type as the vector element type: allow them to be of
a wider integer type than the element type all the way
through the system, and not just as far as LegalizeDAG.
This should be safe because it used to be this way
(the old type legalizer would produce such nodes), so
backends should be able to handle it. In fact only
targets which have legal vector types with an illegal
promoted element type will ever see this (eg: <4 x i16>
on ppc). This fixes a regression with the new type
legalizer (vec_splat.ll). Also, treat SCALAR_TO_VECTOR
the same as BUILD_VECTOR. After all, it is just a
special case of BUILD_VECTOR.
llvm-svn: 69467
add dependencies on nodes with exactly one successor which is a
COPY_TO_REGCLASS node. In the case that the copy is coalesced
away, the dependence should be on the user of the copy, rather
than the copy itself.
llvm-svn: 69309
Insetad of doing ...
if (inlined_subroutine && known_location)
DW_TAG_inline_subroutine
else
DW_TAG_subprogram
do
if (inlined_subroutine) {
if (known_location)
DW_TAG_inline_subroutine
} else {
DW_TAG_subprogram
}
llvm-svn: 69300
to support replacing a node with another that has a superset of
the result types. Use this instead of calling
ReplaceAllUsesOfValueWith for each value.
llvm-svn: 69209
operator is used by a CopyToReg to export the value to a different
block, don't reuse the CopyToReg's register for the subreg operation
result if the register isn't precisely the right class for the
subreg operation.
Also, rename the h-registers.ll test, now that there are more
than one.
llvm-svn: 69087
promoted to legal types without changing the type of the vector. This is
following a suggestion from Duncan
(http://lists.cs.uiuc.edu/pipermail/llvmdev/2009-February/019923.html).
The transformation that used to be done during type legalization is now
postponed to DAG legalization. This allows the BUILD_VECTORs to be optimized
and potentially handled specially by target-specific code.
It turns out that this is also consistent with an optimization done by the
DAG combiner: a BUILD_VECTOR and INSERT_VECTOR_ELT may be combined by
replacing one of the BUILD_VECTOR operands with the newly inserted element;
but INSERT_VECTOR_ELT allows its scalar operand to be larger than the
element type, with any extra high bits being implicitly truncated. The
result is a BUILD_VECTOR where one of the operands has a type larger the
the vector element type.
Any code that operates on BUILD_VECTORs may now need to be aware of the
potential type discrepancy between the vector element type and the
BUILD_VECTOR operands. This patch updates all of the places that I could
find to handle that case.
llvm-svn: 68996
This will be used to replace things like X86's MOV32to32_.
Enhance ScheduleDAGSDNodesEmit to be more flexible and robust
in the presense of subregister superclasses and subclasses. It
can now cope with the definition of a virtual register being in
a subclass of a use.
Re-introduce the code for recording register superreg classes and
subreg classes. This is needed because when subreg extracts and
inserts get coalesced away, the virtual registers are left in
the correct subclass.
llvm-svn: 68961
Create debug_inlined dwarf section using these information. This info is used by gdb, at least on Darwin, to enable better experience debugging inlined functions. See DwarfWriter.cpp for more information on structure of debug_inlined section.
llvm-svn: 68847
in addition to ZERO_EXTEND and SIGN_EXTEND. Fix a bug in the
way it checked for live-out values, and simplify the way it
find users by using SDNode::use_iterator's (relatively) new
features. Also, make it slightly more permissive on targets
with free truncates.
In SelectionDAGBuild, avoid creating ANY_EXTEND nodes that are
larger than necessary. If the target's SwitchAmountTy has
enough bits, use it. This exposes the truncate to optimization
early, enabling more optimizations.
llvm-svn: 68670
eagerly. This helps avoid CopyToReg nodes in some cases where they
aren't needed, and also helps subsequent optimizer heuristics
in cases where the extra nodes would cause the node to appear
to have multiple results. This doesn't have a significant impact
currently; it'll help an upcoming change.
llvm-svn: 68667
with SUBREG_TO_REG, teach SimpleRegisterCoalescing to coalesce
SUBREG_TO_REG instructions (which are similar to INSERT_SUBREG
instructions), and teach the DAGCombiner to take advantage of this on
targets which support it. This eliminates many redundant
zero-extension operations on x86-64.
This adds a new TargetLowering hook, isZExtFree. It's similar to
isTruncateFree, except it only applies to actual definitions, and not
no-op truncates which may not zero the high bits.
Also, this adds a new optimization to SimplifyDemandedBits: transform
operations like x+y into (zext (add (trunc x), (trunc y))) on targets
where all the casts are no-ops. In contexts where the high part of the
add is explicitly masked off, this allows the mask operation to be
eliminated. Fix the DAGCombiner to avoid undoing these transformations
to eliminate casts on targets where the casts are no-ops.
Also, this adds a new two-address lowering heuristic. Since
two-address lowering runs before coalescing, it helps to be able to
look through copies when deciding whether commuting and/or
three-address conversion are profitable.
Also, fix a bug in LiveInterval::MergeInClobberRanges. It didn't handle
the case that a clobber range extended both before and beyond an
existing live range. In that case, multiple live ranges need to be
added. This was exposed by the new subreg coalescing code.
Remove 2008-05-06-SpillerBug.ll. It was bugpoint-reduced, and the
spiller behavior it was looking for no longer occurrs with the new
instruction selection.
llvm-svn: 68576
x * 40
=>
shlq $3, %rdi
leaq (%rdi,%rdi,4), %rax
This has the added benefit of allowing more multiply to be folded into addressing mode. e.g.
a * 24 + b
=>
leaq (%rdi,%rdi,2), %rax
leaq (%rsi,%rax,8), %rax
llvm-svn: 67917
stoppoint nodes around until Legalize; doing this
imposed an ordering on a sequence of loads that
came from different lines, interfering with scheduling.
llvm-svn: 67692
help out the register pressure reduction heuristics in the case of
nodes with multiple uses. Currently this uses very conservative
heuristics, so it doesn't have a broad impact, but in cases where it
does help it can make a big difference.
llvm-svn: 67586
a data dependency on the load node, so it really needs a
data-dependence edge to the load node, even if the load previously
existed.
And add a few comments.
llvm-svn: 67554
size by the array amount as an i32 value instead of promoting from
i32 to i64 then doing the multiply. Not doing this broke wrap-around
assumptions that the optimizers (validly) made. The ultimate real
fix for this is to introduce i64 version of alloca and remove mallocinst.
This fixes PR3829
llvm-svn: 67093
vector shuffle mask. Forced the mask to be built using i32. Note: this will
be irrelevant once vector_shuffle no longer takes a build vector for the
shuffle mask.
llvm-svn: 67076
if FPConstant is legal because if the FPConstant doesn't need to be stored
in a constant pool, the transformation is unlikely to be profitable.
llvm-svn: 66994
ptrtoint and inttoptr in X86FastISel. These casts aren't always
handled in the generic FastISel code because X86 sometimes needs
custom code to do truncation and zero-extension.
llvm-svn: 66988
by inserting explicit zero extensions where necessary. Included
is a testcase where SelectionDAG produces a virtual register
holding an i1 value which FastISel previously mistakenly assumed
to be zero-extended.
llvm-svn: 66941
1. ConstantPoolSDNode alignment field is log2 value of the alignment requirement. This is not consistent with other SDNode variants.
2. MachineConstantPool alignment field is also a log2 value.
3. However, some places are creating ConstantPoolSDNode with alignment value rather than log2 values. This creates entries with artificially large alignments, e.g. 256 for SSE vector values.
4. Constant pool entry offsets are computed when they are created. However, asm printer group them by sections. That means the offsets are no longer valid. However, asm printer uses them to determine size of padding between entries.
5. Asm printer uses expensive data structure multimap to track constant pool entries by sections.
6. Asm printer iterate over SmallPtrSet when it's emitting constant pool entries. This is non-deterministic.
Solutions:
1. ConstantPoolSDNode alignment field is changed to keep non-log2 value.
2. MachineConstantPool alignment field is also changed to keep non-log2 value.
3. Functions that create ConstantPool nodes are passing in non-log2 alignments.
4. MachineConstantPoolEntry no longer keeps an offset field. It's replaced with an alignment field. Offsets are not computed when constant pool entries are created. They are computed on the fly in asm printer and JIT.
5. Asm printer uses cheaper data structure to group constant pool entries.
6. Asm printer compute entry offsets after grouping is done.
7. Change JIT code to compute entry offsets on the fly.
llvm-svn: 66875
related transformations out of target-specific dag combine into the
ARM backend. These were added by Evan in r37685 with no testcases
and only seems to help ARM (e.g. test/CodeGen/ARM/select_xform.ll).
Add some simple X86-specific (for now) DAG combines that turn things
like cond ? 8 : 0 -> (zext(cond) << 3). This happens frequently
with the recently added cp constant select optimization, but is a
very general xform. For example, we now compile the second example
in const-select.ll to:
_test:
movsd LCPI2_0, %xmm0
ucomisd 8(%esp), %xmm0
seta %al
movzbl %al, %eax
movl 4(%esp), %ecx
movsbl (%ecx,%eax,4), %eax
ret
instead of:
_test:
movl 4(%esp), %eax
leal 4(%eax), %ecx
movsd LCPI2_0, %xmm0
ucomisd 8(%esp), %xmm0
cmovbe %eax, %ecx
movsbl (%ecx), %eax
ret
This passes multisource and dejagnu.
llvm-svn: 66779
alignment of the generated constant pool entry to the
desired alignment of a type. If we don't do this, we end up
trying to do movsd from 4-byte alignment memory. This fixes
450.soplex and 456.hmmer.
llvm-svn: 66641
and extern_weak_odr. These are the same as the non-odr versions,
except that they indicate that the global will only be overridden
by an *equivalent* global. In C, a function with weak linkage can
be overridden by a function which behaves completely differently.
This means that IP passes have to skip weak functions, since any
deductions made from the function definition might be wrong, since
the definition could be replaced by something completely different
at link time. This is not allowed in C++, thanks to the ODR
(One-Definition-Rule): if a function is replaced by another at
link-time, then the new function must be the same as the original
function. If a language knows that a function or other global can
only be overridden by an equivalent global, it can give it the
weak_odr linkage type, and the optimizers will understand that it
is alright to make deductions based on the function body. The
code generators on the other hand map weak and weak_odr linkage
to the same thing.
llvm-svn: 66339
with multiple chain operands. This can occur when the scheduler
has added chain operands to a node that already has a chain
operand, in order to handle physical register dependencies.
This fixes an llvm-gcc bootstrap failure on x86-64 introduced
in r66058.
llvm-svn: 66240
so it changed it into a 31 via the TLO.ShrinkDemandedConstant() call. Then it
would go through the DAG combiner again. This time it had a value of 31, which
was turned into a -1 by TLI.SimplifyDemandedBits(). This would ping pong
forever.
Teach the TLO.ShrinkDemandedConstant() call not to lower a value if the demanded
value is an XOR of all ones.
llvm-svn: 65985
arbitrary vector sizes. Add an optional MinSplatBits parameter to specify
a minimum for the splat element size. Update the PPC target to use the
revised interface.
llvm-svn: 65899
extracts + build_vector into a shuffle would fail, because the
type of the new build_vector would not be legal. Try harder to
create a legal build_vector type. Note: this will be totally
irrelevant once vector_shuffle no longer takes a build_vector for
shuffle mask.
New:
_foo:
xorps %xmm0, %xmm0
xorps %xmm1, %xmm1
subps %xmm1, %xmm1
mulps %xmm0, %xmm1
addps %xmm0, %xmm1
movaps %xmm1, 0
Old:
_foo:
xorps %xmm0, %xmm0
movss %xmm0, %xmm1
xorps %xmm2, %xmm2
unpcklps %xmm1, %xmm2
pshufd $80, %xmm1, %xmm1
unpcklps %xmm1, %xmm2
pslldq $16, %xmm2
pshufd $57, %xmm2, %xmm1
subps %xmm0, %xmm1
mulps %xmm0, %xmm1
addps %xmm0, %xmm1
movaps %xmm1, 0
llvm-svn: 65791
overly long ints, e.g. i96, into pieces at PHIs
and the nodes that feed into them; however big-endian
reverses the order of the pieces (for some reason), and
wasn't doing it the same way on both sides, so
the pieces didn't match and runtime failures ensued.
Fixes 188.ammp and sqlite3 on ppc32.
llvm-svn: 65481
them are generic changes.
- Use the "fast" flag that's already being passed into the asm printers instead
of shoving it into the DwarfWriter.
- Instead of calling "MI->getParent()->getParent()" for every MI, set the
machine function when calling "runOnMachineFunction" in the asm printers.
llvm-svn: 65379
a DBG_LABEL or not. We want to fall back to the original way of emitting debug
info when we're in -O0/-fast mode.
- Add plumbing in to pass the "Fast" flag to places that need it.
