor. This makes SimplifyCFG slightly more aggressive,
and makes it unnecessary for LoopUnroll to have its own copy of this code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85667 91177308-0d34-0410-b5e6-96231b3b80d8
PHI operands by the predecessor order, sort them by the order used by the
first PHI in the block. This is still suffucient to expose duplicates.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85634 91177308-0d34-0410-b5e6-96231b3b80d8
This reduces codesize on a variety of codes by 1-2% on x86-64. It also
helps clean up after SSAUpdater.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85626 91177308-0d34-0410-b5e6-96231b3b80d8
ArraySize * ElementSize
ElementSize * ArraySize
ArraySize << log2(ElementSize)
ElementSize << log2(ArraySize)
Refactor isArrayMallocHelper and delete isSafeToGetMallocArraySize, so that there is only 1 copy of the malloc array determining logic.
Update users of getMallocArraySize() to not bother calling isArrayMalloc() as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85421 91177308-0d34-0410-b5e6-96231b3b80d8
Checks on Demand algorithm which looks at arbitrary branches instead of loop
iterations. This is GSoC work by Andre Tavares with only editorial changes
applied!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85382 91177308-0d34-0410-b5e6-96231b3b80d8
In the new world order, BlockAddress can have a BasicBlock operand.
This doesn't permute much, because if you have a ConstantExpr (or
anything more specific than Constant) we still know the operand has
to be a Constant.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85375 91177308-0d34-0410-b5e6-96231b3b80d8
with multiple return values it inserts a PHI to merge them all together.
However, if the return values are all the same, it ends up with a pointless
PHI and this pointless PHI happens to really block SRoA from happening in
at least a silly C++ example written by Doug, but probably others. This
fixes rdar://7339069.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85206 91177308-0d34-0410-b5e6-96231b3b80d8
Remove LowerAllocations pass.
Update some more passes to treate free calls just like they were treating FreeInst.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85176 91177308-0d34-0410-b5e6-96231b3b80d8
the loop exiting block dominates the latch block; if ScalarEvolution
can prove that the trip-count is finite, that's sufficient.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85165 91177308-0d34-0410-b5e6-96231b3b80d8
GEPs (more than one non-zero index) into simple GEPs (at most one
non-zero index). In some simple experiments using this it's not
uncommon to see 3% overall code size wins, because it exposes
redundancies that can be eliminated, however it's tricky to use
because instcombine aggressively undoes the work that this pass does.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85144 91177308-0d34-0410-b5e6-96231b3b80d8
strides for now, because it doesn't handle them correctly. This fixes a
miscompile of SingleSource/Benchmarks/Misc-C++/ray.
This problem was usually hidden because indvars transforms such induction
variables into negations of canonical induction variables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85118 91177308-0d34-0410-b5e6-96231b3b80d8
Chris claims we should never have visibility_hidden inside any .cpp file but
that's still not true even after this commit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85042 91177308-0d34-0410-b5e6-96231b3b80d8
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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@85019 91177308-0d34-0410-b5e6-96231b3b80d8
Update all analysis passes and transforms to treat free calls just like FreeInst.
Remove RaiseAllocations and all its tests since FreeInst no longer needs to be raised.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84987 91177308-0d34-0410-b5e6-96231b3b80d8
exact backedge taken count, when checking for infinite loops. This allows
it to delete loops with multiple exit conditions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84952 91177308-0d34-0410-b5e6-96231b3b80d8
Analysis/ConstantFolding.cpp. This doesn't change the behavior of
instcombine but makes other clients of ConstantFoldInstruction
able to handle loads. This was partially extracted from Eli's patch
in PR3152.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84836 91177308-0d34-0410-b5e6-96231b3b80d8
Most changes are cleanup, but there is 1 correctness fix:
I fixed InstCombine so that the icmp is removed only if the malloc call is removed (which requires explicit removal because the Worklist won't DCE any calls since they can have side-effects).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84772 91177308-0d34-0410-b5e6-96231b3b80d8
in the PHI's Basic Block. This uses a conservative approach, because we don't
have dominator info in instcombine.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84754 91177308-0d34-0410-b5e6-96231b3b80d8
"In the existing code, if the load and the value to replace it with are
of different types *and* target data is available, it tries to use the
target data to coerce the replacement value to the type of the load.
