"private" symbols which the assember shouldn't strip, but which the linker may
remove after evaluation. This is mostly useful for Objective-C metadata.
This is plumbing, so we don't have a use of it yet. More to come, etc.
llvm-svn: 76385
being more consistent with other backends, this makes the SPARC backend
deal with functions with arguments with illegal types correctly, which
fixes some tests in test/CodeGen/Generic.
llvm-svn: 76375
--- Reverse-merging r75799 into '.':
U test/Analysis/PointerTracking
U include/llvm/Target/TargetMachineRegistry.h
U include/llvm/Target/TargetMachine.h
U include/llvm/Target/TargetRegistry.h
U include/llvm/Target/TargetSelect.h
U tools/lto/LTOCodeGenerator.cpp
U tools/lto/LTOModule.cpp
U tools/llc/llc.cpp
U lib/Target/PowerPC/PPCTargetMachine.h
U lib/Target/PowerPC/AsmPrinter/PPCAsmPrinter.cpp
U lib/Target/PowerPC/PPCTargetMachine.cpp
U lib/Target/PowerPC/PPC.h
U lib/Target/ARM/ARMTargetMachine.cpp
U lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
U lib/Target/ARM/ARMTargetMachine.h
U lib/Target/ARM/ARM.h
U lib/Target/XCore/XCoreTargetMachine.cpp
U lib/Target/XCore/XCoreTargetMachine.h
U lib/Target/PIC16/PIC16TargetMachine.cpp
U lib/Target/PIC16/PIC16TargetMachine.h
U lib/Target/Alpha/AsmPrinter/AlphaAsmPrinter.cpp
U lib/Target/Alpha/AlphaTargetMachine.cpp
U lib/Target/Alpha/AlphaTargetMachine.h
U lib/Target/X86/X86TargetMachine.h
U lib/Target/X86/X86.h
U lib/Target/X86/AsmPrinter/X86ATTAsmPrinter.h
U lib/Target/X86/AsmPrinter/X86AsmPrinter.cpp
U lib/Target/X86/AsmPrinter/X86IntelAsmPrinter.h
U lib/Target/X86/X86TargetMachine.cpp
U lib/Target/MSP430/MSP430TargetMachine.cpp
U lib/Target/MSP430/MSP430TargetMachine.h
U lib/Target/CppBackend/CPPTargetMachine.h
U lib/Target/CppBackend/CPPBackend.cpp
U lib/Target/CBackend/CTargetMachine.h
U lib/Target/CBackend/CBackend.cpp
U lib/Target/TargetMachine.cpp
U lib/Target/IA64/IA64TargetMachine.cpp
U lib/Target/IA64/AsmPrinter/IA64AsmPrinter.cpp
U lib/Target/IA64/IA64TargetMachine.h
U lib/Target/IA64/IA64.h
U lib/Target/MSIL/MSILWriter.cpp
U lib/Target/CellSPU/SPUTargetMachine.h
U lib/Target/CellSPU/SPU.h
U lib/Target/CellSPU/AsmPrinter/SPUAsmPrinter.cpp
U lib/Target/CellSPU/SPUTargetMachine.cpp
U lib/Target/Mips/AsmPrinter/MipsAsmPrinter.cpp
U lib/Target/Mips/MipsTargetMachine.cpp
U lib/Target/Mips/MipsTargetMachine.h
U lib/Target/Mips/Mips.h
U lib/Target/Sparc/AsmPrinter/SparcAsmPrinter.cpp
U lib/Target/Sparc/SparcTargetMachine.cpp
U lib/Target/Sparc/SparcTargetMachine.h
U lib/ExecutionEngine/JIT/TargetSelect.cpp
U lib/Support/TargetRegistry.cpp
llvm-svn: 75820
additional bug fixes:
1. The bug that everyone hit was a problem in the asmprinter where it
would remove $stub but keep the L prefix on a name when emitting the
indirect symbol. This is easy to fix by keeping the name of the stub
and the name of the symbol in a StringMap instead of just keeping a
StringSet and trying to reconstruct it late.
2. There was a problem printing the personality function. The current
logic to print out the personality function from the DWARF information
is a bit of a cesspool right now that duplicates a bunch of other
logic in the asm printer. The short version of it is that it depends
on emitting both the L and _ prefix for symbols (at least on darwin)
and until I can untangle it, it is best to switch the mangler back to
emitting both prefixes.
llvm-svn: 75646
This adds location info for all llvm_unreachable calls (which is a macro now) in
!NDEBUG builds.
