as the very last thing before node emission. This should
dramatically reduce the number of times we do 'MatchAddress'
on X86, speeding up compile time. This also improves comments
in the tables and shrinks the table a bit, now down to
80506 bytes for x86.
llvm-svn: 97703
CSE and recursive RAUW calls delete a node from the use list,
invalidating the use list iterator. There's currently no known
way to reproduce this in an unmodified LLVM, however there's no
fundamental reason why a SelectionDAG couldn't be formed which
would trigger this case.
llvm-svn: 97665
entry we're about to process is obviously going to fail, don't
bother pushing a scope only to have it immediately be popped.
This avoids a lot of scope stack traffic in common cases.
Unfortunately, this requires duplicating some of the predicate
dispatch. To avoid duplicating the actual logic I pulled each
predicate out to its own static function which gets used in
both places.
llvm-svn: 97651
SwitchOpcodeMatcher) and have DAGISelMatcherOpt form it. This
speeds up selection, particularly for X86 which has lots of
variants of instructions with only type differences.
llvm-svn: 97645
long test(long x) { return (x & 123124) | 3; }
Currently compiles to:
_test:
orl $3, %edi
movq %rdi, %rax
andq $123127, %rax
ret
This is because instruction and DAG combiners canonicalize
(or (and x, C), D) -> (and (or, D), (C | D))
However, this is only profitable if (C & D) != 0. It gets in the way of the
3-addressification because the input bits are known to be zero.
llvm-svn: 97616
CopyToReg/CopyFromReg/INLINEASM. These are annoying because
they have the same opcode before an after isel. Fix this by
setting their NodeID to -1 to indicate that they are selected,
just like what automatically happens when selecting things that
end up being machine nodes.
With that done, give IsLegalToFold a new flag that causes it to
ignore chains. This lets the HandleMergeInputChains routine be
the one place that validates chains after a match is successful,
enabling the new hotness in chain processing. This smarter
chain processing eliminates the need for "PreprocessRMW" in the
X86 and MSP430 backends and enables MSP to start matching it's
multiple mem operand instructions more aggressively.
I currently #if out the dead code in the X86 backend and MSP
backend, I'll remove it for real in a follow-on patch.
The testcase changes are:
test/CodeGen/X86/sse3.ll: we generate better code
test/CodeGen/X86/store_op_load_fold2.ll: PreprocessRMW was
miscompiling this before, we now generate correct code
Convert it to filecheck while I'm at it.
test/CodeGen/MSP430/Inst16mm.ll: Add a testcase for mem/mem
folding to make anton happy. :)
llvm-svn: 97596
was that we weren't properly handling the case when interior
nodes of a matched pattern become dead after updating chain
and flag uses. Now we handle this explicitly in
UpdateChainsAndFlags.
llvm-svn: 97561
DoInstructionSelection. Inline "SelectRoot" into it from DAGISelHeader.
Sink some other stuff out of DAGISelHeader into SDISel.
Eliminate the various 'Indent' stuff from various targets, which dates
to when isel was recursive.
17 files changed, 114 insertions(+), 430 deletions(-)
llvm-svn: 97555
stuff now that we don't care about emulating the old broken
behavior of the old isel. This eliminates the
'CheckChainCompatible' check (along with IsChainCompatible) which
did an incorrect and inefficient scan *up* the chain nodes which
happened as the pattern was being formed and does the validation
at the end in HandleMergeInputChains when it forms a structural
pattern. This scans "down" the graph, which means that it is
quickly bounded by nodes already selected. This also handles
token factors that get "trapped" in the dag.
Removing the CheckChainCompatible nodes also shrinks the
generated tables by about 6K for X86 (down to 83K).
There are two pieces remaining before I can nuke PreprocessRMW:
1. I xfailed a test because we're now producing worse code in a
case that has nothing to do with the change: it turns out that
our use of MorphNodeTo will leave dead nodes in the graph
which (depending on how the graph is walked) end up causing
bogus uses of chains and blocking matches. This is really
bad for other reasons, so I'll fix this in a follow-up patch.
2. CheckFoldableChainNode needs to be improved to handle the TF.
llvm-svn: 97539
ComplexPattern at the root be generated multiple times, once
for each opcode they are part of. This encourages factoring
because the opcode checks get treated just like everything
else in the matcher.
llvm-svn: 97439
to a scope where every child starts with a CheckOpcode, but
executes more efficiently. Enhance DAGISelMatcherOpt to
form it.
This also fixes a bug in CheckOpcode: apparently the SDNodeInfo
objects are not pointer comparable, we have to compare the
enum name.
llvm-svn: 97438
terms of store and load, which means bitcasting between scalar
integer and vector has endian-specific results, which undermines
this whole approach.
llvm-svn: 97137
the number of value bits, not the number of bits of allocation for in-memory
storage.