- XFAIL DebugInfo/deaddebuglabel.ll. This is finding 11 labels instead of 8. I
need to investigate still.
llvm-svn: 65367
instruction. The class also consolidates the code for detecting constant
splats that's shared across PowerPC and the CellSPU backends (and might be
useful for other backends.) Also introduces SelectionDAG::getBUID_VECTOR() for
generating new BUILD_VECTOR nodes.
llvm-svn: 65296
(Note: Eventually, commits like this will be handled via a pre-commit hook that
does this automagically, as well as expand tabs to spaces and look for 80-col
violations.)
llvm-svn: 64827
U include/llvm/CodeGen/DebugLoc.h
U lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
U lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp
U lib/Target/X86/AsmPrinter/X86ATTAsmPrinter.cpp
Enable debug location generation at -Os. This goes with the reapplication of the
r63639 patch.
llvm-svn: 64715
one bit set, because the bit may be shifted off the end. Instead,
just check for a constant 1 being shifted. This is still sufficient
to handle all the cases in test/CodeGen/X86/bt.ll. This fixes PR3583.
llvm-svn: 64622
Cleanup some warning.
Remark: when struct/class are declared differently than they are defined, this make problem for VC++ since it seems to mangle class differently that struct. These error are very hard to understand and find. So please, try to keep your definition/declaration in sync.
Only tested with VS2008. hope it does not break anything. feel free to revert.
llvm-svn: 64554
in inline asm as signed (what gcc does). Add partial support
for x86-specific "e" and "Z" constraints, with appropriate
signedness for printing.
llvm-svn: 64400
is determined by whether the node has a Flag operand. However, if the
node does have a Flag operand, it will be glued to its register's
def, so the heuristic would end up spuriously applying to whatever
node is the def.
llvm-svn: 64319
It was transforming (x&y)==y to (x&y)!=0 in the case where
y is variable and known to have at most one bit set (e.g. z&1).
This is not correct; the expressions are not equivalent when y==0.
I believe this patch salvages what can be salvaged, including
all the cases in bt.ll. Dan, please review.
Fixes gcc.c-torture/execute/20040709-[12].c
llvm-svn: 64314
instruction index across each part. Instruction indices are used
to make live range queries, and live ranges can extend beyond
scheduling region boundaries.
Refactor the ScheduleDAGSDNodes class some more so that it
doesn't have to worry about this additional information.
llvm-svn: 64288
scheduling, and generalize is so that preserves state across
scheduling regions. This fixes incorrect live-range information around
terminators and labels, which are effective region boundaries.
In place of looking for terminators to anchor inter-block dependencies,
introduce special entry and exit scheduling units for this purpose.
llvm-svn: 64254
Adjust derived classes to pass UnknownLoc where
a DebugLoc does not make sense. Pick one of
DebugLoc and non-DebugLoc variants to survive
for all such classes.
llvm-svn: 64000
Many targets build placeholder nodes for special operands, e.g.
GlobalBaseReg on X86 and PPC for the PIC base. There's no
sensible way to associate debug info with these. I've left
them built with getNode calls with explicit DebugLoc::getUnknownLoc operands.
I'm not too happy about this but don't see a good improvement;
I considered adding a getPseudoOperand or something, but it
seems to me that'll just make it harder to read.
llvm-svn: 63992
SelectionDAGISel::CreateScheduler, and make it just create the
scheduler. Leave running the scheduler to the higher-level code.
This makes the higher-level code a little more explicit and
easier to follow, and will help enable some future refactoring.
llvm-svn: 63944
target directories themselves. This also means that VMCore no longer
needs to know about every target's list of intrinsics. Future work
will include converting the PowerPC target to this interface as an
example implementation.
llvm-svn: 63765
but when legalizing the operation, we split the vector type and generate a library
call whose type needs to be promoted. For example, X86 with SSE on but MMX off,
a divide v2i64 will be scalarized to 2 calls to a library using i64.
llvm-svn: 63760
support GraphViz, I've been using the foo->dump() facility. This
patch is a minor rewrite to the SelectionDAG dump() stuff to make it a
little more helpful. The existing foo->dump() functionality does not
change; this patch adds foo->dumpr(). All of this is only useful when
running LLVM under a debugger.
llvm-svn: 63736
in any old order. Since analyzing a node analyzes its
operands also, this can mean that when we pop a node
off the list of nodes to be analyzed, it may already
have been analyzed.
llvm-svn: 63632
information. This eliminates the need for the Flags field in MemSDNode,
so this makes LoadSDNode and StoreSDNode smaller. Also, it makes
FoldingSetNodeIDs for loads and stores two AddIntegers smaller.
llvm-svn: 63577
crashes or wrong code with codegen of large integers:
eliminate the legacy getIntegerVTBitMask and
getIntegerVTSignBit methods, which returned their
value as a uint64_t, so couldn't handle huge types.
llvm-svn: 63494
returned by getShiftAmountTy may be too small
to hold shift values (it is an i8 on x86-32).
Before and during type legalization, use a large
but legal type for shift amounts: getPointerTy;
afterwards use getShiftAmountTy, fixing up any
shift amounts with a big type during operation
legalization. Thanks to Dan for writing the
original patch (which I shamelessly pillaged).
llvm-svn: 63482
- Modify TableGen to add the DebugLoc when calling getTargetNode.
(The light-weight wrappers are only temporary. The non-DebugLoc version will be
removed once the whole debug info stuff is finished with.)
llvm-svn: 63273
dagcombines that help it match in several more cases. Add
several more cases to test/CodeGen/X86/bt.ll. This doesn't
yet include matching for BT with an immediate operand, it
just covers more register+register cases.
llvm-svn: 63266
new isOperationLegalOrCustom, which does what isOperationLegal
previously did.
Update a bunch of callers to use isOperationLegalOrCustom
instead of isOperationLegal. In some case it wasn't obvious
which behavior is desired; when in doubt I changed then to
isOperationLegalOrCustom as that preserves their previous
behavior.
This is for the second half of PR3376.
llvm-svn: 63212
a uint64_t to verify that the value is in range for the given type,
to help catch accidental overflow. Fix a few places that relied on
getConstant implicitly truncating the value.
llvm-svn: 63128
checking logic. Rather than make the checking more
complicated, I've tweaked some logic to make things
conform to how the checking thought things ought to
be, since this results in a simpler "mental model".
llvm-svn: 63048
tidy up SDUse and related code.
- Replace the operator= member functions with a set method, like
LLVM Use has, and variants setInitial and setNode, which take
care up updating use lists, like LLVM Use's does. This simplifies
code that calls these functions.
- getSDValue() is renamed to get(), as in LLVM Use, though most
places can either use the implicit conversion to SDValue or the
convenience functions instead.
- Fix some more node vs. value terminology issues.
Also, eliminate the one remaining use of SDOperandPtr, and
SDOperandPtr itself.
llvm-svn: 62995
of each use in the SelectionDAG ReplaceAllUses* functions. Thanks
to Chris for spotting this opportunity.
Also, factor out code from all 5 of the ReplaceAllUses* functions
into AddNonLeafNodeToCSEMaps, which is now renamed
AddModifiedNodeToCSEMaps to more accurately reflect its purpose.
llvm-svn: 62964
testcase from PR3376, and in fact is sufficient to completely
avoid the problem in that testcase.
There's an underlying problem though; TLI.isOperationLegal
considers Custom to be Legal, which might be ok in some
cases, but that's what DAGCombiner is using in many places
to test if something is legal when LegalOperations is true.
When DAGCombiner is running after legalize, this isn't
sufficient. I'll address this in a separate commit.
llvm-svn: 62860
to "C ^ 1" is only valid when C is known to be either 0 or 1. Most of the
similar foldings in this function only handle "i1" types, but this one appears
intentionally written to handle larger integer types. If C has an integer
type larger than "i1", this needs to check if the high bits of a boolean
are known to be zero. I also changed the comment to describe this folding as
"C ^ 1" instead of "~C", since that is what the code does and since the latter
would only be valid for "i1" types. The good news is that most LLVM targets
use TargetLowering::ZeroOrOneBooleanContent so this change will not disable
the optimization; the bad news is that I've been unable to come up with a
testcase to demonstrate the problem.
I have also removed a "FIXME" comment for folding "select C, X, 0" to "C & X",
since the code looks correct to me. It could be made more aggressive by not
limiting the type to "i1", but that would then require checking for
TargetLowering::ZeroOrNegativeOneBooleanContent. Similar changes could be
done for the other SELECT foldings, but it was decided to be not worth the
trouble and complexity (see e.g., r44663).
llvm-svn: 62790
Simplify x+0 to x in unsafe-fp-math mode. This avoids a bunch of
redundant work in many cases, because in unsafe-fp-math mode,
ISD::FADD with a constant is considered free to negate, so the
DAGCombiner often negates x+0 to -0-x thinking it's free, when
in reality the end result is -x, which is more expensive than x.
Also, combine x*0 to 0.
This fixes PR3374.
llvm-svn: 62789
special cases after producing the new reduced-width load, because the
new load already has the needed adjustments built into it. This fixes
several bugs due to the special cases, including PR3317.
llvm-svn: 62692
- Ensure that (operation) legalization emits proper FDIV libcall when needed.
- Fix various bugs encountered during llvm-spu-gcc build, along with various
cleanups.
- Start supporting double precision comparisons for remaining libgcc2 build.
Discovered interesting DAGCombiner feature, which is currently solved via
custom lowering (64-bit constants are not legal on CellSPU, but DAGCombiner
insists on inserting one anyway.)
- Update README.
llvm-svn: 62664
SDNode subclasses to keep state that requires non-trivial
destructors, however it was already effectively impossible,
since the destructor isn't actually ever called. There currently
aren't any SDNode subclasses affected by this, and in general
it's desireable to keep SDNode objects light-weight.
This eliminates the last virtual member function in the SDNode
class, so it eliminates the need for a vtable pointer, making
SDNode smaller.
llvm-svn: 62539
uses are added to the From node while it is processing From's
use list, because of automatic local CSE. The fix is to avoid
visiting any new uses.
Fix a few places in the DAGCombiner that assumed that after
a RAUW call, the From node has no users and may be deleted.
This fixes PR3018.
llvm-svn: 62533
and every other instruction in their blocks to keep the terminator
instructions at the end, teach the post-RA scheduler how to operate
on ranges of instructions, and exclude terminators from the range
of instructions that get scheduled.
Also, exclude mid-block labels, such as EH_LABEL instructions, and
schedule code before them separately from code after them. This
fixes problems with the post-RA scheduler moving code past
EH_LABELs.
llvm-svn: 62366
a new toy hazard recognizier heuristic which attempts to direct the
scheduler to avoid clumping large groups of loads or stores too densely.
llvm-svn: 62291
and into the ScheduleDAGInstrs class, so that they don't get
destructed and re-constructed for each block. This fixes a
compile-time hot spot in the post-pass scheduler.
To help facilitate this, tidy and do some minor reorganization
in the scheduler constructor functions.
llvm-svn: 62275
scheduling dependencies. Add assertion checks to help catch
this.
It appears the Mips target defaults to list-td, and it has a
regression test that uses a physreg dependence. Such code was
liable to be miscompiled, and now evokes an assertion failure.
llvm-svn: 62177
aggregate types. Don't increment the current index after reaching
the end of a struct, as it will already be pointing at
one-past-the end. This fixes PR3288.
llvm-svn: 61828
AddPseudoTwoAddrDeps. This lets the scheduling infrastructure
avoid recalculating node heights. In very large testcases this
was a major bottleneck. Thanks to Roman Levenstein for finding
this!
As a side effect, fold-pcmpeqd-0.ll is now scheduled better
and it no longer requires spilling on x86-32.
llvm-svn: 61778
instructions to avoid copies, because TwoAddressInstructionPass
also does this optimization. The scheduler's version didn't
account for live-out values, which resulted in spurious commutes
and missed opportunities.
Now, TwoAddressInstructionPass handles all the opportunities,
instead of just those that the scheduler missed. The result is
usually the same, though there are occasional trivial differences
resulting from the avoidance of spurious commutes.
llvm-svn: 61611
promote from i1 all the way up to the canonical SetCC type.
In order to discover an appropriate type to use, pass
MVT::Other to getSetCCResultType. In order to be able to
do this, change getSetCCResultType to take a type as an
argument, not a value (this is also more logical).
llvm-svn: 61542
This removes all the _8, _16, _32, and _64 opcodes and replaces each
group with an unsuffixed opcode. The MemoryVT field of the AtomicSDNode
is now used to carry the size information. In tablegen, the size-specific
opcodes are replaced by size-independent opcodes that utilize the
ability to compose them with predicates.
This shrinks the per-opcode tables and makes the code that handles
atomics much more concise.
llvm-svn: 61389
temporary workaround for an obscure bug. When node cloning is
used, it is possible that more SUnits will be created, and
if the SUnits std::vector has to reallocate, it will
invalidate all the graph edges.
llvm-svn: 61122
computation code. Also, avoid adding output-depenency edges when both
defs are dead, which frequently happens with EFLAGS defs.
Compute Depth and Height lazily, and always in terms of edge latency
values. For the schedulers that don't care about latency, edge latencies
are set to 1.
Eliminate Cycle and CycleBound, and LatencyPriorityQueue's Latencies array.
These are all subsumed by the Depth and Height fields.
llvm-svn: 61073
width register load followed by a truncating
store for the copy, since the load will not place
the value in the lower bits. Probably partial
loads/stores can never happen here, but fix it
anyway.
llvm-svn: 60972
use of illegal integer types: instead, use a stack slot
and copying via integer registers. The existing code
is still used if the bitconvert is to a legal integer
type.
This fires on the PPC testcases 2007-09-08-unaligned.ll
and vec_misaligned.ll. It looks like equivalent code
is generated with these changes, just permuted, but
it's hard to tell.
With these changes, nothing in LegalizeDAG produces
illegal integer types anymore. This is a prerequisite
for removing the LegalizeDAG type legalization code.