In NDEBUG builds location info and the message is off (it only prints
"UREACHABLE executed").
llvm-svn: 75640
--- Reverse-merging r75615 into '.':
U lib/Target/XCore/XCoreAsmPrinter.cpp
U lib/Target/PIC16/PIC16AsmPrinter.cpp
U lib/Target/X86/AsmPrinter/X86IntelAsmPrinter.cpp
U lib/Target/MSP430/MSP430AsmPrinter.cpp
U lib/Target/IA64/AsmPrinter/IA64AsmPrinter.cpp
U lib/Target/CellSPU/AsmPrinter/SPUAsmPrinter.cpp
U lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
U lib/Target/PowerPC/AsmPrinter/PPCAsmPrinter.cpp
U lib/Target/Alpha/AsmPrinter/AlphaAsmPrinter.cpp
U lib/Target/MSIL/MSILWriter.cpp
U lib/Target/Sparc/AsmPrinter/SparcAsmPrinter.cpp
U lib/Target/Mips/AsmPrinter/MipsAsmPrinter.cpp
llvm-svn: 75637
Make llvm_unreachable take an optional string, thus moving the cerr<< out of
line.
LLVM_UNREACHABLE is now a simple wrapper that makes the message go away for
NDEBUG builds.
llvm-svn: 75379
have the alignment be calculated up front, and have the back-ends obey whatever
alignment is decided upon.
This allows for future work that would allow for precise no-op placement and the
like.
llvm-svn: 74564
C bindings. Change all the backend "Initialize" functions to have C linkage.
Change the "llvm/Config/Targets.def" header to use C-style comments to avoid
compile warnings.
llvm-svn: 74026
initialization of all targets (InitializeAllTargets.h) or assembler
printers (InitializeAllAsmPrinters.h). This is a step toward the
elimination of relinked object files, so that we can build normal
archives.
llvm-svn: 73543
booleans. This gives a better indication of what the "addReg()" is
doing. Remembering what all of those booleans mean isn't easy, especially if you
aren't spending all of your time in that code.
I took Jakob's suggestion and made it illegal to pass in "true" for the
flag. This should hopefully prevent any unintended misuse of this (by reverting
to the old way of using addReg()).
llvm-svn: 71722
Massive check in. This changes the "-fast" flag to "-O#" in llc. If you want to
use the old behavior, the flag is -O0. This change allows for finer-grained
control over which optimizations are run at different -O levels.
Most of this work was pretty mechanical. The majority of the fixes came from
verifying that a "fast" variable wasn't used anymore. The JIT still uses a
"Fast" flag. I'll change the JIT with a follow-up patch.
llvm-svn: 70343
use the old behavior, the flag is -O0. This change allows for finer-grained
control over which optimizations are run at different -O levels.
Most of this work was pretty mechanical. The majority of the fixes came from
verifying that a "fast" variable wasn't used anymore. The JIT still uses a
"Fast" flag. I'm not 100% sure if it's necessary to change it there...
llvm-svn: 70270
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
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
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
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
well as 2 files that use "Registrator"s. These are to be used by the
MSVC builds, as the Win32 linker does not include libs that are
otherwise unreferenced, even if global constructors in the lib have
side-effects.
Patch by Scott Graham!
llvm-svn: 59378
priority function. Instead, just iterate over the AllNodes list, which is
already in topological order. This eliminates a fair amount of bookkeeping,
and speeds up the isel phase by about 15% on many testcases.
The impact on most targets is that AddToISelQueue calls can be simply removed.
In the x86 target, there are two additional notable changes.
The rule-bending AND+SHIFT optimization in MatchAddress that creates new
pre-isel nodes during isel is now a little more verbose, but more robust.
Instead of either creating an invalid DAG or creating an invalid topological
sort, as it has historically done, it can now just insert the new nodes into
the node list at a position where they will be consistent with the topological
ordering.
Also, the address-matching code has logic that checked to see if a node was
"already selected". However, when a node is selected, it has all its uses
taken away via ReplaceAllUsesWith or equivalent, so it won't recieve any
further visits from MatchAddress. This code is now removed.
llvm-svn: 58748
flag. Then in a debugger developers can set breakpoints at these calls
to see waht is about to be selected and what the resulting subgraph
looks like. This really helps when debugging instruction selection.
llvm-svn: 58278
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
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
Instead, just create one, and make sure everything that needs
it can access it. Previously most of the SelectionDAGISel
subclasses all had their own TargetLowering object, which was
redundant with the TargetLowering object in the TargetMachine
subclasses, except on Sparc, where SparcTargetMachine
didn't have a TargetLowering object. Change Sparc to work
more like the other targets here.
llvm-svn: 57016
- 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
isImmediate(), isRegister(), and friends, to avoid confusion
about having two different names with the same meaning. I'm
not attached to the longer names, and would be ok with
changing to the shorter names if others prefer it.
llvm-svn: 56189
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
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
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
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
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
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
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
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
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
are represented as "weak", but there are subtle differences
in some cases on Darwin, so we need both. The intent
is that "common" will behave identically to "weak" unless
somebody changes their target to do something else.
