Commit Graph

847 Commits

Author SHA1 Message Date
Andrew Lenharth 6606f17e50 Use ADDC if it is valid at any smaller size. fixes test/Codegen/Generic/i128-addsub.ll on x86
llvm-svn: 57247
2008-10-07 17:03:15 +00:00
Andrew Lenharth 3a9be150be Expand arith on machines without carry flags
llvm-svn: 57243
2008-10-07 14:15:42 +00:00
Chris Lattner 2416896b3c wrap some long lines and expand i32 mul's to libcalls, inspired by a
patch by Mikael Lepisto!

llvm-svn: 57077
2008-10-04 21:27:46 +00:00
Dale Johannesen 5d60c1ebb1 Pass MemOperand through for 64-bit atomics on 32-bit,
incidentally making the case where the memop is a
pointer deref work.  Fix cmp-and-swap regression.

llvm-svn: 57027
2008-10-03 19:41:08 +00:00
Dale Johannesen 867d549fce Handle some 64-bit atomics on x86-32, some of the time.
llvm-svn: 56963
2008-10-02 18:53:47 +00:00
Dan Gohman 3a293e7404 Fix typos in comments.
llvm-svn: 56919
2008-10-01 15:07:49 +00:00
Dan Gohman 86aa16a69a Optimize SelectionDAG's AssignTopologicalOrder even further.
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
2008-09-30 18:30:35 +00:00
Dale Johannesen f61a84ec43 Remove misuse of ReplaceNodeResults for atomics with
valid types.  No functional change.

llvm-svn: 56808
2008-09-29 22:25:26 +00:00
Dale Johannesen 0e32a2c935 Add "inreg" field to CallSDNode (doesn't increase
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
2008-09-26 19:31:26 +00:00
Richard Pennington 4b35e64504 bug 2812: Segmentation fault on a big emdiam processor.
llvm-svn: 56609
2008-09-25 16:15:10 +00:00
Dan Gohman e2947e1e07 Fix the alignment of loads from constant pool entries when the
load address has an offset from the base of the constant pool
entry.

llvm-svn: 56479
2008-09-22 22:40:08 +00:00
Dan Gohman 64d6c6fe30 Change SelectionDAG::getConstantPool to always set the alignment of the
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
2008-09-16 22:05:41 +00:00
Bill Wendling 24c79f28b1 Reverting r56249. On further investigation, this functionality isn't needed.
Apologies for the thrashing.

llvm-svn: 56251
2008-09-16 21:48:12 +00:00
Bill Wendling 8bc392fb1d - Change "ExternalSymbolSDNode" to "SymbolSDNode".
- 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
2008-09-16 21:12:30 +00:00
Dan Gohman ec270fb640 Change ConstantSDNode and ConstantFPSDNode to use ConstantInt* and
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
2008-09-12 18:08:03 +00:00
Dan Gohman effb894453 Rename ConstantSDNode::getValue to getZExtValue, for consistency
with ConstantInt. This led to fixing a bug in TargetLowering.cpp
using getValue instead of getAPIntValue.

llvm-svn: 56159
2008-09-12 16:56:44 +00:00
Dale Johannesen baf6762e26 The sequence for ppcf128 compares was not IEEE
safe in the presence of NaNs.

llvm-svn: 56136
2008-09-12 00:30:56 +00:00
Evan Cheng 0fff397a13 A few more places where FPOW is being ignored.
llvm-svn: 56032
2008-09-09 23:35:53 +00:00
Evan Cheng f4e5de4583 Legalizer was missing code that expand fpow to a libcall.
llvm-svn: 56028
2008-09-09 23:02:14 +00:00
Dale Johannesen da2d80688b Add intrinsics for log, log2, log10, exp, exp2.
No functional change (and no FE change to generate them).