While there I noticed that the existing code doesn't
handle trunc store of f64 to f32: it turns this into
an i64 store, which represents a 4 byte stack smash.
I added a FIXME about this. Hopefully someone more
motivated than I am will take care of it.
llvm-svn: 60964
ISD::ADD to emit an implicit EFLAGS. This was horribly broken. Instead, replace
the intrinsic with an ISD::SADDO node. Then custom lower that into an
X86ISD::ADD node with a associated SETCC that checks the correct condition code
(overflow or carry). Then that gets lowered into the correct X86::ADDOvf
instruction.
Similar for SUB and MUL instructions.
llvm-svn: 60915
for promoted integer types, eg: i16 on ppc-32, or
i24 on any platform. Complete support for arbitrary
precision integers would require handling expanded
integer types, eg: i128, but I couldn't be bothered.
llvm-svn: 60834
The Cost field is removed. It was only being used in a very limited way,
to indicate when the scheduler should attempt to protect a live register,
and it isn't really needed to do that. If we ever want the scheduler to
start inserting copies in non-prohibitive situations, we'll have to
rethink some things anyway.
A Latency field is added. Instead of giving each node a single
fixed latency, each edge can have its own latency. This will eventually
be used to model various micro-architecture properties more accurately.
The PointerIntPair class and an internal union are now used, which
reduce the overall size.
llvm-svn: 60806
essential problem was that the DAG can contain
random unused nodes which were never analyzed.
When remapping a value of a node being processed,
such a node may become used and need to be analyzed;
however due to operands being transformed during
analysis the node may morph into a different one.
Users of the morphing node need to be updated, and
this wasn't happening. While there I added a bunch
of documentation and sanity checks, so I (or some
other poor soul) won't have to scratch their head
over this stuff so long trying to remember how it
was all supposed to work next time some obscure
problem pops up! The extra sanity checking exposed
a few places where invariants weren't being preserved,
so those are fixed too. Since some of the sanity
checking is expensive, I added a flag to turn it
on. It is also turned on when building with
ENABLE_EXPENSIVE_CHECKS=1.
llvm-svn: 60797
Fix the shift amount when unrolling a vector shift into scalar shifts.
Fix problem in getShuffleScalarElt where it assumes that the input of
a bit convert must be a vector.
llvm-svn: 60740
and use it in x86 address mode folding. Also, make
getRegForValue return 0 for illegal types even if it has a
ValueMap for them, because Argument values are put in the
ValueMap. This fixes PR3181.
llvm-svn: 60696
1. ppcf128 select is expanded to f64 select's.
2. f64 select operand 0 is an i1 truncate, it's promoted to i32 zero_extend.
3. f64 select is updated. It's changed back to a "NewNode" and being re-analyzed.
4. f64 select operands are being processed. Operand 0 is a "NewNode". It's being expunged out of ReplacedValues map.
5. ExpungeNode tries to remap f64 select and notice it's a "NewNode" and assert.
Duncan, please take a look. Thanks.
llvm-svn: 60443
ReplaceNodeResults: rather than returning a node which
must have the same number of results as the original
node (which means mucking around with MERGE_VALUES,
and which is also easy to get wrong since SelectionDAG
folding may mean you don't get the node you expect),
return the results in a vector.
llvm-svn: 60348
multiplies.
Some more cleverness would be nice, though. It would be nice if we
could do this transformation on illegal types. Also, we would
prefer a narrower constant when possible so that we can use a narrower
multiply, which can be cheaper.
llvm-svn: 60283
introduce any new spilling; it just uses unused registers.
Refactor the SUnit topological sort code out of the RRList scheduler and
make use of it to help with the post-pass scheduler.
llvm-svn: 59999
(this doesn't happen that often, since most code
does not use illegal types) then follow it by a
DAG combiner run that is allowed to generate
illegal operations but not illegal types. I didn't
modify the target combiner code to distinguish like
this between illegal operations and illegal types,
so it will not produce illegal operations as well
as not producing illegal types.
llvm-svn: 59960
"It simplifies the type legalization part a bit, and produces better code by
teaching SelectionDAG about the extra bits in an i8 SADDO/UADDO node. In
essence, I spontaneously decided that on x86 this i8 boolean result would be
either 0 or 1, and on other platforms 0/1 or 0/-1, depending on whether the
platform likes it's boolean zero extended or sign extended."
llvm-svn: 59864
g++ -m32 -c -g -DIN_GCC -W -Wall -Wwrite-strings -Wmissing-format-attribute -fno-common -mdynamic-no-pic -DHAVE_CONFIG_H -Wno-unused -DTARGET_NAME=\"i386-apple-darwin9.5.0\" -I. -I. -I../../llvm-gcc.src/gcc -I../../llvm-gcc.src/gcc/. -I../../llvm-gcc.src/gcc/../include -I./../intl -I../../llvm-gcc.src/gcc/../libcpp/include -I../../llvm-gcc.src/gcc/../libdecnumber -I../libdecnumber -I/Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.obj/include -I/Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.src/include -DENABLE_LLVM -I/Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.obj/../llvm.src/include -D_DEBUG -D_GNU_SOURCE -D__STDC_LIMIT_MACROS -D__STDC_CONSTANT_MACROS -I. -I. -I../../llvm-gcc.src/gcc -I../../llvm-gcc.src/gcc/. -I../../llvm-gcc.src/gcc/../include -I./../intl -I../../llvm-gcc.src/gcc/../libcpp/include -I../../llvm-gcc.src/gcc/../libdecnumber -I../libdecnumber -I/Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.obj/include -I/Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.src/include ../../llvm-gcc.src/gcc/llvm-types.cpp -o llvm-types.o
../../llvm-gcc.src/gcc/llvm-convert.cpp: In member function 'void TreeToLLVM::EmitMemCpy(llvm::Value*, llvm::Value*, llvm::Value*, unsigned int)':
../../llvm-gcc.src/gcc/llvm-convert.cpp:1496: error: 'memcpy_i32' is not a member of 'llvm::Intrinsic'
../../llvm-gcc.src/gcc/llvm-convert.cpp:1496: error: 'memcpy_i64' is not a member of 'llvm::Intrinsic'
../../llvm-gcc.src/gcc/llvm-convert.cpp: In member function 'void TreeToLLVM::EmitMemMove(llvm::Value*, llvm::Value*, llvm::Value*, unsigned int)':
../../llvm-gcc.src/gcc/llvm-convert.cpp:1512: error: 'memmove_i32' is not a member of 'llvm::Intrinsic'
../../llvm-gcc.src/gcc/llvm-convert.cpp:1512: error: 'memmove_i64' is not a member of 'llvm::Intrinsic'
../../llvm-gcc.src/gcc/llvm-convert.cpp: In member function 'void TreeToLLVM::EmitMemSet(llvm::Value*, llvm::Value*, llvm::Value*, unsigned int)':
../../llvm-gcc.src/gcc/llvm-convert.cpp:1528: error: 'memset_i32' is not a member of 'llvm::Intrinsic'
../../llvm-gcc.src/gcc/llvm-convert.cpp:1528: error: 'memset_i64' is not a member of 'llvm::Intrinsic'
make[3]: *** [llvm-convert.o] Error 1
make[3]: *** Waiting for unfinished jobs....
rm fsf-funding.pod gcov.pod gfdl.pod cpp.pod gpl.pod gcc.pod
make[2]: *** [all-stage1-gcc] Error 2
make[1]: *** [stage1-bubble] Error 2
make: *** [all] Error 2
llvm-svn: 59809
"ISD::ADDO". ISD::ADDO is lowered into a target-independent form that does the
addition and then checks if the result is less than one of the operands. (If it
is, then there was an overflow.)
llvm-svn: 59779
schedulers. This doesn't have much immediate impact because
targets that use these schedulers by default don't yet provide
pipeline information.
This code also didn't have the benefit of register pressure
information. Also, removing it will avoid problems with list-burr
suddenly starting to do latency-oriented scheduling on x86 when we
start providing pipeline data, which would increase spilling.
llvm-svn: 59775
some of the latency computation logic out of the SDNode
ScheduleDAG code into a TargetInstrItineraries helper method
to help with this.
llvm-svn: 59761
the RR scheduler actually does look at latency values, but it
doesn't use a hazard recognizer so it has no way to know when
a no-op is needed, as opposed to just stalling and incrementing
the cycle count.
llvm-svn: 59759
dedicated "fast" scheduler in -fast mode instead, which is
faster. This speeds up llc -fast by a few percent on some
testcases -- the speedup only happens for code not handled by
fast-isel.
llvm-svn: 59700
is currently off by default, and can be enabled with
-disable-post-RA-scheduler=false.
This doesn't have a significant impact on most code yet because it doesn't
yet do anything to address anti-dependencies and it doesn't attempt to
disambiguate memory references. Also, several popular targets
don't have pipeline descriptions yet.
The majority of the changes here are splitting the SelectionDAG-specific
code out of ScheduleDAG, so that ScheduleDAG can be moved to
libLLVMCodeGen.a. The interface between ScheduleDAG-using code and
the rest of the scheduling code is somewhat rough and will evolve.
llvm-svn: 59676
and FP_ROUND. Not sure what these were doing
here - probably they were sometimes (wrongly)
created with integer operands somewhere that
has since been fixed.
llvm-svn: 59548
SCALAR_TO_VECTOR. I didn't add the testcase, because
once llc gets past scalar-to-vector it hits a SPU target
lowering bug and explodes.
llvm-svn: 59530
new CycleBound value. Instead, just update CycleBound on each call.
Also, make ReleasePred and ReleaseSucc methods more consistent accross
the various schedulers.
This also happens to make ScheduleDAGRRList's CycleBound computation
somewhat more interesting, though it still doesn't have any noticeable
effect, because no current targets that use the register-pressure
reduction scheduler provide pipeline models.
llvm-svn: 59475
to carry a SmallVector of flagged nodes, just calculate the flagged nodes
dynamically when they are needed.
The local-liveness change is due to a trivial scheduling change where
the scheduler arbitrary decision differently.
llvm-svn: 59273
special-purpose hook to a new pass. Also, add check to see if any
x87 virtual registers are used, to avoid doing any work in the
common case that no x87 code is needed.
llvm-svn: 59190
that it no longer handles non-power-of-two vectors.
However it previously only handled them sometimes,
depending on obscure numerical relationships between
the index and vector type. For example, for a vector
of length 6, it would succeed if and only if the
index was an even multiple of 6. I consider this
more confusing than useful.
llvm-svn: 59122
before creating the SUnit for the operation that it was unfolded from. This
allows each SUnit to have all of its predecessor SUnits available at the time
it is created. I don't know yet if this will be absolutely required, but it
is a little tidier to do it this way.
llvm-svn: 59083
The CC was changed, but wasn't checked to see if it was legal if the DAG
combiner was being run after legalization. Threw in a couple of checks just to
make sure that it's okay. As far as the PR is concerned, no back-end target
actually exhibited this problem, so there isn't an associated testcase.
llvm-svn: 59035
support targets that support these conversions. Users should avoid using
this node as the current targets don't generating code for it.
llvm-svn: 59001
inform the optimizers that the result must be zero/
sign extended from the smaller type. For example,
if a fp to unsigned i16 is promoted to fp to i32,
then we are allowed to assume that the extra 16 bits
are zero (because the result of fp to i16 is undefined
if the result does not fit in an i16). This is
quite aggressive, but should help the optimizers
produce better code. This requires correcting a
test which thought that fp_to_uint is some kind
of truncation, which it is not: in the testcase
(which does fp to i1), either the fp value converts
to 0 or 1 or the result is undefined, which is
quite different to truncation.
llvm-svn: 58991
This is Chris' patch from the PR, modified to realize that
SETUGT/SETULT occur legitimately with integers, plus
two fixes in LegalizeDAG to pass a valid result type into
LegalizeSetCC. The argument of TLI.getSetCCResultType is
ignored on PPC, but I think I'm following usage elsewhere.
llvm-svn: 58871
the condition for a BRCOND, according to what is
returned by getSetCCResultContents. Since all
targets return the same thing (ZeroOrOneSetCCResult),
this should be harmless! The point is that all over
the place the result of SETCC is fed directly into
BRCOND. On machines for which getSetCCResultContents
returns ZeroOrNegativeOneSetCCResult, this is a
sign-extended boolean. So it seems dangerous to
also feed BRCOND zero-extended booleans in some
circumstances - for example, when promoting the
condition.
llvm-svn: 58861
(e.g. a bitfield test) narrow the load as much as possible.
The has the potential to avoid unnecessary partial-word
load-after-store conflicts, which cause stalls on several targets.
Also a size win on x86 (testb vs testl).
llvm-svn: 58825
LLVM IR code and not in the selection DAG ISel. This is a cleaner solution.
- Fix the heuristic for determining if protectors are necessary. The previous
one wasn't checking the proper type size.
llvm-svn: 58824
- stackprotector_prologue creates a stack object and stores the guard there.
- stackprotector_epilogue reads the stack guard from the stack position created
by stackprotector_prologue.
- The PrologEpilogInserter was changed to make sure that the stack guard is
first on the stack frame.
llvm-svn: 58791
sized integers like i129, and also reduce the number
of assumptions made about how vaarg is implemented.