No functional change as yet.
llvm-svn: 51118
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
lowering over to SparcCallingConv.td. We can't make the switch
yet because we can't say to pass f64 registers in 2 x i32 registers
with the td file yet.
llvm-svn: 48449
that merely add passes. This allows them to be used with either
FunctionPassManager or PassManager, or even with a custom new
kind of pass manager.
llvm-svn: 48256
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
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
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
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
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
that it is cheap and efficient to get.
Move a variety of predicates from TargetInstrInfo into
TargetInstrDescriptor, which makes it much easier to query a predicate
when you don't have TII around. Now you can use MI->getDesc()->isBranch()
instead of going through TII, and this is much more efficient anyway. Not
all of the predicates have been moved over yet.
Update old code that used MI->getInstrDescriptor()->Flags to use the
new predicates in many places.
llvm-svn: 45674
a header file from libcodegen. This violates a layering order: codegen
depends on target, not the other way around. The fix to this is to
split TII into two classes, TII and TargetInstrInfoImpl, which defines
stuff that depends on libcodegen. It is defined in libcodegen, where
the base is not.
llvm-svn: 45475
that "machine" classes are used to represent the current state of
the code being compiled. Given this expanded name, we can start
moving other stuff into it. For now, move the UsedPhysRegs and
LiveIn/LoveOuts vectors from MachineFunction into it.
Update all the clients to match.
This also reduces some needless #includes, such as MachineModuleInfo
from MachineFunction.
llvm-svn: 45467
e.g. MO.isMBB() instead of MO.isMachineBasicBlock(). I don't plan on
switching everything over, so new clients should just start using the
shorter names.
Remove old long accessors, switching everything over to use the short
accessor: getMachineBasicBlock() -> getMBB(),
getConstantPoolIndex() -> getIndex(), setMachineBasicBlock -> setMBB(), etc.
llvm-svn: 45464
adjustment fields, and an optional flag. If there is a "dynamic_stackalloc" in
the code, make sure that it's bracketed by CALLSEQ_START and CALLSEQ_END. If
not, then there is the potential for the stack to be changed while the stack's
being used by another instruction (like a call).
This can only result in tears...
llvm-svn: 44037
This makes DwarfRegNum to accept list of numbers instead.
Added three different "flavours", but only slightly tested on x86-32/linux.
Please check another subtargets if possible,
llvm-svn: 43997
should only effect x86 when using long double. Now
12/16 bytes are output for long double globals (the
exact amount depends on the alignment). This brings
globals in line with the rest of LLVM: the space
reserved for an object is now always the ABI size.
One tricky point is that only 10 bytes should be
output for long double if it is a field in a packed
struct, which is the reason for the additional
argument to EmitGlobalConstant.
llvm-svn: 43688
address (not just from / to frameindexes).
- Added target hooks to unfold load / store instructions / SDNodes into separate
load, data processing, store instructions / SDNodes.
llvm-svn: 42621
InOperandList. This gives one piece of important information: # of results
produced by an instruction.
An example of the change:
def ADD32rr : I<0x01, MRMDestReg, (ops GR32:$dst, GR32:$src1, GR32:$src2),
"add{l} {$src2, $dst|$dst, $src2}",
[(set GR32:$dst, (add GR32:$src1, GR32:$src2))]>;
=>
def ADD32rr : I<0x01, MRMDestReg, (outs GR32:$dst), (ins GR32:$src1, GR32:$src2),
"add{l} {$src2, $dst|$dst, $src2}",
[(set GR32:$dst, (add GR32:$src1, GR32:$src2))]>;
llvm-svn: 40033
This patch fills the last necessary bits to enable exceptions
handling in LLVM. Currently only on x86-32/linux.
In fact, this patch adds necessary intrinsics (and their lowering) which
represent really weird target-specific gcc builtins used inside unwinder.