Make getTypeStoreSize and getTypeAllocSize work consistently for arrays and
vectors.
Fix several places in CodeGen which compute offsets into in-memory vectors
to use TargetData information.
This fixes PR1784.
llvm-svn: 97064
necessary to swap the operands to handle NaN and negative zero properly.
Also, reintroduce logic for checking for NaN conditions when forming
SSE min and max instructions, fixed to take into consideration NaNs and
negative zeros. This allows forming min and max instructions in more
cases.
llvm-svn: 97025
to adding them in a determinstic order (bottom up from
the root) based on the structure of the graph itself.
This updates tests for some random changes, interesting
bits: CodeGen/Blackfin/promote-logic.ll no longer crashes.
I have no idea why, but that's good right?
CodeGen/X86/2009-07-16-LoadFoldingBug.ll also fails, but
now compiles to have one fewer constant pool entry, making
the expected load that was being folded disappear. Since it
is an unreduced mass of gnast, I just removed it.
This fixes PR6370
llvm-svn: 97023
no id's would cause early exit allowing IsLegalToFold to return true
instead of false, producing a cyclic dag.
This was striking the new isel because it isn't using SelectNodeTo yet,
which theoretically is just an optimization.
llvm-svn: 96972
dragonegg self-host build. I reverted 96640 in order to revert
96556 (96640 goes on top of 96556), but it also looks like with
both of them applied the breakage happens even earlier. The
symptom of the 96556 miscompile is the following crash:
llvm[3]: Compiling AlphaISelLowering.cpp for Release build
cc1plus: /home/duncan/tmp/tmp/llvm/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:4982: void llvm::SelectionDAG::ReplaceAllUsesWith(llvm::SDNode*, llvm::SDNode*, llvm::SelectionDAG::DAGUpdateListener*): Assertion `(!From->hasAnyUseOfValue(i) || From->getValueType(i) == To->getValueType(i)) && "Cannot use this version of ReplaceAllUsesWith!"' failed.
Stack dump:
0. Running pass 'X86 DAG->DAG Instruction Selection' on function '@_ZN4llvm19AlphaTargetLowering14LowerOperationENS_7SDValueERNS_12SelectionDAGE'
g++: Internal error: Aborted (program cc1plus)
This occurs when building LLVM using LLVM built by LLVM (via
dragonegg). Probably LLVM has miscompiled itself, though it
may have miscompiled GCC and/or dragonegg itself: at this point
of the self-host build, all of GCC, LLVM and dragonegg were built
using LLVM. Unfortunately this kind of thing is extremely hard
to debug, and while I did rummage around a bit I didn't find any
smoking guns, aka obviously miscompiled code.
Found by bisection.
r96556 | evancheng | 2010-02-18 03:13:50 +0100 (Thu, 18 Feb 2010) | 5 lines
Some dag combiner goodness:
Transform br (xor (x, y)) -> br (x != y)
Transform br (xor (xor (x,y), 1)) -> br (x == y)
Also normalize (and (X, 1) == / != 1 -> (and (X, 1)) != / == 0 to match to "test on x86" and "tst on arm"
r96640 | evancheng | 2010-02-19 01:34:39 +0100 (Fri, 19 Feb 2010) | 16 lines
Transform (xor (setcc), (setcc)) == / != 1 to
(xor (setcc), (setcc)) != / == 1.
e.g. On x86_64
%0 = icmp eq i32 %x, 0
%1 = icmp eq i32 %y, 0
%2 = xor i1 %1, %0
br i1 %2, label %bb, label %return
=>
testl %edi, %edi
sete %al
testl %esi, %esi
sete %cl
cmpb %al, %cl
je LBB1_2
llvm-svn: 96672
IsLegalToFold and IsProfitableToFold. The generic version of the later simply checks whether the folding candidate has a single use.