This still doesn't work correctly for small integers
like (eg) i1 on x86, since x86 passes each of them
(essentially an i8) in a 4 byte stack slot, so the
pointer needs to be advanced by 4 bytes not by 1 byte
as now. But this is no longer a LegalizeTypes problem
(it was also wrong in LT before): it is a bug in the
operation expansion in LegalizeDAG: now LegalizeTypes
turns an i1 vaarg into an i8 vaarg which would work
fine if only the i8 vaarg was turned into correct code
later.
llvm-svn: 58635
type for the shift amount type. Add a check
that shifts and rotates use the type returned
by getShiftAmountTy for the amount. This
exposed some problems in CellSPU and PPC,
which have already been fixed.
llvm-svn: 58455
other day that PPC custom lowering could create
a BUILD_PAIR of two f64 with a result type of...
f64! - already fixed). Fix a place that triggers
the sanity check.
llvm-svn: 58378
is morphed by AnalyzeNewNode into a previously
processed node, and different result values of
that node are remapped to values with different
nodes, then we could end up using wrong values
here [we were assuming that all results remap
to values with the same underlying node]. This
seems theoretically possible, but I don't have
a testcase. The meat of the patch is in the
changes to AnalyzeNewNode/AnalyzeNewValue and
ReplaceNodeWith. While there, I changed names
like RemapNode to RemapValue, since it really
remaps values. To tell the truth, I would be
much happier if we were only remapping nodes
(it would simplify a bunch of logic, and allow
for some cute speedups) but I haven't yet worked
out how to do that.
llvm-svn: 58372
ppcf128 to i32 conversion and expand it into a code
sequence like in LegalizeDAG. This needs custom
ppc lowering of FP_ROUND_INREG, so turn that on and
make it work with LegalizeTypes. Probably PPC should
simply custom lower the original conversion.
llvm-svn: 58329
(and a bunch of other node types). While there, I
added a doNotCSE predicate and used it to reduce code
duplication (some of the duplicated code was wrong...).
This fixes ARM/cse-libcalls.ll when using LegalizeTypes.
llvm-svn: 58249
worklist twice: UpdateNodeOperands could morph
a new node into a node already on the worklist.
We would then recalculate the NodeId for this
existing node and add it to the worklist. The
testcase is ARM/cse-libcalls.ll, the problem
showing up once UpdateNodeOperands is taught to
do CSE for calls.
llvm-svn: 58246
may return i8, which can result in SELECT nodes for
which the type of the condition is i8, but there are
no patterns for select with i8 condition. Tweak the
LegalizeTypes logic to avoid this as much as possible.
This isn't a real fix because it is still perfectly
possible to end up with such select nodes - CellSPU
needs to be fixed IMHO.
llvm-svn: 57968
that is not of type MVT::i1 in SELECT and SETCC nodes.
Relax the LegalizeTypes SELECT condition promotion
sanity checks to allow other condition types than i1.
llvm-svn: 57966
to have a different type to the vector element
type. This should be fairly harmless because in
the past guys like this were being built all over
the place (and were cleaned up when I added this
check). The reason for relaxing this check is
that it helps LegalizeTypes legalize vector
shuffles: the mask is a BUILD_VECTOR that it is
*not always possible* to legalize while keeping it
a BUILD_VECTOR (vector_shuffle requires the mask
to be a BUILD_VECTOR, as opposed to a vector with
the right vector type). With this check it is even
harder to legalize the mask - turning the check off
means that LegalizeTypes manages to legalize almost
all vector shuffles encountered in practice. The
correct solution is to change vector_shuffle to be a
variadic node with the mask built into it as operands.
While waiting for that change, this hack stops the
problem with vector_shuffle from blocking the turning
on of LegalizeTypes.
llvm-svn: 57965
The same one Apple gcc uses, faster. Also gets the
extreme case in gcc.c-torture/execute/ieee/rbug.c
correct which we weren't before; this is not
sufficient to get the test to pass though, there
is another bug.
llvm-svn: 57926
in the 32-bit signed offset field of addresses. Even though this
may be intended, some linkers refuse to relocate code where the
relocated address computation overflows.
Also, fix the sign-extension of constant offsets to use the
actual pointer size, rather than the size of the GlobalAddress
node, which may be different, for example on x86-64 where MVT::i32
is used when the address is being fit into the 32-bit displacement
field.
llvm-svn: 57885
for strange asm conditions earlier. In this case, we have a
double being passed in an integer reg class. Convert to like
sized integer register so that we allocate the right number
for the class (two i32's for the f64 in this case).
llvm-svn: 57862
result type when the result type is legal but
not the operand type. Add additional support
for EXTRACT_SUBVECTOR and CONCAT_VECTORS,
needed to handle such cases.
llvm-svn: 57840
elements. Otherwise LegalizeTypes will, reasonably
enough, legalize the mask, which may result in it
no longer being a BUILD_VECTOR node (LegalizeDAG
simply ignores the legality or not of vector masks).
llvm-svn: 57782
the previous patch this one actually passes make check.
"Fix PR2356 on PowerPC: if we have an input and output that are tied together
that have different sizes (e.g. i32 and i64) make sure to reserve registers for
the bigger operand."
llvm-svn: 57771
and add a TargetLowering hook for it to use to determine when this
is legal (i.e. not in PIC mode, etc.)
This allows instruction selection to emit folded constant offsets
in more cases, such as the included testcase, eliminating the need
for explicit arithmetic instructions.
This eliminates the need for the C++ code in X86ISelDAGToDAG.cpp
that attempted to achieve the same effect, but wasn't as effective.
Also, fix handling of offsets in GlobalAddressSDNodes in several
places, including changing GlobalAddressSDNode's offset from
int to int64_t.
The Mips, Alpha, Sparc, and CellSPU targets appear to be
unaware of GlobalAddress offsets currently, so set the hook to
false on those targets.
llvm-svn: 57748
shift counts, and patterns that match dynamic shift counts
when the subtract is obscured by a truncate node.
Add DAGCombiner support for recognizing rotate patterns
when the shift counts are defined by truncate nodes.
Fix and simplify the code for commuting shld and shrd
instructions to work even when the given instruction doesn't
have a parent, and when the caller needs a new instruction.
These changes allow LLVM to use the shld, shrd, rol, and ror
instructions on x86 to replace equivalent code using two
shifts and an or in many more cases.
llvm-svn: 57662
i.e. conditions that cannot be checked with a single instruction. For example,
SETONE and SETUEQ on x86.
- Teach legalizer to implement *illegal* setcc as a and / or of a number of
legal setcc nodes. For now, only implement FP conditions. e.g. SETONE is
implemented as SETO & SETNE, SETUEQ is SETUO | SETEQ.
- Move x86 target over.
llvm-svn: 57542
- Move the EH landing-pad code and adjust it so that it works
with FastISel as well as with SDISel.
- Add FastISel support for @llvm.eh.exception and
@llvm.eh.selector.
llvm-svn: 57539
instead of requiring all "short description" strings to begin with
two spaces. This makes these strings less mysterious, and it fixes
some cases where short description strings mistakenly did not
begin with two spaces.
llvm-svn: 57521
than the type an i1 is promoted to (eg: i8). Account
for this. Noticed by Tilmann Scheller on CellSPU; he
will hopefully take care of fixing this in LegalizeDAG
and adding a testcase!
llvm-svn: 56997
`-fno-builtin' flag. Currently, it's used to replace "memset" with "_bzero"
instead of "__bzero" on Darwin10+. This arguably violates the meaning of this
flag, but is currently sufficient. The meaning of this flag should become more
specific over time.
llvm-svn: 56885
Completely eliminate the TopOrder std::vector. Instead, sort
the AllNodes list in place. This also eliminates the need to
call AllNodes.size(), a linear-time operation, before
performing the sort.
Also, eliminate the Sources temporary std::vector, since it
essentially duplicates the sorted result as it is being
built.
This also changes the direction of the topological sort
from bottom-up to top-down. The AllNodes list starts out in
roughly top-down order, so this reduces the amount of
reordering needed. Top-down is also more convenient for
Legalize, and ISel needed only minor adjustments.
llvm-svn: 56867
its size). Adjust various lowering functions to
pass this info through from CallInst. Use it to
implement sseregparm returns on X86. Remove
X86_ssecall calling convention.
llvm-svn: 56677
s/ParamAttr/Attribute/g
s/PAList/AttrList/g
s/FnAttributeWithIndex/AttributeWithIndex/g
s/FnAttr/Attribute/g
This sets the stage
- to implement function notes as function attributes and
- to distinguish between function attributes and return value attributes.
This requires corresponding changes in llvm-gcc and clang.
llvm-svn: 56622
the SelectionDAG and DAGCombiner code. The only functionality change is that now
the DAG combiner is performing the constant folding for these operations instead
of being a no-op.
This is *not* in response to a bug, so there isn't a testcase.
llvm-svn: 56550
RA problem by expanding the live interval of an
earlyclobber def back one slot. Remove
overlap-earlyclobber throughout. Remove
earlyclobber bits and their handling from
live internals.
llvm-svn: 56539
copy of the BURRList scheduler, but with several parts ripped
out, such as backtracking, online topological sort maintenance
(needed by backtracking), the priority queue, and Sethi-Ullman
number computation and maintenance (needed by the priority
queue). As a result of all this, it generates somewhat lower
quality code, but that's its tradeoff for running about 30%
faster than list-burr in -fast mode in many cases.
This is somewhat experimental. Moving forward, major pieces of
this can be refactored with pieces in common with
ScheduleDAGRRList.cpp.
llvm-svn: 56307
with an earlyclobber operand elsewhere. Propagate
this bit and the earlyclobber bit through SDISel.
Change linear-scan RA not to allocate regs in a way
that conflicts with an earlyclobber. See also comments.
llvm-svn: 56290
ConstantPoolSDNode, using the target's preferred alignment for the
constant type.
In LegalizeDAG, when performing loads from the constant pool, the
ConstantPoolSDNode's alignment is used in the calls to getLoad and
getExtLoad.
This change prevents SelectionDAG::getLoad/getExtLoad from incorrectly
choosing the ABI alignment for constant pool loads when Alignment == 0.
The incorrect alignment is only a performance issue when ABI alignment
does not equal preferred alignment (i.e., on x86 it was generating
MOVUPS instead of MOVAPS for v4f32 constant loads when the default ABI
alignment for 128bit vectors is forced to 1 byte.)
Patch by Paul Redmond!
llvm-svn: 56253
- Add linkage to SymbolSDNode (default to external).
- Change ISD::ExternalSymbol to ISD::Symbol.
- Change ISD::TargetExternalSymbol to ISD::TargetSymbol
These changes pave the way to allowing SymbolSDNodes with non-external linkage.
llvm-svn: 56249
Currently it just holds the calling convention and flags
for isVarArgs and isTailCall.
And it has several utility methods, which eliminate magic
5+2*i and similar index computations in several places.
CallSDNodes are not CSE'd. Teach UpdateNodeOperands to handle
nodes that are not CSE'd gracefully.
llvm-svn: 56183
ConstantFP* instead of APInt and APFloat directly.
This reduces the amount of time to create ConstantSDNode
and ConstantFPSDNode nodes when ConstantInt* and ConstantFP*
respectively are already available, as is the case in
SelectionDAGBuild.cpp. Also, it reduces the amount of time
to legalize constants into constant pools, and the amount of
time to add ConstantFP operands to MachineInstrs, due to
eliminating ConstantInt::get and ConstantFP::get calls.
It increases the amount of work needed to create new constants
in cases where the client doesn't already have a ConstantInt*
or ConstantFP*, such as legalize expanding 64-bit integer constants
to 32-bit constants. And it adds a layer of indirection for the
accessor methods. But these appear to be outweight by the benefits
in most cases.
It will also make it easier to make ConstantSDNode and
ConstantFPNode more consistent with ConstantInt and ConstantFP.
llvm-svn: 56162
form:
powf(10.0f, x);
If this is the case, and also we want limited precision floating-point
calculations, then lower to do the limited-precision stuff.
llvm-svn: 56035
to being off by default. Also, add assertion checks to check that
the various fast-isel-related command-line options are only used
when -fast-isel itself is enabled.
llvm-svn: 56029
before. This is taken care of in the selection DAG pass. In my opinion, this
should be in one place or the other. I.e., it should probably be removed from
the DAG combiner (along with the other arithmetic transformations on constants
that are essentially no-ops).
llvm-svn: 55889
out of ScheduleDAGEmit.cpp and into SelectionDAGISel.cpp. This
allows it to be run exactly once per function, even if multiple
SelectionDAG iterations happen in the entry block, as may happen
with FastISel.
llvm-svn: 55863
but less accurate (non-IEEE) code sequences for
certain math library functions. Add the first of
several such expansions. Don't worry, if you don't
turn it on it won't affect you.
llvm-svn: 55823
HandlePHINodesInSuccessorBlocks that works FastISel-style. This
allows PHI nodes to be updated correctly while using FastISel.
This also involves some code reorganization; ValueMap and
MBBMap are now members of the FastISel class, so they needn't
be passed around explicitly anymore. Also, SelectInstructions
is changed to SelectInstruction, and only does one instruction
at a time.
llvm-svn: 55746
the simple fix, materializing the constant before every use. It might be better to either track domination of uses or
to materialize all constants and the beginning of the function and let remat sort when to do materialization at uses.
llvm-svn: 55703
The first can update the SDNode in an SDValue
while the second is called with SDNode* and
returns a possibly updated SDNode*.