Otherwise, it skips all effort to handle the type mismatch and just
feeds the wrongly-typed replacement value to replaceAllUsesWith, which
triggers an assertion.
The patch replaces it with an outer if checking for type mismatch, and
an inner if-else that checks whether target data is available and, if
not, returns false rather than trying to replace the load."
Patch by Kenneth Uildriks!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84739 91177308-0d34-0410-b5e6-96231b3b80d8
the estimated code size and the number of blocks when deciding whether to
do a non-trivial unswitch. This protects it from some very undesirable
worst-case behavior on large numbers of loop-unswitchable conditions, such
as in the testcase in PR5259.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84661 91177308-0d34-0410-b5e6-96231b3b80d8
When an incoming value for a PHI is updated, we must also updated all other
incoming values for the same BB to match, otherwise we create invalid PHIs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84638 91177308-0d34-0410-b5e6-96231b3b80d8
when the invoke had multiple return values: it set the lattice value only on the
extractvalue.
This caused the invoke's lattice value to remain the default (undefined), and
later propagated to extractvalue's operand, which incorrectly introduces
undefined behavior.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84637 91177308-0d34-0410-b5e6-96231b3b80d8
where a loop's header is being split and it has predecessors which are not
contained by the most-nested loop which contains the loop.
This fixes PR5235.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84505 91177308-0d34-0410-b5e6-96231b3b80d8
Update testcases that rely on malloc insts being present.
Also prematurely remove MallocInst handling from IndMemRemoval and RaiseAllocations to help pass tests in this incremental step.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84292 91177308-0d34-0410-b5e6-96231b3b80d8
identifying the malloc as a non-array malloc. This broke GlobalOpt's optimization of stores of mallocs
to global variables.
The fix is to classify malloc's into 3 categories:
1. non-array mallocs
2. array mallocs whose array size can be determined
3. mallocs that cannot be determined to be of type 1 or 2 and cannot be optimized
getMallocArraySize() returns NULL for category 3, and all users of this function must avoid their
malloc optimization if this function returns NULL.
Eventually, currently unexpected codegen for computing the malloc's size argument will be supported in
isArrayMalloc() and getMallocArraySize(), extending malloc optimizations to those examples.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84199 91177308-0d34-0410-b5e6-96231b3b80d8
don't bother every time going around the main worklist. This speeds up a
release-asserts opt -std-compile-opts on 403.gcc by about 4% (1.5s). It
seems to speed up the most expensive instances of instcombine by ~10%.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84171 91177308-0d34-0410-b5e6-96231b3b80d8
instruction (which disqualifies stores, unreachable, etc) and at least the
first operand is a constant. This filters out a lot of obvious cases that
can't be folded. Also, switch the IRBuilder to a TargetFolder, which tries
harder.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@84170 91177308-0d34-0410-b5e6-96231b3b80d8
BasicBlocks, so that it doesn't blindly procede in the presence of
large individual BasicBlocks. This addresses a class of code-size
expansion problems.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83992 91177308-0d34-0410-b5e6-96231b3b80d8
it to visit instructions from the start of the function to the
end of the function in the first path. This greatly speeds up
some pathological cases (e.g. PR5150).
Try #3, this time with some unneeded debug info stuff removed
which was causing dead pointers to be added to the worklist.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83818 91177308-0d34-0410-b5e6-96231b3b80d8
it to visit instructions from the start of the function to the
end of the function in the first path. This greatly speeds up
some pathological cases (e.g. PR5150).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83814 91177308-0d34-0410-b5e6-96231b3b80d8
into a shuffle even if it was used by another insertelement. If the
visitation order of instcombine was wrong, this would turn a chain of
insertelements into a chain of shufflevectors, which was quite painful.