llvm-svn: 55753
2008-09-04 00:47:13 +00:00
Gabor Greif f304a7aa4d erect abstraction boundaries for accessing SDValue members, rename Val -> Node to reflect semantics
llvm-svn: 55504
2008-08-28 21:40:38 +00:00
Dale Johannesen 41be0d4445 Split the ATOMIC NodeType's to include the size, e.g.
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
2008-08-28 02:44:49 +00:00
Gabor Greif abfdf928d8 disallow direct access to SDValue::ResNo, provide a getter instead
llvm-svn: 55394
2008-08-26 22:36:50 +00:00
Dan Gohman d56f73f2f2 Optimize SelectionDAG's topological sort to use one pass instead
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
2008-08-26 21:42:18 +00:00
Dan Gohman 2af34bd309 Add libcalls for the new rounding opcodes.
llvm-svn: 55133
2008-08-21 18:38:14 +00:00
Dan Gohman c6337ac069 Add libm-oriented ISD opcodes for rounding operations.
llvm-svn: 55130
2008-08-21 17:55:02 +00:00
Dan Gohman 550c9af91f Improve support for vector casts in LLVM IR and CodeGen.
llvm-svn: 54784
2008-08-14 20:04:46 +00:00
Nate Begeman 82f1925708 Fix broken CellSPU lowering, re-instate braces in Legalize
llvm-svn: 54168
2008-07-29 19:07:27 +00:00
Nate Begeman d63495ff25 Disable a fix in the previous patch, since it breaks CellSPU.
The CellSPU codegen is broken, but needs to be fixed before we can
put this back in.

llvm-svn: 54164
2008-07-29 18:28:31 +00:00
Nate Begeman fecbc8cff1 Add vector shifts to the IR, patch by Eli Friedman.
CodeGen & Clang work coming next.

llvm-svn: 54161
2008-07-29 15:49:41 +00:00
Dan Gohman 2ce6f2ad5e Rename SDOperand to SDValue.
llvm-svn: 54128
2008-07-27 21:46:04 +00:00
Dan Gohman 91e5dcb680 Tidy SDNode::use_iterator, and complete the transition to have it
parallel its analogue, Value::value_use_iterator. The operator* method
now returns the user, rather than the use.

llvm-svn: 54127
2008-07-27 20:43:25 +00:00
Mon P Wang 7334350d31 When splitting a vector shuffle, fixed which type we used for the hi part
llvm-svn: 54007
2008-07-25 01:30:26 +00:00
Dan Gohman 581cc87f57 Add titles to the various SelectionDAG viewGraph calls
that include useful information like the name of the
block being viewed and the current phase of compilation.

llvm-svn: 53872
2008-07-21 20:00:07 +00:00
Duncan Sands b0e3938651 Add VerifyNode, a place to put sanity checks on
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
2008-07-21 10:20:31 +00:00
Duncan Sands 77a3d05f1e Factorize some code for determining which libcall to use.
llvm-svn: 53713
2008-07-17 02:36:29 +00:00
Mon P Wang 97432f4f1b Fixed potential bug if the source and target of a bit convert have different alignment
llvm-svn: 53590
2008-07-15 05:28:34 +00:00
Dan Gohman 02c7c6cb33 Include a frame index in the "fixed stack" pseudo source value
instead of using the frame index for the SVOffset, which was
inconsistent.

llvm-svn: 53486
2008-07-11 22:44:52 +00:00
Chris Lattner 87909d0629 Fix a bug in the soft-float handling of FCOPYSIGN that Duncan noticed
when working on legalizetypes.  Both legalizetypes and legalizeops now
produce hte same code for CodeGen/ARM/fcopysign.ll.

llvm-svn: 53435
2008-07-10 23:46:13 +00:00
Duncan Sands 5e6d1402c2 Add a mysteriously missing libcall, FPTOSINT_F80_I32.
Be nice to 16 bit machines by supporting FP_TO_XINT
expansion for these.