This allows the target isel routines more flexibility in deciding whether folding makes sense. The specific case we are interested in is folding constant pool loads with multiple uses.
llvm-svn: 96255
lowering and requires that certain types exist in ValueTypes.h. Modified widening to
check if an op can trap and if so, the widening algorithm will apply only the op on
the defined elements. It is safer to do this in widening because the optimizer can't
guarantee removing unused ops in some cases.
llvm-svn: 95823
into TargetOpcodes.h. #include the new TargetOpcodes.h
into MachineInstr. Add new inline accessors (like isPHI())
to MachineInstr, and start using them throughout the
codebase.
llvm-svn: 95687
type is the same as the element type of the vector. EXTRACT_VECTOR_ELT can
be used to extended the width of an integer type. This fixes a bug for
Generic/vector-casts.ll on a ppc750.
llvm-svn: 94990
"visit*" method is called, take the newly created nodes, walk them in a DFS
fashion, and if they don't have an ordering set, then give it one.
llvm-svn: 94757
This allows code gen and the exception table writer to cooperate to make sure
landing pads are associated with the correct invoke locations.
llvm-svn: 94726
Target independent isel should always pass along the "tail call" property. Change
target hook LowerCall's parameter "isTailCall" into a refernce. If the target
decides it's impossible to honor the tail call request, it should set isTailCall
to false to make target independent isel happy.
llvm-svn: 94626
which is more convenient, and change getPICJumpTableRelocBaseExpr
to take a MachineFunction to match.
Next, move the X86 code that create a PICBase symbol to
X86TargetLowering::getPICBaseSymbol from
X86MCInstLower::GetPICBaseSymbol, which was an asmprinter specific
library. This eliminates a 'gross hack', and allows us to
implement X86ISelLowering::getPICJumpTableRelocBaseExpr which now
calls it.
This in turn allows us to eliminate the
X86AsmPrinter::printPICJumpTableSetLabel method, which was the
only overload of printPICJumpTableSetLabel.
llvm-svn: 94526
1. MachineJumpTableInfo is now created lazily for a function the first time
it actually makes a jump table instead of for every function.
2. The encoding of jump table entries is now described by the
MachineJumpTableInfo::JTEntryKind enum. This enum is determined by the
TLI::getJumpTableEncoding() hook, instead of by lots of code scattered
throughout the compiler that "knows" that jump table entries are always
32-bits in pic mode (for example).
3. The size and alignment of jump table entries is now calculated based on
their kind, instead of at machinefunction creation time.
Future work includes using the EntryKind in more places in the compiler,
eliminating other logic that "knows" the layout of jump tables in various
situations.
llvm-svn: 94470
the '-pre-RA-sched' flag. It actually makes more sense to do it this way. Also,
keep track of the SDNode ordering by default. Eventually, we would like to make
this ordering a way to break a "tie" in the scheduler. However, doing that now
breaks the "CodeGen/X86/abi-isel.ll" test for 32-bit Linux.
llvm-svn: 94308
missing ones are libsupport, libsystem and libvmcore. libvmcore is
currently blocked on bugpoint, which uses EH. Once it stops using
EH, we can switch it off.
This #if 0's out 3 unit tests, because gtest requires RTTI information.
Suggestions welcome on how to fix this.
llvm-svn: 94164
comments (fast isel, X86). This doesn't seem
to break any functionality, but will introduce
cases where -g affects the generated code. I'll
be fixing that.
llvm-svn: 93811
Instcombine does this but apparently there are situations where this pattern will escape the optimizer and / or created by isel. Here is a case that's seen in JavaScriptCore:
%t1 = sub i32 0, %a
%t2 = add i32 %t1, -1
The dag combiner pattern: ((c1-A)+c2) -> (c1+c2)-A
will fold it to -1 - %a.
llvm-svn: 93773
print/dumpWithDepth allows one to dump a DAG up to N levels deep.
dump/printWithFullDepth prints the whole DAG, subject to a depth limit
on 100 in the default case (to prevent infinite recursion).
Have CannotYetSelect to a dumpWithFullDepth so it is clearer exactly
what the non-matching DAG looks like.
llvm-svn: 93538
This patch also cleans up code that expects there to be a bitcast in the first argument and testcases that call llvm.dbg.declare.
It also strips old llvm.dbg.declare intrinsics that did not pass metadata as the first argument.
llvm-svn: 93531
really does need to be a vector type, because
TargetLowering::getOperationAction for SIGN_EXTEND_INREG uses that type,
and it needs to be able to distinguish between vectors and scalars.
Also, fix some more issues with legalization of vector casts.
llvm-svn: 93043
When folding a and(any_ext(load)) both the any_ext and the
load have to have only a single use.
This removes the anyext-uses.ll testcase which started failing
because it is unreduced and unclear what it is testing.
llvm-svn: 92950
(OP (trunc x), (trunc y)) -> (trunc (OP x, y))
Unfortunately this simple change causes dag combine to infinite looping. The problem is the shrink demanded ops optimization tend to canonicalize expressions in the opposite manner. That is badness. This patch disable those optimizations in dag combine but instead it is done as a late pass in sdisel.
This also exposes some deficiencies in dag combine and x86 setcc / brcond lowering. Teach them to look pass ISD::TRUNCATE in various places.
llvm-svn: 92849
clear what information these functions are actually using.