Since CollectShuffleElements handles these cases, the code can just
be nuked.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83810 91177308-0d34-0410-b5e6-96231b3b80d8
input the the mul is a zext from bool, just that it is all zeros
other than the low bit. This fixes some phase ordering issues
that would cause us to miss some xforms in mul.ll when the worklist
is visited differently.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83794 91177308-0d34-0410-b5e6-96231b3b80d8
it to visit instructions from the start of the function to the
end of the function in the first path. This greatly speeds up
some pathological cases (e.g. PR5150).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83790 91177308-0d34-0410-b5e6-96231b3b80d8
For now the metadata of sinked/hoisted instructions is still wrong, but that'll
be fixed when instructions will have debug metadata directly attached.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83786 91177308-0d34-0410-b5e6-96231b3b80d8
done by condprop, but do it in a much more general form. The
basic idea is that we can do a limited form of tail duplication
in the case when we have a branch on a phi. Moving the branch
up in to the predecessor block makes instruction selection
much easier and encourages chained jump threadings.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83759 91177308-0d34-0410-b5e6-96231b3b80d8
inserted only once, just use vector. Don't compute ExitBlocks unless we
need it, change std::sort to array_pod_sort.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83747 91177308-0d34-0410-b5e6-96231b3b80d8
from GVN, this also speeds it up, inserts fewer PHI nodes (see the
testcase) and allows it to remove more loads (due to fewer PHI nodes
standing in the way).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83746 91177308-0d34-0410-b5e6-96231b3b80d8
not just at the end. Add a big comment explaining when this could
be useful (which never happens for jump threading).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83741 91177308-0d34-0410-b5e6-96231b3b80d8
DemoteRegToStack. This makes it more efficient (because it isn't
creating a ton of load/stores that are eventually removed by a later
mem2reg), and more slightly more effective (because those load/stores
don't get in the way of threading).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83706 91177308-0d34-0410-b5e6-96231b3b80d8
constants used in inlining heuristics (especially
those used in more than one file). No functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83675 91177308-0d34-0410-b5e6-96231b3b80d8
and that will make Caller too big to inline, see if it
might be better to inline Caller into its callers instead.
This situation is described in PR 2973, although I haven't
tried the specific case in SPASS.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83602 91177308-0d34-0410-b5e6-96231b3b80d8
to declare that they preserve other passes without needing to pull in
additional header file or library dependencies. Convert MachineFunctionPass
and CodeGenLICM to make use of this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83555 91177308-0d34-0410-b5e6-96231b3b80d8
already on the worklist, and print Visited when an instruction is about to be
visited. Net, on one input, this reduced the output size by at least 9x.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83510 91177308-0d34-0410-b5e6-96231b3b80d8
for inlining.
When MallocInst goes away this code will be subsumed as part of
calls and work just fine...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83434 91177308-0d34-0410-b5e6-96231b3b80d8
out of it, and jump threading, condprop and gvn are now getting
most of the benefit. This was approved by Nicholas and Nicolas.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83390 91177308-0d34-0410-b5e6-96231b3b80d8
the new predicates I added) instead of going through a context and doing a
pointer comparison. Besides being cheaper, this allows a smart compiler
to turn the if sequence into a switch.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@83297 91177308-0d34-0410-b5e6-96231b3b80d8
phi nodes. Make sure to phi translate from the right block.
This fixes a llvm-building-llvm failure on GVN-PRE.cpp
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82970 91177308-0d34-0410-b5e6-96231b3b80d8
simple constants for the true/false value of the select. We now
do phi translation etc. This really fixes PR4895 :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82917 91177308-0d34-0410-b5e6-96231b3b80d8
that are phi nodes. Also tighten up FoldOpIntoPhi to treat constantexpr
operands to phis just like other variables, avoiding moving constantexpr
computations around.
Patch by Daniel Dunbar.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82913 91177308-0d34-0410-b5e6-96231b3b80d8
This doesn't kick in too much because of phi translation issues,
but this can be resolved in the future.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82447 91177308-0d34-0410-b5e6-96231b3b80d8
from a piece of a large store when both are in the same block.