After corresponding llvm-gcc patch will land (easy) exceptions should be
more or less workable. However, exceptions handling support should not be
thought as 'finished': I expect many small and not so small glitches
everywhere.
llvm-svn: 39855
TargetLowering to SelectionDAG so that they have more convenient
access to the current DAG, in preparation for the ValueType routines
being changed from standalone functions to members of SelectionDAG for
the pre-legalize vector type changes.
llvm-svn: 37704
Three changes:
1. Convert signed integer types to signless versions.
2. Implement the @sext and @zext parameter attributes. Previously the
type of an function parameter was used to determine whether it should
be sign extended or zero extended before the call. This information is
now communicated via the function type's parameter attributes.
3. The interface to LowerCallTo had to be changed in order to accommodate
the parameter attribute information. Although it would have been
convenient to pass in the FunctionType itself, there isn't always one
present in the caller. Consequently, a signedness indication for the
result type and for each parameter was provided for in the interface
to this method. All implementations were changed to make the adjustment
necessary.
llvm-svn: 32788
Turn on -Wunused and -Wno-unused-parameter. Clean up most of the resulting
fall out by removing unused variables. Remaining warnings have to do with
unused functions (I didn't want to delete code without review) and unused
variables in generated code. Maintainers should clean up the remaining
issues when they see them. All changes pass DejaGnu tests and Olden.
llvm-svn: 31380
DLL* linkages got full (I hope) codegeneration support in C & both x86
assembler backends.
External weak linkage added for future use, we don't provide any
codegeneration, etc. support for it.
llvm-svn: 30374
This pass:
1. Splits TargetMachine into TargetMachine (generic targets, can be implemented
any way, like the CBE) and LLVMTargetMachine (subclass of TM that is used by
things using libcodegen and other support).
2. Instead of having each target fully populate the passmgr for file or JIT
output, move all this to common code, and give targets hooks they can
implement.
3. Commonalize the target population stuff between file emission and JIT
emission.
4. All (native code) codegen stuff now happens in a FunctionPassManager, which
paves the way for "fast -O0" stuff in the CFE later, and now LLC could
lazily stream .bc files from disk to use less memory.
5. There are now many fewer #includes and the targets don't depend on the
scalar xforms or libanalysis anymore (but codegen does).
6. Changing common code generator pass ordering stuff no longer requires
touching all targets.
7. The JIT now has the option of "-fast" codegen or normal optimized codegen,
which is now orthogonal to the fact that JIT'ing is being done.
llvm-svn: 30081
in the start of an array and a count of operands where applicable. In many
cases, the number of operands is known, so this static array can be allocated
on the stack, avoiding the heap. In many other cases, a SmallVector can be
used, which has the same benefit in the common cases.
I updated a lot of code calling getNode that takes a vector, but ran out of
time. The rest of the code should be updated, and these methods should be
removed.
We should also do the same thing to eliminate the methods that take a
vector of MVT::ValueTypes.
It would be extra nice to convert the dagiselemitter to avoid creating vectors
for operands when calling getTargetNode.
llvm-svn: 29566
x86 and ppc for 100% dense switch statements when relocations are non-PIC.
This support will be extended and enhanced in the coming days to support
PIC, and less dense forms of jump tables.
llvm-svn: 27947
manner that the LowerSwitch LLVM to LLVM pass does: emitting a binary
search tree of basic blocks. The new approach has several advantages:
it is faster, it generates significantly smaller code in many cases, and
it paves the way for implementing dense switch tables as a jump table by
handling switches directly in the instruction selector.
This functionality is currently only enabled on x86, but should be safe for
every target. In anticipation of making it the default, the cfg is now
properly updated in the x86, ppc, and sparc select lowering code.
llvm-svn: 27156
Make the PPC backend not dependent on BRTWOWAY_CC and make the branch
selector smarter about the code it generates, fixing a case in the
readme.
llvm-svn: 26814
The ABI specifies that there is a register save area at the bottom of the
stack, which means the actual used pointer needs to be an offset from
the subtracted value.
llvm-svn: 26202
switch statements in the constructors and simplifies the
implementation of the getUseType() member function. You will have to
specify defs using MachineOperand::Def instead of MOTy::Def though
(similarly for Use and UseAndDef).
llvm-svn: 11715
ilist of MachineInstr objects. This allows constant time removal and
insertion of MachineInstr instances from anywhere in each
MachineBasicBlock. It also allows for constant time splicing of
MachineInstrs into or out of MachineBasicBlocks.
llvm-svn: 11340
It's not clear why the code was looking for signed chars < 0, but it can't
matter to the assembler anyway, so the check goes away. This also fixes
compatibility with arrays of [us]byte that have constantexprs in them.