This patch has no intended functional impact,
but helps eliminating ugly temporary SDValues.
llvm-svn: 55608
assignment when selecting the def. This is the naive solution to the problem: insert a copy to the pre-chosen
vreg. Other solutions might be preferable, such as:
1) Passing the dest reg into FastEmit_. However, this would require the higher level code to know about reg classes, which they don't currently.
2) Selecting blocks in reverse postorder. This has some compile time cost for computing the order, and we'd need to measure its impact.
llvm-svn: 55555
its work by putting all nodes in the worklist, requiring a big
dynamic allocation. Now, DAGCombiner just iterates over the AllNodes
list and maintains a worklist for nodes that are newly created or
need to be revisited. This allows the worklist to stay small in most
cases, so it can be a SmallVector.
This has the side effect of making DAGCombine not miss a folding
opportunity in alloca-align-rounding.ll.
llvm-svn: 55498
SelectionDAGLowering instead of being in an anonymous namespace.
This fixes warnings about SelectionDAGLowering having fields
using anonymous namespaces.
llvm-svn: 55497
ATOMIC_LOAD_ADD_{8,16,32,64} instead of ATOMIC_LOAD_ADD.
Increased the Hardcoded Constant OpActionsCapacity to match.
Large but boring; no functional change.
This is to support partial-word atomics on ppc; i8 is
not a valid type there, so by the time we get to lowering, the
ATOMIC_LOAD nodes looks the same whether the type was i8 or i32.
The information can be added to the AtomicSDNode, but that is the
largest SDNode; I don't fully understand the SDNode allocation,
but it is sensitive to the largest node size, so increasing
that must be bad. This is the alternative.
llvm-svn: 55457
works with.
SelectionDAG, FunctionLoweringInfo, and SelectionDAGLowering
objects now get created once per SelectionDAGISel instance, and
can be reused across blocks and across functions. Previously,
they were created and destroyed each time they were needed.
This reorganization simplifies the handling of PHI nodes, and
also SwitchCases, JumpTables, and BitTestBlocks. This
simplification has the side effect of fixing a bug in FastISel
where successor PHI nodes weren't being updated correctly.
This is also a step towards making the transition from FastISel
into and out of SelectionDAG faster, and also making
plain SelectionDAG faster on code with lots of little blocks.
llvm-svn: 55450
of two, and to not need a scratch std::vector. Also, compute the ordering
immediately in the result array, instead of in another scratch std::vector
that is copied to the result array.
llvm-svn: 55421
of two, and to not need a scratch std::vector. Also, use the
SelectionDAG's topological sort in LegalizeDAG instead of having
a separate implementation.
llvm-svn: 55389
was inserted or not. This allows bitcast in fast isel to properly handle the case
where an appropriate reg-to-reg copy is not available.
llvm-svn: 55375
RecyclingAllocator, but this change is needed for the nodes
to actually be recycled. This cuts SelectionDAG's memory
usage high-water-mark in half in some cases.
llvm-svn: 55351
use raw_ostream instead of std::ostream. Among other goodness,
this speeds up llvm-dis of kc++ with a release build from 0.85s
to 0.49s (88% faster).
Other interesting changes:
1) This makes Value::print be non-virtual.
2) AP[S]Int and ConstantRange can no longer print to ostream directly,
use raw_ostream instead.
3) This fixes a bug in raw_os_ostream where it didn't flush itself
when destroyed.
4) This adds a new SDNode::print method, instead of only allowing "dump".
A lot of APIs have both std::ostream and raw_ostream versions, it would
be useful to go through and systematically anihilate the std::ostream
versions.
This passes dejagnu, but there may be minor fallout, plz let me know if
so and I'll fix it.
llvm-svn: 55263
process up to a higher level. This allows FastISel to leverage
more of SelectionDAGISel's infastructure, such as updating Machine
PHI nodes.
Also, implement transitioning from SDISel back to FastISel in
the middle of a block, so it's now possible to go back and
forth. This allows FastISel to hand individual CallInsts and other
complicated things off to SDISel to handle, while handling the rest
of the block itself.
To help support this, reorganize the SelectionDAG class so that it
is allocated once and reused throughout a function, instead of
being completely reallocated for each block.
llvm-svn: 55219
and use it in FastISelEmitter.cpp, and make FastISel
subtarget aware. Among other things, this lets it work
properly on x86 targets that don't have SSE, where it
successfully selects x87 instructions.
llvm-svn: 55156
class hold a MachineRegisterInfo member, and make the
MachineBasicBlock be passed in to SelectInstructions rather
than the FastISel constructor.
llvm-svn: 55076
In particular, Collector was confusing to implementors. Several
thought that this compile-time class was the place to implement
their runtime GC heap. Of course, it doesn't even exist at runtime.
Specifically, the renames are:
Collector -> GCStrategy
CollectorMetadata -> GCFunctionInfo
CollectorModuleMetadata -> GCModuleInfo
CollectorRegistry -> GCRegistry
Function::getCollector -> getGC (setGC, hasGC, clearGC)
Several accessors and nested types have also been renamed to be
consistent. These changes should be obvious.
llvm-svn: 54899
returning an std::string by value, it fills in a SmallString/SmallVector
passed in. This significantly reduces string thrashing in some cases.
More specifically, this:
- Adds an operator<< and a print method for APInt that allows you to
directly send them to an ostream.
- Reimplements APInt::toString to be much simpler and more efficient
algorithmically in addition to not thrashing strings quite as much.
This speeds up llvm-dis on kc++ by 7%, and may also slightly speed up the
asmprinter. This also fixes a bug I introduced into the asmwriter in a
previous patch w.r.t. alias printing.
llvm-svn: 54873
FPROUND_F80_F32, FPROUND_PPCF128_F32,
FPROUND_F80_F64, FPROUND_PPCF128_F64
Support for soften float fp_round operands is added, Mips
needs this to round f64->f32.
Also added support to soften float FABS result, Mips doesn't
support double fabs results while in 'single float only' mode.
llvm-svn: 54484
- Add a basic machine-level dead block eliminator.
These two have to go together, since many other parts of the code generator are unable to handle the unreachable blocks otherwise created.
llvm-svn: 54333
switches use the binary search algorithm) for
environments that don't support it. PPC64 JIT
is such an environment; turn the flag on for that.
llvm-svn: 54248
a new ilist_node class, and remove them. Unlike alist_node,
ilist_node doesn't attempt to manage storage itself, so it avoids
the associated problems, including being opaque in gdb.
Adjust the Recycler class so that it doesn't depend on alist_node.
Also, change it to use explicit Size and Align parameters, allowing
it to work when the largest-sized node doesn't have the greatest
alignment requirement.
Change MachineInstr's MachineMemOperand list from a pool-backed
alist to a std::list for now.
llvm-svn: 54146
Remove the GetResultInst instruction. It is still accepted in LLVM assembly
and bitcode, where it is now auto-upgraded to ExtractValueInst. Also, remove
support for return instructions with multiple values. These are auto-upgraded
to use InsertValueInst instructions.
The IRBuilder still accepts multiple-value returns, and auto-upgrades them
to InsertValueInst instructions.
llvm-svn: 53941
SelectionDAG graph writer to make use of them. Now, nodes with multiple
values are displayed as such, with incoming edges pointing to the
specific value they use.
llvm-svn: 53875
generic SDNode's (nodes with their own constructors
should do sanity checking in the constructor). Add
sanity checks for BUILD_VECTOR and fix all the places
that were producing bogus BUILD_VECTORs, as found by
"make check". My favorite is the BUILD_VECTOR with
only two operands that was being used to build a
vector with four elements!
llvm-svn: 53850
the night realising that it was wrong :) I
think the reason the same type was being used
for the shufflevec of indices as for the actual
indices is so that if one of them needs splitting
then so does the other. After my patch it might
be that the indices need splitting but not the
rest, yet there is no good way of handling that.
I think the right solution is to not have the
shufflevec be an operand at all: just have it
be the list of numbers it actually is, stored
as extra info in the node.
llvm-svn: 53768
PseudoSourceValue values, which never have names. Use getName()
for all other values, because we want to print just a short summary
of the value, not the entire instruction.
llvm-svn: 53738
mask. These are just indices into the shuffled vector
so their type is unrelated to the type of the
shuffled elements (which is what was being used before).
This fixes vec_shuffle-11.ll when using LegalizeTypes.
What seems to have happened is that Dan's recent change
r53687, which corrected the result type of the shuffle,
somehow caused LegalizeTypes to notice that the mask
operand was a BUILD_VECTOR with a legal type but elements
of an illegal type (i64). LegalizeTypes legalized this
by introducing a new BUILD_VECTOR of i32 and bitcasting
it to the old type. But the mask operand is not supposed
to be a bitcast but a straight BUILD_VECTOR of constants,
causing a crash.
llvm-svn: 53729
replacement of multiple values. This is slightly more efficient
than doing multiple ReplaceAllUsesOfValueWith calls, and theoretically
could be optimized even further. However, an important property of this
new function is that it handles the case where the source value set and
destination value set overlap. This makes it feasible for isel to use
SelectNodeTo in many very common cases, which is advantageous because
SelectNodeTo avoids a temporary node and it doesn't require CSEMap
updates for users of values that don't change position.
Revamp MorphNodeTo, which is what does all the work of SelectNodeTo, to
handle operand lists more efficiently, and to correctly handle a number
of corner cases to which its new wider use exposes it.
This commit also includes a change to the encoding of post-isel opcodes
in SDNodes; now instead of being sandwiched between the target-independent
pre-isel opcodes and the target-dependent pre-isel opcodes, post-isel
opcodes are now represented as negative values. This makes it possible
to test if an opcode is pre-isel or post-isel without having to know
the size of the current target's post-isel instruction set.
These changes speed up llc overall by 3% and reduce memory usage by 10%
on the InstructionCombining.cpp testcase with -fast and -regalloc=local.
llvm-svn: 53728
In LegalizeDAG the value is zero-extended to
the new type before byte swapping. It doesn't
matter how the extension is done since the new
bits are shifted off anyway after the swap, so
extend by any old rubbish bits. This results
in the final assembler for the testcase being
one line shorter.
llvm-svn: 53604
extending load of a vector. Handle this case when
splitting vector loads. I'm not completely sure
what is supposed to happen, but I think it means
hi should be set to undef. LegalizeDAG does not
consider this case.
llvm-svn: 53555
stores of one-element vectors. Also, neaten the
handling of INSERT_VECTOR_ELT when the inserted
type is larger than the vector element type.
llvm-svn: 53554
are used for passing huge immediates in inline ASM
from the front-end straight down to the ASM writer.
Of course this is a hack, but it is simple, limited
in scope, works in practice, and is what LegalizeDAG
does.
llvm-svn: 53553
SINT_TO_FP libcall plus additional operations:
it might as well be a direct UINT_TO_FP libcall.
So only turn it into an SINT_TO_FP if the target
has special handling for SINT_TO_FP.
llvm-svn: 53461
Lack of these caused a bootstrap failure with Fortran
on x86-64 with LegalizeTypes turned on. While there,
be nice to 16 bit machines and support expansion of
i32 too.
llvm-svn: 53408
makes their special-case checks of use_size() less beneficial,
so remove them. This eliminates all but one use of use_size(),
which is in AssignTopologicalOrder, which uses it only once for
each node, and so can reasonably afford to recompute it, as
this allows the UsesSize field of SDNode to be removed
altogether.
llvm-svn: 53377
class, and store IsVolatile and Alignment in a more compact form.
This makes AtomicSDNode slightly larger, but it shrinks LoadSDNode
and StoreSDNode, which are much more common and are the largest of
the SDNode subclasses. Also, this lets the isVolatile() and
getAlignment() accessors be non-virtual.
llvm-svn: 53361
SINT_TO_FP and UINT_TO_FP. This now produces
the same code as LegalizeDAG (the previous
code was based on a mistaken idea of what
LegalizeDAG did in this case).
llvm-svn: 53288
soft float: experiments show that targets aren't
expecting this for results or for operands. Add
support select/select_cc result soft float and
correct operand soft float for these.
llvm-svn: 53245
MachineMemOperands. The pools are owned by MachineFunctions.
This drastically reduces the number of calls to malloc/free made
during the "Emit" phase of scheduling, as well as later phases
in CodeGen. Combined with other changes, this speeds up the
"instruction selection" phase of CodeGen by 10% in some cases.
llvm-svn: 53212
and reused across SelectionDAGs.
This drastically reduces the number of calls to malloc/free made during
instruction selection, and improves memory locality.
llvm-svn: 53211
simple const SDOperand*, which is what's usually needed.
For AddNodeIDOperands, which is small, just duplicate the function to
accept an SDUse*.
For SelectionDAG::getNode - Add an overload that accepts SDUse* that
copies the operands into a temporary SDOperand array, but also has
special-case checks for 0 through 3 operands to avoid the copy in
the common cases.
llvm-svn: 53183
that fixed problems in EmitStackConvert where the source and target type
have different alignment by creating a stack slot with the max
alignment of source and target type.
llvm-svn: 53150
hook for each way in which a result type can be
legalized (promotion, expansion, softening etc),
just use one: ReplaceNodeResults, which returns
a node with exactly the same result types as the
node passed to it, but presumably with a bunch of
custom code behind the scenes. No change if the
new LegalizeTypes infrastructure is not turned on.
llvm-svn: 53137
SelectionDAG::SelectNodeTo in the instruction selector. This
updates existing nodes in place instead of creating new ones.