This allows clang to compile the testcase in PR4216 to this code:
_test_bitfield:
movl 4(%esp), %eax
movl %eax, %ecx
andl $-65536, %ecx
orl $32962, %eax
andl $40186, %eax
orl %ecx, %eax
ret
This is not ideal, but is a whole lot better than the code produced
by llvm-gcc:
_test_bitfield:
movw $-32574, %ax
orw 4(%esp), %ax
andw $-25350, %ax
movw %ax, 4(%esp)
movw 7(%esp), %cx
shlw $8, %cx
movzbl 6(%esp), %edx
orw %cx, %dx
movzwl %dx, %ecx
shll $16, %ecx
movzwl %ax, %eax
orl %ecx, %eax
ret
and dramatically better than that produced by gcc 4.2:
_test_bitfield:
pushl %ebx
call L3
"L00000000001$pb":
L3:
popl %ebx
movl 8(%esp), %eax
leal 0(,%eax,4), %edx
sarb $7, %dl
movl %eax, %ecx
andl $7168, %ecx
andl $-7201, %ebx
movzbl %dl, %edx
andl $1, %edx
sall $5, %edx
orl %ecx, %ebx
orl %edx, %ebx
andl $24, %eax
andl $-58336, %ebx
orl %eax, %ebx
orl $32962, %ebx
movl %ebx, %eax
popl %ebx
ret
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82439 91177308-0d34-0410-b5e6-96231b3b80d8
so that nonlocal and partially redundant loads can use it as well.
The testcase shows examples of craziness this can handle. This triggers
*many* times in 176.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82403 91177308-0d34-0410-b5e6-96231b3b80d8
(and load -> load) when the base pointers must alias but when
they are different types. This occurs very very frequently in
176.gcc and other code that uses bitfields a lot.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82399 91177308-0d34-0410-b5e6-96231b3b80d8
In getMallocArraySize(), fix bug in the case that array size is the product of 2 constants.
Extend isArrayMalloc() and getMallocArraySize() to handle case where malloc is used as char array.
Ensure that ArraySize in LowerAllocations::runOnBasicBlock() is correct type.
Extend Instruction::isSafeToSpeculativelyExecute() to handle malloc calls.
Add verification for malloc calls.
Reviewed by Dan Gohman.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82257 91177308-0d34-0410-b5e6-96231b3b80d8
constants out of loops. These aren't covered by the regular LICM
pass, because in LLVM IR constants don't require separate
instructions. They're not always covered by the MachineLICM pass
either, because it doesn't know how to unfold folded constant-pool
loads. This is somewhat experimental at this point, and off by
default.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82076 91177308-0d34-0410-b5e6-96231b3b80d8
phis, similar to the FoldPHIArgGEPIntoPHI change.
Also, delete some comments that don't reflect the code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@82053 91177308-0d34-0410-b5e6-96231b3b80d8
more than one phi, since that leads to higher register pressure on
entry to the phi. This is especially problematic when the phi is in
a loop header, as it increases register pressure throughout the loop.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81993 91177308-0d34-0410-b5e6-96231b3b80d8
argpromote to avoid invalidating an iterator. This fixes PR4977.
All clang tests now pass with expensive checking (on my system
at least).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81843 91177308-0d34-0410-b5e6-96231b3b80d8
within the notional bounds of the static type of the getelementptr (which
is not the same as "inbounds") from GlobalOpt into a utility routine,
and use it in ConstantFold.cpp to check whether there are any mis-behaved
indices.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81478 91177308-0d34-0410-b5e6-96231b3b80d8
loop exit edge -- new PHIs may be needed not only for the additional
splits that are made to preserve LoopSimplify form, but also for the
original split. Factor out the code that inserts new PHIs so that it
can be used for both. Remove LoopRotation.cpp's code for manually
updating LCSSA form, as it is now redundant. This fixes PR4934.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81363 91177308-0d34-0410-b5e6-96231b3b80d8
that get created during loop unswitching, and fix SplitBlockPredecessors'
LCSSA updating code to create new PHIs instead of trying to just move
existing ones.