llvm-svn: 53407
2008-07-10 15:33:02 +00:00
Evan Cheng 34ef1db87c Do not CSE DEBUG_LOC, DBG_LABEL, DBG_STOPPOINT, DECLARE, and EH_LABEL SDNode's. This improves compile time slightly at -O0 -g.
llvm-svn: 53246
2008-07-08 20:06:39 +00:00
Dan Gohman 56e3f63ec5 Add explicit keywords.
llvm-svn: 53179
2008-07-07 18:00:37 +00:00
Dan Gohman 38740a98b2 Make DenseMap's insert return a pair, to more closely resemble std::map.
llvm-svn: 53177
2008-07-07 17:46:23 +00:00
Evan Cheng d8b83e1292 LegalizeSetCCOperands should legalize the result of ExpandLibCall. Patch by Richard Osborne.
llvm-svn: 53169
2008-07-07 07:18:09 +00:00
Mon P Wang 5c755ff51b Fixed generating incorrect aligned stores that I backout of r53031
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
2008-07-05 20:40:31 +00:00
Duncan Sands 93e180342a Rather than having a different custom legalization
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
2008-07-04 11:47:58 +00:00
Evan Cheng fad8be450d Backed out 53031.
llvm-svn: 53110
2008-07-03 18:20:14 +00:00
Duncan Sands 739a0548c4 Add a new getMergeValues method that does not need
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
2008-07-02 17:40:58 +00:00
Mon P Wang 4b7c1acf26 Fixed problem 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: 53031
2008-07-02 17:07:12 +00:00
Evan Cheng 4c609abd90 Eliminate a compile time warning.
llvm-svn: 52982
2008-07-01 21:35:46 +00:00
Dan Gohman fb19f9402b Split ISD::LABEL into ISD::DBG_LABEL and ISD::EH_LABEL, eliminating
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
2008-07-01 00:05:16 +00:00
Dan Gohman 5c73a886b4 Rename ISD::LOCATION to ISD::DBG_STOPPOINT to better reflect its
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
2008-06-30 20:59:49 +00:00
Duncan Sands 1ae6ef83ee Revert the SelectionDAG optimization that makes
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
2008-06-30 10:19:09 +00:00
Chris Lattner 9d3740ed1c Implement split and scalarize for SELECT_CC, fixing PR2504
llvm-svn: 52887
2008-06-30 02:43:01 +00:00
Dan Gohman aa01afd47c Remove the OrigVT member from AtomicSDNode, as it is redundant with
the base SDNode's VTList.

llvm-svn: 52722
2008-06-25 16:07:49 +00:00
Mon P Wang 6a490371c9 Added MemOperands to Atomic operations since Atomics touches memory.
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
2008-06-25 08:15:39 +00:00
Duncan Sands 5fb92e58de Make custom lowering of ADD work correctly. This
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
2008-06-22 09:42:16 +00:00
Dan Gohman 3792c470d5 Clean up some uses of std::distance, now that we have allnodes_size.
llvm-svn: 52545
2008-06-20 17:15:19 +00:00
Evan Cheng be0429c558 ISD::UNDEF should be expanded recursively / iteratively.
llvm-svn: 52508
2008-06-19 22:01:11 +00:00
Andrew Lenharth f88d50bfcc add missing atomic intrinsic from gcc
llvm-svn: 52270
2008-06-14 05:48:15 +00:00
Duncan Sands 8651e9c584 Disable some DAG combiner optimizations that may be
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
2008-06-13 19:07:40 +00:00
Duncan Sands 11dd424539 Remove comparison methods for MVT. The main cause
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
2008-06-08 20:54:56 +00:00
Duncan Sands 13237ac3b9 Wrap MVT::ValueType in a struct to get type safety
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
2008-06-06 12:08:01 +00:00
Scott Michel a7d8649f78 Fix spellnig error
llvm-svn: 51917
2008-06-03 19:13:20 +00:00
Scott Michel d831cc49e5 Add necessary 64-bit support so that gcc frontend compiles (mostly). Current
issue is operand promotion for setcc/select... but looks like the fundamental
stuff is implemented for CellSPU.

llvm-svn: 51884
2008-06-02 22:18:03 +00:00
Dan Gohman 8807147ada Remove an unused variable.
llvm-svn: 51721
2008-05-30 00:56:36 +00:00
Evan Cheng 9ac3631fa3 If the result of a BIT_CONVERT is a v1* vector, it doesn't mean its source is a v1* vector.
llvm-svn: 51192
2008-05-16 17:19:05 +00:00
Nate Begeman f79f52282c Actually scalarize the operand to BIT_CONVERT instead of asking someone to do
something with a v1 type.

llvm-svn: 51160
2008-05-15 20:40:58 +00:00
Dan Gohman fd3e3003f3 Whitespace cleanups.
llvm-svn: 51089
2008-05-14 00:43:10 +00:00
Nate Begeman b87e63a730 Teach Legalize how to scalarize VSETCC
Teach X86 a few more vsetcc patterns.  Custom lowering for unsupported ones is next.