Also slightly restructure some code to be cleaner.
llvm-svn: 10854
implementation of a Target{RegInfo, InstrInfo, Machine, etc} now has a separate
header and a separate implementation file.
This means that instead of a massive SparcInternals.h that forces a
recompilation of the whole target whenever a minor detail is changed, you should
only recompile a few files.
Note that SparcInternals.h is still around; its contents should be minimized.
llvm-svn: 10500
a) remove opIsUse(), opIsDefOnly(), opIsDefAndUse()
b) add isUse(), isDef()
c) rename opHiBits32() to isHiBits32(),
opLoBits32() to isLoBits32(),
opHiBits64() to isHiBits64(),
opLoBits64() to isLoBits64().
This results to much more readable code, for example compare
"op.opIsDef() || op.opIsDefAndUse()" to "op.isDef()" a pattern used
very often in the code.
llvm-svn: 10461
* There is now only one pass to print out assembly instead of two
* It is a FunctionPass
* The Module-level printing of globals is now in doFinalization() method of the
FunctionPass
* The code has been reformatted to follow LLVM coding standards
* Some comments, not all, were doxygenified
* Last but not least, the function to create an instance of this pass is also no
longer a method in the UltraSparc class.
llvm-svn: 9953
for the Sparc backend: breaking up constant expressions. Thus, we cannot have it
guarded by a conditional, it should never be disabled.
Also, it's now available for the JIT since it is a FunctionPass.
llvm-svn: 9791
MachineConstantPool. This involved refactoring the two classes involved in
printing out Sparc assembly. In fact, they should share all this code anyway.
llvm-svn: 9776
it will be converted to a MachineConstantPool index during instruction
selection
* This is now eligible to become a FunctionPass since it does not have any side
effects outside of the function it is processing.
llvm-svn: 9773
* It's no longer a BasicBlock pass: update comment on run() method
* Fix placement of braces to be consistent
* Delete extraneous whitespace
llvm-svn: 9361
the module. This change converts it from being a basic block pass to being
a simple pass. This allows elimination of the annotation and simplification
of the logic for moving constants into global variables.
llvm-svn: 9320
Make insertFarJumpAtAddr() return void, because nothing uses its return value.
Remove some commented-out code.
Implement replaceMachineCodeForFunction() for SPARC.
llvm-svn: 9203
Rename SlotCalculator::getValSlot() to SlotCalculator::getSlot(),
SlotCalculator::insertValue() to SlotCalculator::getOrCreateSlot(),
SlotCalculator::insertVal() to SlotCalculator::insertValue(), and
SlotCalculator::doInsertVal() to SlotCalculator::doInsertValue().
llvm-svn: 9190
functionality of FunctionInfo pass as doFinalization method.
Rename pass to match names of other passes like it.
Rename the pass creator fn to mimic the other creator fn names.
Include StringExtras for utostr().
Make symbol prologue/epilogue stuff redundant with
EmitBytecodeToAssembly, in preparation for refactoring.
llvm-svn: 8597
some confusion when trying to generate files (it probably couldn't tell that
./file and $(SourceDir)/file may be the same file).
Now, just let VPATH find everything, and list the primary tblgen file first in
the list of dependencies so that we can just use $< to reference it in the make
rule.
This should hopefully fix the nightly tester.
llvm-svn: 8433
much cleaner and easier.
Labeled .td as a suffix for tblgen files in Makefile.rules.
Modified build rules so that source files generated during the build are placed
in the build directory and not the source directory (and not in a Debug
directory). This makes the system cleaner and allows us to have a read-only
source tree.
llvm-svn: 8424
* Generate a single BA instead of 6-instruction JUMP if possible
(this occurs both in the creation and overwriting of the stub code)
* If possible, rewrite the ORIGINAL call to call the generated function
directly, thus bypassing the stub entirely
Also added some statistics on how often calls are overwritten and how often the
CompilationCallback is invoked.
llvm-svn: 8376
* Break apart saving and restoring registers into separate functions
* Instead of saving single and double FP registers, just save the double
registers -- aliasing helps preserve the single FP registers as well.
* Request just as much memory for a stub as we actually use
llvm-svn: 8200
registers are global, are NOT windowed, and hence can be clobbered!
* Removed unused register shorthand notations
* Fixed and cleaned up comments in inline assembly
llvm-svn: 7853
because saving i1 and i2 to their ``designated'' stack slots corrupts unknown
memory in other functions, standard libraries, and worse.