Also, optimize Loop::verifyLoop, since it gets called a lot. Use
searches on a sorted list of blocks instead of calling the "contains"
function, as is done in other places in the Loop class, since "contains"
does a linear search. Also, don't call verifyLoop from LoopSimplify or
LCSSA, as the PassManager is already calling verifyLoop as part of
LoopInfo's verifyAnalysis.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81221 91177308-0d34-0410-b5e6-96231b3b80d8
extractelement operations into a bitcast of the pointer,
then a gep, then a scalar load. Disable this when the vector
only has one element, because it leads to infinite loops in
instcombine (PR4908).
This transformation seems like a really bad idea to me, as it
will likely disable CSE of vector load/stores etc and can be
better done in the code generator when profitable. This
goes all the way back to the first days of packed types,
r25299 specifically.
I'll let those people who care about the performance of vector
code decide what to do with this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81185 91177308-0d34-0410-b5e6-96231b3b80d8
compile-time constant integers or that are out of bounds for their
corresponding static array types. These can cause aliasing that
GlobalOpt assumes won't happen.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81165 91177308-0d34-0410-b5e6-96231b3b80d8
an aggregate store overlapping a different aggregate store, despite
the stores having distinct addresses.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81164 91177308-0d34-0410-b5e6-96231b3b80d8
is missing the inbounds flag. This is slightly conservative, but it
avoids problems with two constants pointing to the same address but
getting distinct entries in the Memory DenseMap.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81163 91177308-0d34-0410-b5e6-96231b3b80d8
- I think there are more instances of this, but I think they are fixed in Dan's
incoming patch. This one was preventing me from doing a bugpoint reduction
though.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81103 91177308-0d34-0410-b5e6-96231b3b80d8
Constant uniquing tables. This allows distinct ConstantExpr objects
with the same operation and different flags.
Even though a ConstantExpr "a + b" is either always overflowing or
never overflowing (due to being a ConstantExpr), it's still necessary
to be able to represent it both with and without overflow flags at
the same time within the IR, because the safety of the flag may
depend on the context of the use. If the constant really does overflow,
it wouldn't ever be safe to use with the flag set, however the use
may be in code that is never actually executed.
This also makes it possible to merge all the flags tests into a single test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80998 91177308-0d34-0410-b5e6-96231b3b80d8
that these passes are properly preserved.
Fix several transformation passes that claimed to preserve LoopSimplify
form but weren't.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80926 91177308-0d34-0410-b5e6-96231b3b80d8
instead of a bool argument, and to do the dominator check itself.
This makes it eaiser to use when DominatorTree information is
available.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80920 91177308-0d34-0410-b5e6-96231b3b80d8
Removed inverted flag form MaximumSpanningTree, also do not handle so much
information to MaximumSpanningTree.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80911 91177308-0d34-0410-b5e6-96231b3b80d8
simplifylibcalls optimization is thus valid for C++ but not C.
It's not important enough to worry about for C++ apps, so just
remove it.
rdar://7191924
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80887 91177308-0d34-0410-b5e6-96231b3b80d8
don't alias. Remove an old and poorly reduced testcase that fails
with this transform for reasons unrelated to the original test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80693 91177308-0d34-0410-b5e6-96231b3b80d8
Add statistics for regular edge profiling, this enables the comparation of the
number of edges inserted by regular and optimal edge profiling.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80668 91177308-0d34-0410-b5e6-96231b3b80d8
for sanity. This didn't turn up any bugs.