Go back to selecting ISD::DBG_LABEL nodes into
TargetInstrInfo::DBG_LABEL nodes instead of leaving them
unselected, now that SelectNodeTo allows us to update them
in place.
llvm-svn: 53057
to be passed the list of value types, and use this
where appropriate. Inappropriate places are where
the value type list is already known and may be
long, in which case the existing method is more
efficient.
llvm-svn: 53035
the need for a flavor operand, and add a new SDNode subclass,
LabelSDNode, for use with them to eliminate the need for a label id
operand.
Change instruction selection to let these label nodes through
unmodified instead of creating copies of them. Teach the MachineInstr
emitter how to emit a MachineInstr directly from an ISD label node.
This avoids the need for allocating SDNodes for the label id and
flavor value, as well as SDNodes for each of the post-isel label,
label id, and label flavor.
llvm-svn: 52943
SelectionDAG::allnodes_size is linear, but that doesn't appear to
outweigh the benefit of reducing heap traffic. If it does become a
problem, we should teach SelectionDAG to keep a count of how many
nodes are live, because there are several other places where that
information would be useful as well.
llvm-svn: 52926
purpose, and give it a custom SDNode subclass so that it doesn't
need to have line number, column number, filename string, and
directory string, all existing as individual SDNodes to be the
operands.
This was the only user of ISD::STRING, StringSDNode, etc., so
remove those and some associated code.
This makes stop-points considerably easier to read in
-view-legalize-dags output, and reduces overhead (creating new
nodes and copying std::strings into them) on code containing
debugging information.
llvm-svn: 52924
SmallVectors. Change the signature of TargetLowering::LowerArguments
to avoid returning a vector by value, and update the two targets
which still use this directly, Sparc and IA64, accordingly.
llvm-svn: 52917
only needs one bit for each register. UsedRegs is a SmallVector
sized at 16, so this eliminates a heap allocation/free for every
call and return processed by Legalize on most targets.
llvm-svn: 52915
it impossible to create a MERGE_VALUES node with
only one result: sometimes it is useful to be able
to create a node with only one result out of one of
the results of a node with more than one result, for
example because the new node will eventually be used
to replace a one-result node using ReplaceAllUsesWith,
cf X86TargetLowering::ExpandFP_TO_SINT. On the other
hand, most users of MERGE_VALUES don't need this and
for them the optimization was valuable. So add a new
utility method getMergeValues for creating MERGE_VALUES
nodes which by default performs the optimization.
Change almost everywhere to use getMergeValues (and
tidy some stuff up at the same time).
llvm-svn: 52893
Move GetConstantStringInfo to lib/Analysis. Remove
string output routine from Constant. Update all
callers. Change debug intrinsic api slightly to
accomodate move of routine, these now return values
instead of strings.
This unbreaks llvm-gcc bootstrap.
llvm-svn: 52884
string output routine from Constant. Update all
callers. Change debug intrinsic api slightly to
accomodate move of routine, these now return values
instead of strings.
llvm-svn: 52748
For this it is convenient to permit floats to
be used with EXTRACT_ELEMENT, so I tweaked
things to allow that. I also added libcalls
for ppcf128 to i32 forms of FP_TO_XINT, since
they exist in libgcc and this case can certainly
occur (and does occur in the testsuite) - before
the i64 libcall was being used. Also, the
XINT_TO_FP result seemed to be wrong when
the argument is an i128: the wrong fudge
factor was added (the i32 and i64 cases were
handled directly, but the i128 code fell
through to some generic softening code which
seemed to think it was i64 to f32!). So I
fixed it by adding a fudge factor that I
found in my breakfast cereal.
llvm-svn: 52739
Added abstract class MemSDNode for any Node that have an associated MemOperand
Changed atomic.lcs => atomic.cmp.swap, atomic.las => atomic.load.add, and
atomic.lss => atomic.load.sub
llvm-svn: 52706
fixes PR2476; patch by Richard Osborne. The same
problem exists for a bunch of other operators, but
I'm ignoring this because they will be automagically
fixed when the new LegalizeTypes infrastructure lands,
since it already solves this problem centrally.
llvm-svn: 52610
to DenseMap<SDNode*, SUnit*>, and adjust the way cloned SUnit nodes are
handled so that only the original node needs to be in the map.
This speeds up llc on 447.dealII.llvm.bc by about 2%.
llvm-svn: 52576
integer of the same type. Before it was "promotion",
but this is confusing because it is quite different
to promotion of integers. Call it "softening" instead,
inspired by "soft float".
llvm-svn: 52546
rather than bundling them together. Rename FloatToInt
to PromoteFloat (better, if not perfect). Reorganize
files by types rather than by operations.
llvm-svn: 52408
of value info (sign/zero ext info) from one MBB to another. This doesn't
handle much right now because of two limitations:
1) only handles zext/sext, not random bit propagation (no assert exists
for this)
2) doesn't handle phis.
llvm-svn: 52383
still excluding types like i1 (not byte sized)
and i120 (loading an i120 requires loading an i64,
an i32, an i16 and an i8, which is expensive).
llvm-svn: 52310
wrong for volatile loads and stores. In fact this
is almost all of them! There are three types of
problems: (1) it is wrong to change the width of
a volatile memory access. These may be used to
do memory mapped i/o, in which case a load can have
an effect even if the result is not used. Consider
loading an i32 but only using the lower 8 bits. It
is wrong to change this into a load of an i8, because
you are no longer tickling the other three bytes. It
is also unwise to make a load/store wider. For
example, changing an i16 load into an i32 load is
wrong no matter how aligned things are, since the
fact of loading an additional 2 bytes can have
i/o side-effects. (2) it is wrong to change the
number of volatile load/stores: they may be counted
by the hardware. (3) it is wrong to change a volatile
load/store that requires one memory access into one
that requires several. For example on x86-32, you
can store a double in one processor operation, but to
store an i64 requires two (two i32 stores). In a
multi-threaded program you may want to bitcast an i64
to a double and store as a double because that will
occur atomically, and be indivisible to other threads.
So it would be wrong to convert the store-of-double
into a store of an i64, because this will become two
i32 stores - no longer atomic. My policy here is
to say that the number of processor operations for
an illegal operation is undefined. So it is alright
to change a store of an i64 (requires at least two
stores; but could be validly lowered to memcpy for
example) into a store of double (one processor op).
In short, if the new store is legal and has the same
size then I say that the transform is ok. It would
also be possible to say that transforms are always
ok if before they were illegal, whether after they
are illegal or not, but that's more awkward to do
and I doubt it buys us anything much.
However this exposed an interesting thing - on x86-32
a store of i64 is considered legal! That is because
operations are marked legal by default, regardless of
whether the type is legal or not. In some ways this
is clever: before type legalization this means that
operations on illegal types are considered legal;
after type legalization there are no illegal types
so now operations are only legal if they really are.
But I consider this to be too cunning for mere mortals.
Better to do things explicitly by testing AfterLegalize.
So I have changed things so that operations with illegal
types are considered illegal - indeed they can never
map to a machine operation. However this means that
the DAG combiner is more conservative because before
it was "accidentally" performing transforms where the
type was illegal because the operation was nonetheless
marked legal. So in a few such places I added a check
on AfterLegalize, which I suppose was actually just
forgotten before. This causes the DAG combiner to do
slightly more than it used to, which resulted in the X86
backend blowing up because it got a slightly surprising
node it wasn't expecting, so I tweaked it.
llvm-svn: 52254
maps can be deleted. This happens when RAUW
replaces a node N with another equivalent node
E, deleting the first node. Solve this by
adding (N, E) to ReplacedNodes, which is already
used to remap nodes to replacements. This means
that deleted nodes are being allowed in maps,
which can be delicate: the memory may be reused
for a new node which might get confused with the
old deleted node pointer hanging around in the
maps, so detect this and flush out maps if it
occurs (ExpungeNode). The expunging operation
is expensive, however it never occurs during
a llvm-gcc bootstrap or anywhere in the nightly
testsuite. It occurs three times in "make check":
Alpha/illegal-element-type.ll,
PowerPC/illegal-element-type.ll and
X86/mmx-shift.ll. If expunging proves to be too
expensive then there are other more complicated
ways of solving the problem.
In the normal case this patch adds the overhead
of a few more map lookups, which is hopefully
negligable.
llvm-svn: 52214
of integer types. Fix the isMask APInt method to
actually work (hopefully) rather than crashing
because it adds apints of different bitwidths.
It looks like isShiftedMask is also broken, but
I'm leaving that one to the APInt people (it is
not used anywhere).
llvm-svn: 52142
of apint codegen failure is the DAG combiner doing
the wrong thing because it was comparing MVT's using
< rather than comparing the number of bits. Removing
the < method makes this mistake impossible to commit.
Instead, add helper methods for comparing bits and use
them.
llvm-svn: 52098
no visible functionality change, but enables a future patch where node creation
will update the CFG if it decides to create an unconditional rather than a conditional branch.
llvm-svn: 52067
and better control the abstraction. Rename the type
to MVT. To update out-of-tree patches, the main
thing to do is to rename MVT::ValueType to MVT, and
rewrite expressions like MVT::getSizeInBits(VT) in
the form VT.getSizeInBits(). Use VT.getSimpleVT()
to extract a MVT::SimpleValueType for use in switch
statements (you will get an assert failure if VT is
an extended value type - these shouldn't exist after
type legalization).
This results in a small speedup of codegen and no
new testsuite failures (x86-64 linux).
llvm-svn: 52044
Move platform independent code (lowering of possibly overwritten
arguments, check for tail call optimization eligibility) from
target X86ISelectionLowering.cpp to TargetLowering.h and
SelectionDAGISel.cpp.
Initial PowerPC tail call implementation:
Support ppc32 implemented and tested (passes my tests and
test-suite llvm-test).
Support ppc64 implemented and half tested (passes my tests).
On ppc tail call optimization is performed if
caller and callee are fastcc
call is a tail call (in tail call position, call followed by ret)
no variable argument lists or byval arguments
option -tailcallopt is enabled
Supported:
* non pic tail calls on linux/darwin
* module-local tail calls on linux(PIC/GOT)/darwin(PIC)
* inter-module tail calls on darwin(PIC)
If constraints are not met a normal call will be emitted.
A test checking the argument lowering behaviour on x86-64 was added.
llvm-svn: 50477
This removes the existing bottleneck related to the removal of elements from
the middle of the queue.
Also fixes a subtle bug in ScheduleDAGRRList::CapturePred:
It was updating the state of the SUnit before removing it. As a result, the
comparison operators were working incorrectly and this SUnit could not be removed
from the queue properly.
Reviewed by Evan and Dan. Approved by Dan.
llvm-svn: 50412
We now compile test2/test3 to:
_test2:
## InlineAsm Start
set %xmm0, %xmm1
## InlineAsm End
addps %xmm1, %xmm0
ret
_test3:
## InlineAsm Start
set %xmm0, %xmm1
## InlineAsm End
paddd %xmm1, %xmm0
ret
as expected.
llvm-svn: 50389
towards PR2094. It now compiles the attached .ll file to:
_sad16_sse2:
movslq %ecx, %rax
## InlineAsm Start
%ecx %rdx %rax %rax %r8d %rdx %rsi
## InlineAsm End
## InlineAsm Start
set %eax
## InlineAsm End
ret
which is pretty decent for a 3 output, 4 input asm.
llvm-svn: 50386
c1, f1 = CopyToReg
c2, f2 = CopyToReg
c3 = TokenFactor c1, c2
...
= user c3, ..., f2
Now that the two CopyToReg's and the user are "flagged" together. They effectively forms a single scheduling unit. The TokenFactor is now both an operand and a successor of the Flagged nodes.
llvm-svn: 50376
ComputeMaskedBits knows about cttz, ctlz, and ctpop. Teach
SelectionDAG's ComputeMaskedBits what InstCombine's knows
about SRem. And teach them both some things about high bits
in Mul, UDiv, URem, and Sub. This allows instcombine and
dagcombine to eliminate sign-extension operations in
several new cases.
llvm-svn: 50358
conversion open the door for many nasty implicit conversion issues, and
can be easily solved by initializing with (V.begin(), V.end()) when
needed.
This patch includes many small cleanups for sdisel also.
llvm-svn: 50340
When choosing between constraints with multiple options,
like "ir", test to see if we can use the 'i' constraint and
go with that if possible. This produces more optimal ASM in
all cases (sparing a register and an instruction to load it),
and fixes inline asm like this:
void test () {
asm volatile (" %c0 %1 " : : "imr" (42), "imr"(14));
}
Previously we would dump "42" into a memory location (which
is ok for the 'm' constraint) which would cause a problem
because the 'c' modifier is not valid on memory operands.
Isn't it great how inline asm turns 'missed optimization'
into 'compile failed'??
Incidentally, this was the todo in
PowerPC/2007-04-24-InlineAsm-I-Modifier.ll
Please do NOT pull this into Tak.
llvm-svn: 50315
- Make targetlowering.h fit in 80 cols.
- Make LowerAsmOperandForConstraint const.
- Make lowerXConstraint -> LowerXConstraint
- Make LowerXConstraint return a const char* instead of taking a string byref.
llvm-svn: 50312
to the block that defines their operands. This doesn't work in the
case that the operand is an invoke, because invoke is a terminator
and must be the last instruction in a block.