llvm-svn: 51009
2008-05-12 23:09:43 +00:00
Nate Begeman cfcb56091b Add support for vicmp/vfcmp codegen, more legalize support coming.
This is necessary to unbreak the build.

llvm-svn: 50988
2008-05-12 19:40:03 +00:00
Dan Gohman ecb77385ab Fix a missing break in the ISD::FLT_ROUNDS_ handling. Patch by giuma!
llvm-svn: 50967
2008-05-12 16:07:15 +00:00
Mon P Wang 3e58393c3d Added addition atomic instrinsics and, or, xor, min, and max.
llvm-svn: 50663
2008-05-05 19:05:59 +00:00
Scott Michel be940424b3 Fix custom target lowering for zero/any/sign_extend: make sure that
DAG.UpdateNodeOperands() is called before (not after) the call to
TLI.LowerOperation().

llvm-svn: 50461
2008-04-30 00:26:38 +00:00
Nate Begeman 6f94f61317 Pull the code to perform an INSERT_VECTOR_ELT in memory out into its own
function, and then use it to fix a bug in SplitVectorOp that expected inserts
to always have constant insertion indices.

llvm-svn: 50273
2008-04-25 18:07:40 +00:00
Chris Lattner 3b18762f40 Switch to using Simplified ConstantFP::get API.
llvm-svn: 49977
2008-04-20 00:41:09 +00:00
Dan Gohman 9752a8f3b4 Correct the SrcValue information in the Expand code for va_copy.
llvm-svn: 49839
2008-04-17 02:09:26 +00:00
Roman Levenstein a3ee1a38a3 Ongoing work on improving the instruction selection infrastructure:
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
2008-04-16 16:15:27 +00:00
Duncan Sands 844d55a42a Factor some libcall code.
llvm-svn: 49583
2008-04-12 17:14:18 +00:00
Dan Gohman 544ab2c50b Drop ISD::MEMSET, ISD::MEMMOVE, and ISD::MEMCPY, which are not Legal
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
2008-04-12 04:36:06 +00:00
Roman Levenstein 51f532f92d Re-commit of the r48822, where the infinite looping problem discovered
by Dan Gohman is fixed.

llvm-svn: 49330
2008-04-07 10:06:32 +00:00
Evan Cheng 025cea1126 Backing out 48222 temporarily.
llvm-svn: 49124
2008-04-03 03:13:16 +00:00
Evan Cheng 0bd72c5ccd More soft fp fixes.
llvm-svn: 49016
2008-04-01 02:18:22 +00:00
Evan Cheng 4cabe4b452 Pasto.
llvm-svn: 49014
2008-04-01 02:00:09 +00:00
Evan Cheng 611abc03ed Add comment.
llvm-svn: 49013
2008-04-01 01:51:26 +00:00
Evan Cheng 86e476b7cb Unbreak ARM / Thumb soft FP support.
llvm-svn: 49012
2008-04-01 01:50:16 +00:00
Roman Levenstein 358e04a185 Use a linked data structure for the uses lists of an SDNode, just like
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
2008-03-26 12:39:26 +00:00
Duncan Sands d97eea372a Introduce a new node for holding call argument
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
2008-03-21 09:14:45 +00:00
Dale Johannesen 12c76db312 Make conversions of i8/i16 to ppcf128 work.
llvm-svn: 48493
2008-03-18 17:28:38 +00:00
Nate Begeman 63eb03f800 Tabs -> spaces
Use getIntPtrConstant in a couple places to shorten stuff up
Handle splitting vector shuffles with undefs in the mask

llvm-svn: 48351
2008-03-14 00:53:31 +00:00
Dan Gohman b72127ac4c More APInt-ification.
llvm-svn: 48344
2008-03-13 22:13:53 +00:00
Dan Gohman d6819da453 Generalize ExpandIntToFP to handle the case where the operand is legal
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
2008-03-11 01:59:03 +00:00
Dan Gohman 10f7d850cf More APInt-ification.
llvm-svn: 48201
2008-03-11 00:11:06 +00:00
Dan Gohman f4300950f1 Implement more support for fp-to-i128 and i128-to-fp conversions.
llvm-svn: 48189
2008-03-10 23:03:31 +00:00
Dan Gohman 272e234477 Fix mul expansion to check the correct number of bits for
zero extension when checking if an unsigned multiply is
safe.

llvm-svn: 48171
2008-03-10 20:42:19 +00:00
Scott Michel a6729e8666 Give TargetLowering::getSetCCResultType() a parameter so that ISD::SETCC's
return ValueType can depend its operands' ValueType.