In addition, this has the benefit of improving JIT performance because we
eliminate writing out 4 instructions in CompilationCallback() and 2 loads and 2
stores.
llvm-svn: 7653
2. Handle fp-to-uint conversions directly here instead of relying on
a pre-transformation to replace them with the 2-step conversion.
3. Use size rather than explicitly checking types when deciding what
opcodes to use, wherever possible. This is less error prone (the
bug fix above was not the first time!).
4. Float-to-pointer casts shd now work though this hasn't been tested.
llvm-svn: 7645
* Doxygen-ified comments
* Added capability to make far calls (i.e., beyond 30 bits in CALL instr)
which implies that we need to delete function references that were added by
the call to addFunctionReference() because the actual call instruction is 10
instructions away (thanks to 64-bit address construction)
* Cleaned up code that generates far jumps by using an array+loop
SparcV9CodeEmitter.h:
* Explained more of the side-effects of emitFarCall()
llvm-svn: 7639
Note that some generated operators (like &, | or ^) may
not be supported by the assembler -- but if they've got
this far, it's better to generate them and let the assembler decide.
llvm-svn: 7476
since it is *necessary* for correct code generation. Only optional
transformations belong in the PreOpts pass (which needs to be renamed
from PreSelection to PreOpts).
llvm-svn: 7474
that depends on machine register size.
Moved insertCallerSavingCode() to PhyRegAlloc and
moved isRegVolatile and modifiedByCall to TargetRegInfo: they are all
machine independent. Remove several dead functions.
llvm-svn: 7392
Single and FP double reg types (which share the same reg class).
Now all methods marking/finding unused regs consider the regType
within the reg class, and SparcFloatRegClass specializes this code.
(2) Remove machine-specific regalloc. methods that are no longer needed.
In particular, arguments and return value from a call do not need
machine-specific code for allocation.
(3) Rename TargetRegInfo::getRegType variants to avoid unintentional
overloading when an include file is omitted.
llvm-svn: 7334
now works in instructions which require a 2-bit or 3-bit INTcc code.
Incidentally, that means that the representation of INTcc registers is now the
same in both integer and FP instructions. Thus, code became much simpler and
cleaner.
llvm-svn: 7185
allow, i.e. make a sequence of instructions to enable an indirect call using
jump-and-link and 2 temporary registers (which we save and ultimately restore).
Warning: if the delay slot of a function call is used to do meaningful work and
not just a NOP, this behavior is incorrect. However, the Sparc backend does not
yet utilize the delay slots effectively, so it is not necessary to make an
overly complicated algorithm for something that's not used.
llvm-svn: 7178
* FP double registers are now coded correctly
* Removed function which converted registers based on register types, it was
broken (because regTypes are broken)
llvm-svn: 7175
(1) Cannot use ANDN(ot), ORN, and XORN for boolean ops, only bitwise ops.
(2) Conditional move instructions must distinguish signed and unsigned
condition codes, e.g., MOVLE vs. MOVLEU.
(3) Conditional-move-on-register was using the cond-move-on-cc opcodes,
which produces a valid-looking instruction with bogus registers!
(4) Here's a really cute one: dividing-by-2^k for negative numbers needs to
add 2^k-1 before shifting, not add 1 after shifting. Sadly, these
are the same when k=0 so our poor test case worked fine.
(5) Casting between signed and unsigned values was not correct:
completely reimplemented.
(6) Zero-extension on unsigned values was bogus: I was only doing the
SRL and not the SLLX before it. Don't know WHAT I was thinking!
(7) And the most important class of changes: Sign-extensions on signed values.
Signed values are not sign-extended after ordinary operations,
so they must be sign-extended before the following cases:
-- passing to an external or unknown function
-- returning from a function
-- using as operand 2 of DIV or REM
-- using as either operand of condition-code setting operation
(currently only SUBCC), with smaller than 32-bit operands
Also, a couple of improvements:
(1) Fold cast-to-bool into Not(bool). Need to do this for And, Or, XOR also.
(2) Convert SetCC-Const into a conditional-move-on-register (case 41)
if the constant is 0. This was only being done for branch-on-SetCC-Const
when the branch is folded with the SetCC-Const.
llvm-svn: 7159
(1) An int CC live range must be spilled if there are any interferences,
even if no other "neighbour" in the interf. graph has been allocated
that reg. yet. This is actually true of any class with only one reg!
(2) SparcIntCCRegClass::colorIGNode sets the color even if the LR must
be spilled so that the machine-independent spill code doesn't have to
make the machine-dependent decision of which CC name to use based on
operand type: %xcc or %icc. (These are two halves of the same register.)