Replace it with support in SelectionDAGISel for copying struct values
into sequences of virtual registers.
llvm-svn: 50279
rather than having it suck them out of a node. Add
a bunch of new libcalls, and remove dead softfloat
code (dead, because FloatToInt is used not Expand
in this case). Note that indexed stores probably
aren't handled properly, likewise for loads.
llvm-svn: 49915
Rename SDOperandImpl back to SDOperand.
Introduce the SDUse class that represents a use of the SDNode referred by
an SDOperand. Now it is more similar to Use/Value classes.
Patch is approved by Dan Gohman.
llvm-svn: 49795
ScheduleDAG; they don't correspond to any actual instructions so they
don't need to be scheduled.
This fixes a bug where the EntryToken was being scheduled multiple
times in some cases, though it ended up not causing any trouble because
EntryToken doesn't expand into anything. With this fixed the schedulers
reliably schedule the expected number of units, so we can check this
with an assertion.
This requires a tweak to test/CodeGen/X86/loop-hoist.ll because it
ends up getting scheduled differently in a trivial way, though it was
enough to fool the prcontext+grep that the test does.
llvm-svn: 49701
much simpler than in LegalizeDAG because calls are
not yet expanded into call sequences: that happens
after type legalization has finished.
llvm-svn: 49634
in its maps. Add some sanity checks that catch
this kind of thing. Hopefully these can be
removed one day (once all problems are fixed!)
but for the moment it seems wise to have them in.
llvm-svn: 49612
on any current target and aren't optimized in DAGCombiner. Instead
of using intermediate nodes, expand the operations, choosing between
simple loads/stores, target-specific code, and library calls,
immediately.
Previously, the code to emit optimized code for these operations
was only used at initial SelectionDAG construction time; now it is
used at all times. This fixes some cases where rep;movs was being
used for small copies where simple loads/stores would be better.
This also cleans up code that checks for alignments less than 4;
let the targets make that decision instead of doing it in
target-independent code. This allows x86 to use rep;movs in
low-alignment cases.
Also, this fixes a bug that resulted in the use of rep;stos for
memsets of 0 with non-constant memory size when the alignment was
at least 4. It's better to use the library in this case, which
can be significantly faster when the size is large.
This also preserves more SourceValue information when memory
intrinsics are lowered into simple loads/stores.
llvm-svn: 49572
before an invoke. Failure to do this causes references in
the landing pad to variables that were not set. Fixes
g++.dg/eh/delayslot1.C
g++.dg/eh/fp-regs.C
g++.old-deja/g++.brendan/eh1.C
llvm-svn: 49243
There is no point in creating a long live range defined by an implicit_def. Scheduler now duplicates implicit_def instruction for each of its uses. Therefore, if an implicit_def node has multiple uses, it will become a number of very short live ranges, rather than a long one. This will make coalescer's job easier.
llvm-svn: 49164
review feedback.
-enable-eh is still accepted but doesn't do anything.
EH intrinsics use Dwarf EH if the target supports that,
and are handled by LowerInvoke otherwise.
The separation of the EH table and frame move data is,
I think, logically figured out, but either one still
causes full EH info to be generated (not sure how to
split the metadata correctly).
MachineModuleInfo::needsFrameInfo is no longer used and
is removed.
llvm-svn: 49064
not marked nounwind, or for all functions when -enable-eh
is set, provided the target supports Dwarf EH.
llvm-gcc generates nounwind in the right places; other FEs
will need to do so also. Given such a FE, -enable-eh should
no longer be needed.
llvm-svn: 49006
In order to handle indexed nodes I had to introduce
a new constructor, and since I was there I factorized
the code in the various load constructors.
llvm-svn: 48894
nodes. This doesn't currently have much impact the generated code, but it
does produce simpler-looking SelectionDAGs, and consequently
simpler-looking ScheduleDAGs, because there are fewer spurious
dependencies.
In particular, CopyValueToVirtualRegister now uses the entry node as the
input chain dependency for new CopyToReg nodes instead of calling getRoot
and depending on the most recent memory reference.
Also, rename UnorderedChains to PendingExports and pull it up from being
a local variable in SelectionDAGISel::BuildSelectionDAG to being a
member variable of SelectionDAGISel, so that it doesn't have to be
passed around to all the places that need it.
llvm-svn: 48893
called LimitedSumOfUnscheduledPredsOfSuccs. It terminates the computation
after a given treshold is reached. This new function is always faster, but
brings real wins only on bigger test-cases.
The old function SumOfUnscheduledPredsOfSuccs is left in-place for now and therefore a warning about an unused static function is produced.
llvm-svn: 48872
LLVM Value/Use does and MachineRegisterInfo/MachineOperand does.
This allows constant time for all uses list maintenance operations.
The idea was suggested by Chris. Reviewed by Evan and Dan.
Patch is tested and approved by Dan.
On normal use-cases compilation speed is not affected. On very big basic
blocks there are compilation speedups in the range of 15-20% or even better.
llvm-svn: 48822
This fixes Bugzilla #1835 (http://llvm.org/bugs/show_bug.cgi?id=1835).
This patched is reviewed by Tanya and Dan. Dan tested and approved it.
The reason for the bad performance of the old algorithm is that it is very naive and scans every
time all nodes of the DAG in the worst case.
This patch introduces a new algorithm based on the paper "Online algorithms
for maintaining the topological order of a directed acyclic graph" by
David J.Pearce and Paul H.J.Kelly. This is the MNR algorithm. It has a
linear time worst-case and performs much better in most situations.
The paper can be found here:
http://fano.ics.uci.edu/cites/Document/Online-algorithms-for-maintaining-the-topological-order-of-a-directed-acyclic-graph.html
The main idea of the new algorithm is to compute the topological ordering of the SNodes in the
DAG and to maintain it even after DAG modifications. The topological ordering allows for very fast
node reachability checks.
Tests on very big input files with tens of thousands of instructions in a BB indicate huge
speed-ups (up to 10x compilation time improvement) compared to the old version.
llvm-svn: 48817
flags. This is needed by the new legalize types
infrastructure which wants to expand the 64 bit
constants previously used to hold the flags on
32 bit machines. There are two functional changes:
(1) in LowerArguments, if a parameter has the zext
attribute set then that is marked in the flags;
before it was being ignored; (2) PPC had some bogus
code for handling two word arguments when using the
ELF 32 ABI, which was hard to convert because of
the bogusness. As suggested by the original author
(Nicolas Geoffray), I've disabled it for the moment.
Tested with "make check" and the Ada ACATS testsuite.
llvm-svn: 48640
the fcopysign expansion from LegalizeDAG to get rid of
what seems to be a bug: the use of sign extension means
that when copying the sign bit from an f32 to an f64,
the upper 32 bits of the f64 (now an i64) are set, not
just the top bit... I also generalized it to work for
any sized floating point types, and removed the bogosity:
SDOperand Mask1 = (SrcVT == MVT::f64)
? DAG.getConstantFP(BitsToDouble(1ULL << 63), SrcVT)
: DAG.getConstantFP(BitsToFloat(1U << 31), SrcVT);
Mask1 = DAG.getNode(ISD::BIT_CONVERT, SrcNVT, Mask1);
(here SrcNVT is an integer with the same size as SrcVT).
As far as I can see this takes a 1 << 63, converts to
a double, converts that to a floating point constant
then converts that to an integer constant, ending up
with... 1 << 63 as an integer constant! So I just
generate this integer constant directly.
llvm-svn: 48305
getCopyToParts problem was noticed by the new
LegalizeTypes infrastructure. In order to avoid
this kind of thing in the future I've added a
check that EXTRACT_ELEMENT is only used with
integers. Once LegalizeTypes is up and running
most likely BUILD_PAIR and EXTRACT_ELEMENT can
be removed, in favour of using apints instead.
llvm-svn: 48294
X86 lowering normalize vector 0 to v4i32. However DAGCombine can fold (sub x, x) -> 0 after legalization. It can create a zero vector of a type that's not expected (e.g. v8i16). We don't want to disable the optimization since leaving a (sub x, x) is really bad. Add isel patterns for other types of vector 0 to ensure correctness. It's highly unlikely to happen other than in bugpoint reduced test cases.
llvm-svn: 48279
and it's the result that requires expansion. This code is a little confusing
because the TargetLoweringInfo tables for [US]INT_TO_FP use the operand type
(the integer type) rather than the result type.
llvm-svn: 48206
Change insert/extract subreg instructions to be able to be used in TableGen patterns.
Use the above features to reimplement an x86-64 pseudo instruction as a pattern.
llvm-svn: 48130
field to 32 bits, thus enabling correct handling of ByVal
structs bigger than 0x1ffff. Abstract interface a bit.
Fixes gcc.c-torture/execute/pr23135.c and
gcc.c-torture/execute/pr28982b.c in gcc testsuite (were ICE'ing
on ppc32, quietly producing wrong code on x86-32.)
llvm-svn: 48122
they are produced by calls (which are known exact) and by cross block copies
which are known to be produced by extends.
This improves:
define double @test2() {
%tmp85 = call double asm sideeffect "fld0", "={st(0)}"()
ret double %tmp85
}
from:
_test2:
subl $20, %esp
# InlineAsm Start
fld0
# InlineAsm End
fstpl 8(%esp)
movsd 8(%esp), %xmm0
movsd %xmm0, (%esp)
fldl (%esp)
addl $20, %esp
#FP_REG_KILL
ret
to:
_test2:
# InlineAsm Start
fld0
# InlineAsm End
#FP_REG_KILL
ret
by avoiding a f64 <-> f80 trip
llvm-svn: 48108
an RFP register class.
Teach ScheduleDAG how to handle CopyToReg with different src/dst
reg classes.
This allows us to compile trivial inline asms that expect stuff
on the top of x87-fp stack.
llvm-svn: 48107
in different register classes, e.g. copy of ST(0) to RFP*. This gets
some really trivial inline asm working that plops things on the top of
stack (PR879)
llvm-svn: 48105
of BUILD_VECTORS that only have two unique elements:
1. The previous code was nondeterminstic, because it walked a map in
SDOperand order, which isn't determinstic.
2. The previous code didn't handle the case when one element was undef
very well. Now we ensure that the generated shuffle mask has the
undef vector on the RHS (instead of potentially being on the LHS)
and that any elements that refer to it are themselves undef. This
allows us to compile CodeGen/X86/vec_set-9.ll into:
_test3:
movd %rdi, %xmm0
punpcklqdq %xmm0, %xmm0
ret
instead of:
_test3:
movd %rdi, %xmm1
#IMPLICIT_DEF %xmm0
punpcklqdq %xmm1, %xmm0
ret
... saving a register.
llvm-svn: 48060
_test3:
movd %rdi, %xmm1
#IMPLICIT_DEF %xmm0
punpcklqdq %xmm1, %xmm0
ret
instead of:
_test3:
#IMPLICIT_DEF %rax
movd %rax, %xmm0
movd %rdi, %xmm1
punpcklqdq %xmm1, %xmm0
ret
This is still not ideal. There is no reason to two xmm regs.
llvm-svn: 48058
except ppc long double. This allows us to shrink constant pool
entries for x86 long double constants, which in turn allows us to
use flds/fldl instead of fldt.
llvm-svn: 47938
bug in r47928 (Int64Ty is the correct type for the constant
pool entry here) and removes the asserts, now that the code
is capable of handling i128.
llvm-svn: 47932
For x86, if sse2 is available, it's not a good idea since cvtss2sd is slower than a movsd load and it prevents load folding. On x87, it's important to shrink fp constant since fldt is very expensive.
llvm-svn: 47931
The basic idea is that all these algorithms are computing the longest paths from the root node or to the exit node. Therefore the existing implementation that uses and iterative and potentially
exponential algorithm was changed to a well-known graph algorithm based on dynamic programming. It has a linear run-time.
llvm-svn: 47884
same size as an int type by doing a bitconvert of
load/store of the int type (same algorithm as floating point).
This makes them work for ppc Altivec. There was some
code that purported to handle loads of (some) vectors
by splitting them into two smaller vectors, but getExtLoad
rejects subvector loads, so this could never have worked;
the patch removes it.
llvm-svn: 47696
approach taken is different to that in LegalizeDAG
when it is a question of expanding or promoting the
result type: for example, if extracting an i64 from
a <2 x i64>, when i64 needs expanding, it bitcasts
the vector to <4 x i32>, extracts the appropriate
two i32's, and uses those for the Lo and Hi parts.
Likewise, when extracting an i16 from a <4 x i16>,
and i16 needs promoting, it bitcasts the vector to
<2 x i32>, extracts the appropriate i32, twiddles
the bits if necessary, and uses that as the promoted
value. This puts more pressure on bitcast legalization,
and I've added the appropriate cases. They needed to
be added anyway since users can generate such bitcasts
too if they want to. Also, when considering various
cases (Legal, Promote, Expand, Scalarize, Split) it is
a pain that expand can correspond to Expand, Scalarize
or Split, so I've changed the LegalizeTypes enum so it
lists those different cases - now Expand only means
splitting a scalar in two.
The code produced is the same as by LegalizeDAG for
all relevant testcases, except for
2007-10-31-extractelement-i64.ll, where the code seems
to have improved (see below; can an expert please tell
me if it is better or not).
Before < vs after >.