This is a cosmetic change, no functionality impacted.

llvm-svn: 48145
2008-03-10 15:42:14 +00:00
Chris Lattner 322c826c9d Fix two problems in SelectionDAGLegalize::ExpandBUILD_VECTOR's handling
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
2008-03-09 00:29:42 +00:00
Evan Cheng 95cf661534 Implement x86 support for @llvm.prefetch. It corresponds to prefetcht{0|1|2} and prefetchnta instructions.
llvm-svn: 48042
2008-03-08 00:58:38 +00:00
Dale Johannesen 8ee39c61f2 Clarify that CALLSEQ_START..END may not be nested,
and add some protection against creating such.

llvm-svn: 47957
2008-03-05 19:14:03 +00:00
Chris Lattner 3dc3899007 Improve comment, pass in the original VT so that we can shrink a long double constant
all the way to float, not stopping at double.

llvm-svn: 47937
2008-03-05 06:46:58 +00:00
Dan Gohman da7897c4e1 Codegen support for i128 UINT_TO_FP. This just fixes a
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
2008-03-05 02:07:31 +00:00
Evan Cheng 0a62cb44ce Add a target lowering hook to control whether it's worthwhile to compress fp constant.
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
2008-03-05 01:30:59 +00:00
Andrew Lenharth 357061a74d 64bit CAS on 32bit x86.
llvm-svn: 47929
2008-03-05 01:15:49 +00:00
Dan Gohman d9d874b0cd Codegen support for i128 SINT_TO_FP.
llvm-svn: 47928
2008-03-05 01:08:17 +00:00
Evan Cheng 38caf77419 Refactor ExpandConstantFP so it can optimize load from constpool of types larger than f64 into extload from smaller types.
llvm-svn: 47883
2008-03-04 08:05:30 +00:00
Dan Gohman f2bbfa3ba0 More APInt-ification.
llvm-svn: 47864
2008-03-03 22:20:46 +00:00
Andrew Lenharth d032c33300 all but CAS working on x86
llvm-svn: 47798
2008-03-01 21:52:34 +00:00
Dale Johannesen 208cc8f1b9 Add MVT::is128BitVector and is64BitVector. Shrink
unaligned load/store code using them.  Per review
of unaligned load/store vector patch.

llvm-svn: 47782
2008-03-01 03:40:57 +00:00
Dan Gohman 837a6dccd7 Use the new convertFromAPInt instead of convertFromZeroExtendedInteger,
which allows more of the surrounding arithmetic to be done with APInt
instead of uint64_t.

llvm-svn: 47745
2008-02-29 01:44:25 +00:00
Dale Johannesen c4c3de2b52 Fix an assertion message.
llvm-svn: 47722
2008-02-28 18:36:51 +00:00
Chris Lattner 9824ffef0c implement expand for ISD::DECLARE by just deleting it.
llvm-svn: 47708
2008-02-28 05:53:40 +00:00
Dale Johannesen bf76a08e7c Handle load/store of misaligned vectors that are the
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
2008-02-27 22:36:00 +00:00
Dan Gohman e5e32ec8f7 Remove the `else', at Evan's insistence.
llvm-svn: 47686
2008-02-27 19:44:57 +00:00
Duncan Sands 96658d0189 Support for legalizing MEMBARRIER.
llvm-svn: 47667
2008-02-27 08:53:44 +00:00
Dan Gohman 66272a545b Teach Legalize how to expand an EXTRACT_ELEMENT.
llvm-svn: 47656
2008-02-27 01:52:30 +00:00
Dan Gohman 432e4a6742 Make some static variables const.
llvm-svn: 47566
2008-02-25 21:39:34 +00:00
Dan Gohman 1f372edd97 Convert MaskedValueIsZero and all its users to use APInt. Also add
a SignBitIsZero function to simplify a common use case.

llvm-svn: 47561
2008-02-25 21:11:39 +00:00
Dan Gohman f3057a939d Fix a regression in 403.gcc and 186.crafty introduced in 47383. To test
that a value is >= 32, check that all of the high bits are zero, not
just one or more.