This is also a micro-optimization, as passing a SDNode * around is
simpler than passing a { SDNode *, int } by value or reference.
llvm-svn: 92564
(X != null) | (Y != null) --> (X|Y) != 0
(X == null) & (Y == null) --> (X|Y) == 0
so that instcombine can stop doing this for pointers. This is part of PR3351,
which is a case where instcombine doing this for pointers (inserting ptrtoint)
is pessimizing code.
llvm-svn: 92406
multiply sequence when the power is a constant integer. Before, our
codegen for std::pow(.., int) always turned into a libcall, which was
really inefficient.
This should also make many gfortran programs happier I'd imagine.
llvm-svn: 92388
getMDKindID/getMDKindNames methods to LLVMContext (and add
convenience methods to Module), eliminating MetadataContext.
Move the state that it maintains out to LLVMContext.
llvm-svn: 92259
I asked Devang to do back on Sep 27. Instead of going through the
MetadataContext class with methods like getMD() and getMDs(), just
ask the instruction directly for its metadata with getMetadata()
and getAllMetadata().
This includes a variety of other fixes and improvements: previously
all Value*'s were bloated because the HasMetadata bit was thrown into
value, adding a 9th bit to a byte. Now this is properly sunk down to
the Instruction class (the only place where it makes sense) and it
will be folded away somewhere soon.
This also fixes some confusion in getMDs and its clients about
whether the returned list is indexed by the MDID or densely packed.
This is now returned sorted and densely packed and the comments make
this clear.
This introduces a number of fixme's which I'll follow up on.
llvm-svn: 92235
compare. On other targets we end up with a call to memcmp because we don't
want 16 individual byte loads. We should be able to use movups as well, but
we're failing to select the generated icmp.
llvm-svn: 92107
SDISel. This optimization was causing simplifylibcalls to
introduce type-unsafe nastiness. This is the first step, I'll be
expanding the memcmp optimizations shortly, covering things that
we really really wouldn't want simplifylibcalls to do.
llvm-svn: 92098
return partial registers. This affected the back-end lowering code some.
Also patch up some places I missed before in the "get" functions.
llvm-svn: 91880
- Move DisableScheduling flag into TargetOption.h
- Move SDNodeOrdering into its own header file. Give it a minimal interface that
doesn't conflate construction with storage.
- Move assigning the ordering into the SelectionDAGBuilder.
This isn't used yet, so there should be no functional changes.
llvm-svn: 91727
The change in SelectionDAGBuilder is needed to allow using bitcasts to convert
between f64 (the default type for ARM "d" registers) and 64-bit Neon vector
types. Radar 7457110.
llvm-svn: 91649
LegalizeDAG.cpp. Unlike the code it replaces, which simply decrements the simple
type by one, getHalfSizedIntegerVT() searches for the smallest simple integer
type that is at least half the size of the type it is called on. This approach
has the advantage that it will continue working if a new value type (such as
i24) is added to MVT.
Also, in preparation for new value types, remove the assertions that
non-power-of-2 8-bit-mutiple types are Extended when legalizing extload and
truncstore operations.
llvm-svn: 91614
Fold (zext (and x, cst)) -> (and (zext x), cst)
DAG combiner likes to optimize expression in the other way so this would end up cause an infinite looping.
llvm-svn: 91574
1. Only perform (zext (shl (zext x), y)) -> (shl (zext x), y) when y is a constant. This makes sure it remove at least one zest.
2. If the shift is a left shift, make sure the original shift cannot shift out bits.
llvm-svn: 91399
stuff isn't used just yet.
We want to model the GCC `-fno-schedule-insns' and `-fno-schedule-insns2'
flags. The hypothesis is that the people who use these flags know what they are
doing, and have hand-optimized the C code to reduce latencies and other
conflicts.
The idea behind our scheme to turn off scheduling is to create a map "on the
side" during DAG generation. It will order the nodes by how they appeared in the
code. This map is then used during scheduling to get the ordering.
llvm-svn: 91392
Pointed out by Javier Martinez (who also provided a patch). Since
this logic is not used on (for example) x86, I guess nobody noticed.
Tested by generating SHL, SRL, SRA on various choices of i64 for all
possible shift amounts, and comparing with gcc. Since I did this on
x86-32, I had to force the use of ExpandShiftWithUnknownAmountBit.
What I'm saying here is that I don't have a testcase I can add to the
repository.
llvm-svn: 90482
Note that "hasDotLocAndDotFile"-style debug info was already broken;
people wanting this functionality should implement it in the
AsmPrinter/DwarfWriter code.
llvm-svn: 89711
Dan Gohman