< subl $92, %esp
< movaps %xmm0, 64(%esp)
< movaps %xmm0, (%esp)
< movl 4(%esp), %eax
< movl %eax, 28(%esp)
< movl (%esp), %eax
< movl %eax, 24(%esp)
< movq 24(%esp), %mm0
< movq %mm0, 56(%esp)
---
> subl $44, %esp
> movaps %xmm0, 16(%esp)
> pshufd $1, %xmm0, %xmm1
> movd %xmm1, 4(%esp)
> movd %xmm0, (%esp)
> movq (%esp), %mm0
> movq %mm0, 8(%esp)
< subl $92, %esp
< movaps %xmm0, 64(%esp)
< movaps %xmm0, (%esp)
< movl 12(%esp), %eax
< movl %eax, 28(%esp)
< movl 8(%esp), %eax
< movl %eax, 24(%esp)
< movq 24(%esp), %mm0
< movq %mm0, 56(%esp)
---
> subl $44, %esp
> movaps %xmm0, 16(%esp)
> pshufd $3, %xmm0, %xmm1
> movd %xmm1, 4(%esp)
> movhlps %xmm0, %xmm0
> movd %xmm0, (%esp)
> movq (%esp), %mm0
> movq %mm0, 8(%esp)
< subl $92, %esp
< movaps %xmm0, 64(%esp)
---
> subl $44, %esp
< movl 16(%esp), %eax
< movl %eax, 48(%esp)
< movl 20(%esp), %eax
< movl %eax, 52(%esp)
< movaps %xmm0, (%esp)
< movl 4(%esp), %eax
< movl %eax, 60(%esp)
< movl (%esp), %eax
< movl %eax, 56(%esp)
---
> pshufd $1, %xmm0, %xmm1
> movd %xmm1, 4(%esp)
> movd %xmm0, (%esp)
> movd %xmm1, 12(%esp)
> movd %xmm0, 8(%esp)
< subl $92, %esp
< movaps %xmm0, 64(%esp)
---
> subl $44, %esp
< movl 24(%esp), %eax
< movl %eax, 48(%esp)
< movl 28(%esp), %eax
< movl %eax, 52(%esp)
< movaps %xmm0, (%esp)
< movl 12(%esp), %eax
< movl %eax, 60(%esp)
< movl 8(%esp), %eax
< movl %eax, 56(%esp)
---
> pshufd $3, %xmm0, %xmm1
> movd %xmm1, 4(%esp)
> movhlps %xmm0, %xmm0
> movd %xmm0, (%esp)
> movd %xmm1, 12(%esp)
> movd %xmm0, 8(%esp)
llvm-svn: 47672
operand of a VECTOR_SHUFFLE. The mask is a
vector of constant integers. The code in
LegalizeDAG doesn't bother to legalize the
mask, since it's basically just storage for
a bunch of constants, however LegalizeTypes
is more picky. The problem is that there may
not exist any legal vector-of-integers type
with a legal element type, so it is impossible
to create a legal mask! Unless of course you
cheat by creating a BUILD_VECTOR where the
operands have a different type to the element
type of the vector being built... This is
pretty ugly but works - all relevant tests in
the testsuite pass, and produce the same
assembler with and without LegalizeTypes.
llvm-svn: 47670
Change several cases in SimplifyDemandedMask that don't ever do any
simplifying to reuse the logic in ComputeMaskedBits instead of
duplicating it.
llvm-svn: 47648
CodeGen/PowerPC/illegal-element-type.ll): suppose
a node X is processed, and processing maps it to
a node Y. Then X continues to exist in the DAG,
but with no users. While processing some other
node, a new node may be created that happens to
be equal to X, and thus X will be reused rather
than a truly new node. This can cause X to
"magically reappear", and since it is in the
Processed state in will not be reprocessed, so
at the end of type legalization the illegal node
X can still be present. The solution is to replace
X with Y whenever X gets resurrected like this.
llvm-svn: 47601
after legalize. Just because a constant is legal (e.g. 0.0 in SSE)
doesn't mean that its negated value is legal (-0.0). We could make
this stronger by checking to see if the negated constant is actually
legal post negation, but it doesn't seem like a big deal.
llvm-svn: 47591
out of illegal elements (BUILD_VECTOR). Uses and beefs
up BUILD_PAIR, though it didn't really have to. Like
most of LegalizeTypes, does not support soft-float.
This cures all "make check" vector building failures.
llvm-svn: 47537
inline asms.
Fix PR2078 by marking aliases of registers used when a register is
marked used. This prevents EAX from being allocated when AX is listed
in the clobber set for the asm.
llvm-svn: 47426
and splitting extract_subvector. This fixes nine
"make check" testcases, for example
2008-02-04-ExtractSubvector.ll and (partially)
CodeGen/Generic/vector.ll.
llvm-svn: 47384
AddNodeIDNode does profiling for a ConstantSDNode, but so does
SelectionDAG::getConstant. This profiling should be moved to a common
static function in ConstantSDNode.
llvm-svn: 47359
tblgen will complain if a sign-extended constant does not fit into a
data type smaller than i32, e.g., i16. This causes a problem when certain
hex constants are used, such as 0xff for byte masks or immediate xor
values.
tblgen will try the sign-extended value first and, if the sign extended
value would overflow, it tries to see if the unsigned value will fit.
Consequently, a software developer can now safely incant:
(XORHIr16 R16C:$rA, 0xffff)
which is somewhat clearer and more informative than incanting:
(XORHIr16 R16C:$rA, (i16 -1))
even if the two are bitwise equivalent.
Tblgen also outputs the 64-bit unsigned constant in the generated ISel code
when getTargetConstant() is invoked.
llvm-svn: 47188
in a ret node. These are created as i32 constants
but on some platforms i32 is not legal. This
fixes 26 "make check" failures, for example
Alpha/2005-07-12-TwoMallocCalls.ll.
llvm-svn: 47172
the return value is zero-extended if it isn't
sign-extended. It may also be any-extended.
Also, if a floating point value was returned
in a larger floating point type, pass 1 as the
second operand to FP_ROUND, which tells it
that all the precision is in the original type.
I think this is right but I could be wrong.
Finally, when doing libcalls, set isZExt on
a parameter if it is "unsigned". Currently
isSExt is set when signed, and nothing is
set otherwise. This should be right for all
calls to standard library routines.
llvm-svn: 47122
1) ConstantFP is now expand by default
2) ConstantFP is not turned into TargetConstantFP during Legalize
if it is legal.
This allows ConstantFP to be handled like Constant, allowing for
targets that can encode FP immediates as MachineOperands.
As a bonus, fix up Itanium FP constants, which now correctly match,
and match more constants! Hooray.
llvm-svn: 47121
CTTZ and CTPOP. The expansion code differs from
that in LegalizeDAG in that it chooses to take the
CTLZ/CTTZ count from the Hi/Lo part depending on
whether the Hi/Lo value is zero, not on whether
CTLZ/CTTZ of Hi/Lo returned 32 (or whatever the
width of the type is) for it. I made this change
because the optimizers may well know that Hi/Lo
is zero and exploit it. The promotion code for
CTTZ also differs from that in LegalizeDAG: it
uses an "or" to get the right result when the
original value is zero, rather than using a compare
and select. This also means the value doesn't
need to be zero extended.
llvm-svn: 47075
node as soon as we create it in SDISel. Previously we would lower it in
legalize. The problem with this is that it only exposes the argument
loads implied by FORMAL_ARGUMENTs after legalize, so that only dag combine 2
can hack on them. This causes us to miss some optimizations because
datatype expansion also happens here.
Exposing the loads early allows us to do optimizations on them. For example
we now compile arg-cast.ll to:
_foo:
movl $2147483647, %eax
andl 8(%esp), %eax
ret
where we previously produced:
_foo:
subl $12, %esp
movsd 16(%esp), %xmm0
movsd %xmm0, (%esp)
movl $2147483647, %eax
andl 4(%esp), %eax
addl $12, %esp
ret
It might also make sense to do this for ISD::CALL nodes, which have implicit
stores on many targets.
llvm-svn: 47054
handle arbitrary precision integers and any number
of parts. For example, on a 32 bit machine an i50
corresponds to two i32 parts. getCopyToParts will
extend the i50 to an i64 then write half of the i64
to each part; getCopyFromParts will combine the two
i32 parts into an i64 then truncate the result to
i50.
llvm-svn: 47024
Added member template "Add" to FoldingSetNodeID that allows "adding" arbitrary
objects to a profile via dispatch to FoldingSetTrait<T>::Profile().
Removed FoldingSetNodeID::AddAPFloat and FoldingSetNodeID::APInt, as their
functionality is now replaced using the above mentioned member template.
llvm-svn: 46957
initializer problem, a minor tweak to the way the
DAGISelEmitter finds load/store nodes, and a renaming of the
new PseudoSourceValue objects.
llvm-svn: 46827
ReadyToProcess node - add an assertion to check
this. Add an assertion to NodeDeleted that checks
that processed/ready nodes are indeed not deleted.
It is because they are never deleted that none of
the maps can have a deleted node as the source of
a mapping. It does however seem to be possible in
theory to have a deleted value as the target of a
mapping, however this has not yet been spotted in
the wild. Still mulling on what to do about this.
[The theoretical situation is this: a node A is
expanded/promoted/whatever to a newly created node
B. Thus A->B is added to a map. When the subtree
rooted at B is legalized it is conceivable that B
is deleted due to RAUW on a node somewhere above
it].
llvm-svn: 46705
keep the LegalizeTypes node flags up to date when doing a RAUW.
This fixes a nasty bug that Duncan ran into and makes the
previous (nonbuggy case) more efficent.
llvm-svn: 46679
DAGUpdateListener object pointer instead of just returning a vector
of deleted nodes. This makes the interfaces more efficient (no more
allocating a vector [at least a malloc], filling it in, then walking
it) and more clean. This also allows the client to be notified of
nodes that are *changed* but not deleted.
llvm-svn: 46677
SelectionDAG::ReplaceAllUsesWith to handle replacement of
an SDOperand with *any* sdoperand, not just one for a node with
a single result. Note that this has a horrible FIXME'd hack in it
to work around PR1975. This should be removed when PR1975 is fixed.
llvm-svn: 46674
Added ISD::DECLARE node type to represent llvm.dbg.declare intrinsic. Now the intrinsic calls are lowered into a SDNode and lives on through out the codegen passes.
For now, since all the debugging information recording is done at isel time, when a ISD::DECLARE node is selected, it has the side effect of also recording the variable. This is a short term solution that should be fixed in time.
llvm-svn: 46659
in the backend. Introduce a new SDNode type, MemOperandSDNode, for
holding a MemOperand in the SelectionDAG IR, and add a MemOperand
list to MachineInstr, and code to manage them. Remove the offset
field from SrcValueSDNode; uses of SrcValueSDNode that were using
it are all all using MemOperandSDNode now.
Also, begin updating some getLoad and getStore calls to use the
PseudoSourceValue objects.
Most of this was written by Florian Brander, some
reorganization and updating to TOT by me.
llvm-svn: 46585
Note this solution might be somewhat fragile since ISD::LABEL may be used for other
purposes. If that ends up to be an issue, we may need to introduce a different node
for debug labels.
llvm-svn: 46571
and StoreSDNode into their common base class LSBaseSDNode. Member
functions getLoadedVT and getStoredVT are replaced with the common
getMemoryVT to simplify code that will handle both loads and stores.
llvm-svn: 46538
type that matters but the operand type. This fixes
2008-01-08-IllegalCMP.ll which crashed with the new
legalize infrastructure because SETCC with result
type i8 and operand type i64 was being custom expanded
by the X86 backend. With this fix, the gcc build gets
as far as the first libcall.
llvm-svn: 46525
registers if used by a bitconvert or using a bitconvert. This allows us to
avoid constant pool loads and use cheaper integer instructions when the
values come from or end up in integer regs anyway. For example, we now
compile CodeGen/X86/fp-in-intregs.ll to:
_test1:
movl $2147483648, %eax
xorl 4(%esp), %eax
ret
_test2:
movl $1065353216, %eax
orl 4(%esp), %eax
andl $3212836864, %eax
ret
Instead of:
_test1:
movss 4(%esp), %xmm0
xorps LCPI2_0, %xmm0
movd %xmm0, %eax
ret
_test2:
movss 4(%esp), %xmm0
andps LCPI3_0, %xmm0
movss LCPI3_1, %xmm1
andps LCPI3_2, %xmm1
orps %xmm0, %xmm1
movd %xmm1, %eax
ret
bitconverts can happen due to various calling conventions that require
fp values to passed in integer regs in some cases, e.g. when returning
a complex.
llvm-svn: 46414
delete a node even if it was not dead in some cases. Instead, just add it to
the worklist. Also, make sure to use the CombineTo methods, as it was doing
things that were unsafe: the top level combine loop could touch dangling memory.
This fixes CodeGen/Generic/2008-01-25-dag-combine-mul.ll
llvm-svn: 46384
1. we already know the value is dead, so don't bother replacing
it with undef.
2. The very case the comment describes actually makes the load
live which asserts in deletenode. If we do the replacement
and the node becomes live, just treat it as new. This fixes
a failure on X86/2008-01-16-InvalidDAGCombineXform.ll with
some local changes in my tree.
llvm-svn: 46306
dead stuff around. This gets fed into the isel pass and causes certain foldings from
happening because nodes have extraneous uses floating around. For example, if we turned
foo(bar(x)) -> baz(x), we sometimes left bar(x) around.
llvm-svn: 46305
Dan Gohman