llvm-svn: 47467
2008-02-22 01:12:31 +00:00
Andrew Lenharth 95528943e9 Atomic op support. If any gcc test uses __sync builtins, it might start failing on archs that haven't implemented them yet
llvm-svn: 47430
2008-02-21 06:45:13 +00:00
Dan Gohman 34fc7dbf5b Convert Legalize to use the APInt form of ComputeMaskedBits.
llvm-svn: 47383
2008-02-20 16:57:27 +00:00
Andrew Lenharth fedcf477b5 I cannot find a libgcc function for this builtin. Therefor expanding it to a noop (which is how it use to be treated). If someone who knows the x86 backend better than me could tell me how to get a lock prefix on an instruction, that would be nice to complete x86 support.
llvm-svn: 47213
2008-02-16 14:46:26 +00:00
Andrew Lenharth 9b254eed32 llvm.memory.barrier, and impl for x86 and alpha
llvm-svn: 47204
2008-02-16 01:24:58 +00:00
Scott Michel a3cefeaf0c Make tblgen a little smarter about constants smaller than i32. Currently,
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
2008-02-15 23:05:48 +00:00
Dan Gohman a36ade5595 Use StoreSDNode::getValue instead of calling getOperand directly
with a hard-coded operand number.

llvm-svn: 47163
2008-02-15 18:11:59 +00:00
Duncan Sands 4c95dbd69f In TargetLowering::LowerCallTo, don't assert that
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
2008-02-14 17:28:50 +00:00
Nate Begeman 53e1b3f9d5 Change how FP immediates are handled.
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
2008-02-14 08:57:00 +00:00
Nate Begeman 735ab3ce67 Support legalizing insert_vector_elt on targets where the element
type is not legal.

llvm-svn: 47048
2008-02-13 06:43:04 +00:00
Dan Gohman 54d3b5a1f5 From Chris' review: use cast instead of dyn_cast with an assert.
llvm-svn: 46962
2008-02-11 18:58:42 +00:00
Duncan Sands 7377f5fbe3 Add a isBigEndian method to complement isLittleEndian.
llvm-svn: 46954
2008-02-11 10:37:04 +00:00
Dan Gohman 16d4bc3dc0 Follow Chris' suggestion; change the PseudoSourceValue accessors
to return pointers instead of references, since this is always what
is needed.

llvm-svn: 46857
2008-02-07 18:41:25 +00:00
Dan Gohman 2d489b5081 Re-apply the memory operand changes, with a fix for the static
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
2008-02-06 22:27:42 +00:00
Evan Cheng efd142a920 SDIsel processes llvm.dbg.declare by recording the variable debug information descriptor and its corresponding stack frame index in MachineModuleInfo. This only works if the local variable is "homed" in the stack frame. It does not work for byval parameter, etc.
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
2008-02-02 04:07:54 +00:00
Evan Cheng 263070ea2b Rename RecordLabel to RecordSourceLine because that's what it is doing.
llvm-svn: 46628
2008-02-01 02:05:57 +00:00
Evan Cheng 27b32b87ed Revert 46556 and 46585. Dan please fix the PseudoSourceValue problem and re-commit.
llvm-svn: 46623
2008-01-31 21:00:00 +00:00
Evan Cheng 1c6c16ea11 Add an extra operand to LABEL nodes which distinguishes between debug, EH, or misc labels. This fixes the EH breakage. However I am not convinced this is *the* solution.
llvm-svn: 46609
2008-01-31 09:59:15 +00:00
Dan Gohman 9ba4d76816 Rename ISD::FLT_ROUNDS to ISD::FLT_ROUNDS_ to avoid conflicting
with the real FLT_ROUNDS (defined in <float.h>).