(3) LR->isMarkedForSpill() is no longer the same as LR->hasColor().
These should never have been the same, and this is necessary now for #2.
(4) All RDCCR and WRCCR instructions are directly generated with the
phony number for %ccr so that EmitAssembly/EmitBinary doesn't have to
deal with this.
llvm-svn: 7151
correct: empirically, "regType" is wrong for a number of registers. Thus, one
can only rely on the "regClass" to figure out what kind of register one is
dealing with.
This change switches to using only "regClass" and adds a few extra DEBUG() print
statements and a few clean-ups in comments and code, mostly minor.
llvm-svn: 7103
the TableGen descriptions; all unset bits are thus errors.
* As a result, found and fixed instructions where some operands were not
actually assigned into the right portion of the instruction.
llvm-svn: 7074
area to avoid using up precious stack space within the 4095 offset limit
from %fp. Such objects that would themselves live at a large offset
were being put there already so this is a simple change.
llvm-svn: 7066
deletes it, but we were merrily trying to fix the operands of that
instruction anyway! Instead, fix the replacement instruction.
(2) An Improvement: Check for and extract global values in all operands,
not just in known pointer operands. For example, they can occur in
call arguments, and probably other unforeseeable places as well.
This also eliminates the special-case handling of Load and Store.
llvm-svn: 7053
(1) select: Ok to convert a pointer to a float or double.
(2) regalloc: Some MachineInstr* for caller-saving code before a call
were being inserted before and after the call!
(3) Don't insert the caller-saving instructions in the
MachineCodeForInstruction for the Call instruction.
*All* instructions generated by register allocation need to be
recorded in those maps, but it needs to be done uniformly.
llvm-svn: 7051
Move LowerAllocations, PrintFunction, and SymbolStripping passes, and
the corresponding -disable-strip and -d options, over here to the SPARC
target-specific bits of llc. Rename -d to -dump-asm.
tools/llc/Makefile:
Reindent. Add x86 library so that llc compiles again.
tools/llc/llc.cpp:
Remove support for running arbitrary optimization passes. Use opt instead.
Remove LowerAllocations, PrintFunction, and SymbolStripping passes, as noted
above.
Allow user to select a backend (x86 or SPARC); default to guessing from
the endianness/pointer size of the input bytecode file.
Fix typos.
Delete empty .s file and exit with error status if target does not support
static compilation.
llvm-svn: 6776
* BPA and BPN do not take a %cc register as a parameter
* SLL/SRL/SRA{r,i}5 are there for a reason - they are ONLY 32-bit instructions
* Likewise, SLL/SRL/SRAX{r,i}6 are only 64-bit
* Added WRCCR{r,i} opcodes
llvm-svn: 6655
* Removed instruction scheduling as it is too slow to run in a JIT environment
* Removed other passes because they aren't necessary and can slow JIT down
llvm-svn: 6652
Special cases: STFSRx and STXFSRx - they operate on predefined rd=0 or rd=1, and
expect %fsr as the parameter in assembly. They are disabled (since not used)
until an encoding, both for code generation and output, is chosen.
llvm-svn: 6619
Update file comment to contain a bunch of the overview mapping-info
documentation previously buried within the file.
Remove some unnecessary include/using stmts.
Rename pass to MappingInfoCollector.
Rewrite a lot of it so it doesn't use global instance variables and so
it outputs into MappingInfo objects and then dumps those out, instead of going
straight to an assembly file.
Change name of factory to getMappingInfoCollector.
Fold prologue & epilogue writers into MappingInfo methods.
lib/Target/Sparc/FInfo.cpp:
Correct file comment to reflect above change
lib/Target/Sparc/Sparc.cpp:
Change name of factory to getMappingInfoCollector.
llvm-svn: 6617
in Emitter.cpp, just convert the Sparc version of the constant pool into
what's already supported and inter-operate.
* Implemented a first pass at lazy function resolution in the JITResolver. That
required adding a SparcV9CodeEmitter pointer to simplify generating
bit-patterns of the instructions.
* SparcV9CodeEmitter now creates and destroys static TheJITResolver, which makes
sense because the SparcV9CodeEmitter is the only user of TheJITResolver, and
lives for the entire duration of the JIT (via PassManager which lives in VM).
* Changed all return values in the JITResolver to uint64_t because of the 64-bit
Sparc architecture.
* Added a new version of getting the value of a GlobalValue in the
SparcV9CodeEmitter, which now works for already-generated functions (JITted or
library functions).