Change CallGraphNode to maintain its "callsite" information in the
call edges list as a WeakVH instead of as an instruction*. This fixes
a broad class of dangling pointer bugs, and makes CallGraph have a number
of useful invariants again. This fixes the class of problem indicated
by PR4029 and PR3601.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80663 91177308-0d34-0410-b5e6-96231b3b80d8
changes: SimplifyDemandedBits can't use the builder yet because it
has the wrong insertion point. This fixes a crash building
MultiSource/Benchmarks/PAQ8p
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80537 91177308-0d34-0410-b5e6-96231b3b80d8
instead of CallGraphNode*'s. This also papers over a callgraph
problem where a pass (in this case, MemCpyOpt) introduces a new
function into the module (llvm.memset.i64) but doesn't add it to
the call graph (nor should it, since it is a function pass).
While it might be a good idea for MemCpyOpt to not synthesize
functions in a runOnFunction(), there is no need for FunctionAttrs
to be boneheaded, so fix it there. This fixes an assertion building
176.gcc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80535 91177308-0d34-0410-b5e6-96231b3b80d8
indirect function pointer, inline it, then go to delete the body.
The problem is that the callgraph had other references to the function,
though the inliner had no way to know it, so we got a dangling pointer
and an invalid iterator out of the deal.
The fix to this is pretty simple: stop the inliner from deleting the
function by knowing that there are references to it. Do this by making
CallGraphNodes contain a refcount. This requires moving deletion of
available_externally functions to the module-level cleanup sweep where
it belongs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80533 91177308-0d34-0410-b5e6-96231b3b80d8
argpromotion and structretpromote. Basically, when replacing
a function, they used the 'changeFunction' api which changes
the entry in the function map (and steals/reuses the callgraph
node).
This has some interesting effects: first, the problem is that it doesn't
update the "callee" edges in any callees of the function in the call graph.
Second, this covers for a major problem in all the CGSCC pass stuff, which
is that it is completely broken when functions are deleted if they *don't*
reuse a CGN. (there is a cute little fixme about this though :).
This patch changes the protocol that CGSCC passes must obey: now the CGSCC
pass manager copies the SCC and preincrements its iterator to avoid passes
invalidating it. This allows CGSCC passes to mutate the current SCC. However
multiple passes may be run on that SCC, so if passes do this, they are now
required to *update* the SCC to be current when they return.
Other less interesting parts of this patch are that it makes passes update
the CG more directly, eliminates changeFunction, and requires clients of
replaceCallSite to specify the new callee CGN if they are changing it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80527 91177308-0d34-0410-b5e6-96231b3b80d8
is itself a bitcast. Since we have gep(bitcast(bitcast(y))) in this
case, just wait for the two bitcasts to get zapped. This prevents
instcombine from confusing some aliasing stuff, and allows it to
directly eliminate the load in the testcase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80508 91177308-0d34-0410-b5e6-96231b3b80d8
workslist and is set to insert new instructions before the current one.
Convert a bunch of stuff that used to call InsertNewInstBefore over to
use it, greatly simplifying code and making it more natural.
There is still a lot more to go, but this is a good start.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80492 91177308-0d34-0410-b5e6-96231b3b80d8
if the operand is not an instruction.
Simplify most uses of AddOperandsToWorkList to use AddValue and
inline it into the one remaining callsite.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80488 91177308-0d34-0410-b5e6-96231b3b80d8
former looks too much like AddUsersToWorkList and keeps
confusing me.
Remove AddSoonDeadInstToWorklist and change its two callers
to do the same thing in a simpler way.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80486 91177308-0d34-0410-b5e6-96231b3b80d8
into their callers. simplify ReplaceInstUsesWith. Make
EraseInstFromFunction only add operands to the worklist if
there aren't too many of them (this was a scalability win
for crazy programs that was only infrequently enforced).
Switch more code to using EraseInstFromFunction instead of
duplicating it inline. Change some fcmp/icmp optimizations
to modify fcmp/icmp in place instead of creating a new one
and deleting the old one just to change the predicate.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80483 91177308-0d34-0410-b5e6-96231b3b80d8
does constant folding of gep's: this is already handled in
a more general way.