llvm-svn: 46587
2008-01-31 00:41:03 +00:00
Dan Gohman 3646fdda67 Create a new class, MemOperand, for describing memory references
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
2008-01-31 00:25:39 +00:00
Dan Gohman 47a7d6fafe Factor the addressing mode and the load/store VT out of LoadSDNode
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
2008-01-30 00:15:11 +00:00
Nate Begeman ef33767efb Properly expand extract-element for non-power-of-2 codegen
llvm-svn: 46486
2008-01-29 02:24:00 +00:00
Duncan Sands 95d46ef887 The last pieces needed for loading arbitrary
precision integers.  This won't actually work
(and most of the code is dead) unless the new
legalization machinery is turned on.  While
there, I rationalized the handling of i1, and
removed some bogus (and unused) sextload patterns.
For i1, this could result in microscopically
better code for some architectures (not X86).
It might also result in worse code if annotating
with AssertZExt nodes turns out to be more harmful
than helpful.

llvm-svn: 46280
2008-01-23 20:39:46 +00:00
Duncan Sands 88de26cffb The final piece needed for storing arbitrary precision
integers.  Handle truncstore of a legal type to an unusual
number of bits.  Most of this code is not reachable unless
the new legalize infrastructure is turned on.

llvm-svn: 46249
2008-01-22 07:17:34 +00:00
Dale Johannesen 949e5a2f8a Do not generate a FP_ROUND of f64 to f64.
llvm-svn: 46195
2008-01-20 01:18:38 +00:00
Chris Lattner 1ea55cf816 This commit changes:
1. Legalize now always promotes truncstore of i1 to i8. 
2. Remove patterns and gunk related to truncstore i1 from targets.
3. Rename the StoreXAction stuff to TruncStoreAction in TLI.
4. Make the TLI TruncStoreAction table a 2d table to handle from/to conversions.
5. Mark a wide variety of invalid truncstores as such in various targets, e.g.
   X86 currently doesn't support truncstore of any of its integer types.
6. Add legalize support for truncstores with invalid value input types.
7. Add a dag combine transform to turn store(truncate) into truncstore when
   safe.

The later allows us to compile CodeGen/X86/storetrunc-fp.ll to:

_foo:
	fldt	20(%esp)
	fldt	4(%esp)
	faddp	%st(1)
	movl	36(%esp), %eax
	fstps	(%eax)
	ret

instead of:

_foo:
	subl	$4, %esp
	fldt	24(%esp)
	fldt	8(%esp)
	faddp	%st(1)
	fstps	(%esp)
	movl	40(%esp), %eax
	movss	(%esp), %xmm0
	movss	%xmm0, (%eax)
	addl	$4, %esp
	ret

llvm-svn: 46140
2008-01-17 19:59:44 +00:00
Chris Lattner 72733e573b * Introduce a new SelectionDAG::getIntPtrConstant method
and switch various codegen pieces and the X86 backend over
  to using it.

* Add some comments to SelectionDAGNodes.h

* Introduce a second argument to FP_ROUND, which indicates
  whether the FP_ROUND changes the value of its input. If
  not it is safe to xform things like fp_extend(fp_round(x)) -> x.

llvm-svn: 46125
2008-01-17 07:00:52 +00:00
Chris Lattner 7ca4d5b1f3 merge a few pieces of code that do the store/load to stack
pattern to use EmitStackConvert now.

llvm-svn: 46066
2008-01-16 07:51:34 +00:00
Chris Lattner 87bc3e7ece rename ExpandBIT_CONVERT to EmitStackConvert, generalizing
it to allow it to emit different load and store kinds.

llvm-svn: 46065
2008-01-16 07:45:30 +00:00
Chris Lattner a2c7ff3386 simplify a bunch of code by using SelectionDAG::CreateStackTemporary
instead of inlining its body.

llvm-svn: 46062
2008-01-16 07:03:22 +00:00
Chris Lattner 91d86242f9 Change legalizeop of FP_ROUND and FP_EXTEND to not fall through
into the ANY_EXTEND/ZERO_EXTEND/SIGN_EXTEND code to simplify it.

Unmerge the code for FP_ROUND and FP_EXTEND from each other to 
make each one simpler.

llvm-svn: 46061
2008-01-16 06:57:07 +00:00
Chris Lattner ec224888a6 The type of the 'abort' node should be pointer type (because
it's a function pointer) not MVT::Other.  This fixes builtin_trap
lowering on ppc, alpha, ia64

llvm-svn: 46018
2008-01-15 22:09:33 +00:00
Chris Lattner ee8df1f4d3 Add support for targets that have a legal ISD::TRAP.
llvm-svn: 46014
2008-01-15 21:58:08 +00:00