* Removed little-used and unused functions, cleaning up the internal view of the
SparcV9CodeEmitter.
llvm-svn: 6612
rewriting it. I also vacuumed out all the commented-out code and
inaccurate comments, etc.
(We need to put the mapping information in a data structure so that we can
pass it out to the JIT, instead of automagically converting it to .byte
directives.)
llvm-svn: 6574
they prefer the destination register to be last. Thus, two new classes were made
for them that accomodate for having this layout of operands (F3_1rd, F3_2rd).
llvm-svn: 6564
* Added instruction classes which start building from rs1, then rs2, and rd.
* Fixed order of operands in classes 4.1 and 4.2; added 4.6 .
llvm-svn: 6561
* Added new classes which start building from rs1, adding rs2, and then rd.
* Fixed order of operands in classes 3.11, 3.12, 3.16, and 3.17 .
* Fixed comments to reflect Real Life (tm).
* Removed "don't care" commented out assignments and dead classes (#if 0).
llvm-svn: 6560
SparcInstrSelection.cpp:
* Fixed opcodes to return correct 'i' version since the two functions are each
only used in one place.
* Changed name of function to have an 'i' in the name to signify that they each
return an immediate form of the opcode.
* Added a warning if either of the functions is ever used in a context which
requires a register-version opcode.
SparcV9_F4.td: fixed class F4_3, added F4_4 and notes that F4_{1,2} need fixing
SparcV9.td: added the MOV(F)cc instructions
llvm-svn: 6548
globals in some other places may not have been pulled out either;
globals in phi operands were being put just before the phi instead of
in the predecessor basic blocks.
llvm-svn: 6466
call arguments and return values:
Now all copy operations before and after a call are generated during
selection instead of during register allocation.
The values are copied to virtual registers (or to the stack), but
in the former case these operands are marked with the correct physical
registers according to the calling convention.
Although this complicates scheduling and does not work well with
live range analysis, it simplifies the machine-dependent part of
register allocation.
llvm-svn: 6465
instructions of format 3.12 and 3.13 cannot inherit from F3rdrs1, because that
implies that the two registers are the first two parameters to the instruction.
Thus I made the instructions inherit from F3rd again, and manually added an rs1
field AFTER the shcnt field in the instruction, which maps to the appropriate
place in the instruction.
The other changes are just elimination of unnecessary spaces.
llvm-svn: 6437
SparcV9_F3.td: F3_12 and F3_13 instructions have rd and rs1 fields. Also,
their fields were totally screwed up. This seems to fix the problem.
llvm-svn: 6429
exit code. This, in turn, makes an empty file SparcV9CodeEmitter.inc, and only
much later, produces a link error because the key function that TableGen creates
isn't found.
Using a temporary file in the middle forces a good .INC file to be generated by
TableGen, and it will keep trying until you fix the input file.
llvm-svn: 6392
list (rd, rs1, imm), in that order (bit-wise), the actual assembly syntax is
instr rd, imm, rs1, and that is how they are constructed in the instruction
selector. This fixes the discrepancy.
Also fixed some comments along the same lines and fixed page numbers referring
to where instructions are described in the Sparc manual.
llvm-svn: 6384
Non-obvious change: since I have changed ST and STD to be STF and STDF to
(a) closer resemble their name (NOT assembly text) in the Sparc manual, and
(b) clearly specify that they they are floating-point opcodes,
I made the same changes in this file.
llvm-svn: 6376
Here I had to make one non-trivial change: add a function to get a version of
the opcode that takes an immediate, given an opcode that takes all registers.
This is required because sometimes it is not known at construction time which
opcode is used because opcodes are passed around between functions.
llvm-svn: 6375
annotations on instructions to specify which format they are (i.e., do they take
2 registers and 1 immediate or just 3 registers) as that changes their binary
representation and hence, code emission.
This makes instructions more like how X86 defines them to be. Now, writers of
instruction selection must choose the correct opcode based on what instruction
type they are building, which they already know. Thus, the JIT doesn't have to
do the same work by `discovering' which operands an instruction really has.
As this involves lots of small changes to a lot of files in lib/target/Sparc,
I'll commit them individually because otherwise the diffs will be unreadable.
llvm-svn: 6371
a default value is set in the header file.
* Fixed some code layout to make it more consistent with the rest of codebase
* Added addPassesToJITCompile() with relevant passes
llvm-svn: 6369
Also, added annotations to how instructions are modified (reg/imm operands).
Added prototype for adding register numbers to values pass for interfacing with
the target-independent register allocators in the JIT.
llvm-svn: 6366
Bill Wendling