No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80475 91177308-0d34-0410-b5e6-96231b3b80d8
This implements the maximum spanning tree algorithm on CFGs according to
weights given by the ProfileEstimator. This is then used to implement Optimal
Edge Profiling.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80358 91177308-0d34-0410-b5e6-96231b3b80d8
calls into a function and if the calls bring in arrays, try to merge
them together to reduce stack size. For example, in the testcase
we'd previously end up with 4 allocas, now we end up with 2 allocas.
As described in the comments, this is not really the ideal solution
to this problem, but it is surprisingly effective. For example, on
176.gcc, we end up eliminating 67 arrays at "gccas" time and another
24 at "llvm-ld" time.
One piece of concern that I didn't look into: at -O0 -g with
forced inlining this will almost certainly result in worse debug
info. I think this is acceptable though given that this is a case
of "debugging optimized code", and we don't want debug info to
prevent the optimizer from doing things anyway.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80215 91177308-0d34-0410-b5e6-96231b3b80d8
and introduce a new Instruction::isIdenticalTo which tests for full
identity, including the SubclassOptionalData flags. Also, fix the
Instruction::clone implementations to preserve the SubclassOptionalData
flags. Finally, teach several optimizations how to handle
SubclassOptionalData correctly, given these changes.
This fixes the counterintuitive behavior of isIdenticalTo not comparing
the full value, and clone not returning an identical clone, as well as
some subtle bugs that could be caused by these.
Thanks to Nick Lewycky for reporting this, and for an initial patch!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80038 91177308-0d34-0410-b5e6-96231b3b80d8
sinking code, since they are special. If the loop preheader happens
to be the entry block of a function, don't sink static allocas
out of it. This fixes PR4775.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@80010 91177308-0d34-0410-b5e6-96231b3b80d8
This change speeds up llvm-gcc by more then 6% at "-O0 -g" (measured by compiling InstructionCombining.cpp!)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79977 91177308-0d34-0410-b5e6-96231b3b80d8
the command line. This gives llvm-gcc developers
a way to control inlining (documented as "not intended
for end users").
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79966 91177308-0d34-0410-b5e6-96231b3b80d8
of an extracted block contains a PHI using a value defined in the extracted region.
With this patch, the partial inliner now passes MultiSource/Applications.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79963 91177308-0d34-0410-b5e6-96231b3b80d8
member out of line. ftostr is not particularly speedy,
so that method is presumably not perf sensitive.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79885 91177308-0d34-0410-b5e6-96231b3b80d8
by hand, and fix a few places that were using hasInitializer() that
appear to depend on the initializer value.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79441 91177308-0d34-0410-b5e6-96231b3b80d8
try to use i686-darwin to build for arm-eabi, you'll quickly run into
several false assumptions that the target OS must be the same as the
host OS. These patches split $(OS) into $(HOST_OS) and $(TARGET_OS) to
help builds like "make check" and the test-suite able to cross
compile. Along the way a target of *-unknown-eabi is defined as
"Freestanding" so that TARGET_OS checks have something to work with.
Patch by Sandeep Patel!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79296 91177308-0d34-0410-b5e6-96231b3b80d8
vector (&Formals[0]). With this change llvm-gcc builds
with expensive checking enabled for C, C++ and Fortran.
While there, change a std::vector into a SmallVector.
This is partly gratuitous, but mostly because not all
STL vector implementations define the data method (and
it should be faster).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79237 91177308-0d34-0410-b5e6-96231b3b80d8
unfoldable references to a PHI node in the block being folded, and disable
the transformation in that case. The correct transformation of such PHI
nodes depends on whether BB dominates Succ, and dominance is expensive
to compute here. (Alternatively, it's possible to check whether any
uses are live, but that's also essentially a dominance calculation.
Another alternative is to use reg2mem, but it probably isn't a good idea to
use that in simplifycfg.)
Also, remove some incorrect code from CanPropagatePredecessorsForPHIs
which is made unnecessary with this patch: it didn't consider the case
where a PHI node in BB has multiple uses.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@79174 91177308-0d34-0410-b5e6-96231b3b80d8
