This is done by pushing physical register definitions close to their
use, which happens to handle flag definitions if they're not glued to
the branch. This seems to be generally a good thing though, so I
didn't need to add a target hook yet.
The primary motivation is to generate code closer to what people
expect and rule out missed opportunity from enabling macro-op
fusion. As a side benefit, we get several 2-5% gains on x86
benchmarks. There is one regression:
SingleSource/Benchmarks/Shootout/lists slows down be -10%. But this is
an independent scheduler bug that will be tracked separately.
See rdar://problem/9283108.
Incidentally, pre-RA scheduling is only half the solution. Fixing the
later passes is tracked by:
<rdar://problem/8932804> [pre-RA-sched] on x86, attempt to schedule CMP/TEST adjacent with condition jump
Fixes:
<rdar://problem/9262453> Scheduler unnecessary break of cmp/jump fusion
llvm-svn: 129508
will allow multiple context with different loop unroll parameters to run. This is a minor change and no effect
on existing application.
llvm-svn: 129449
Additional fixes:
Do something reasonable for subtargets with generic
itineraries by handle node latency the same as for an empty
itinerary. Now nodes default to unit latency unless an itinerary
explicitly specifies a zero cycle stage or it is a TokenFactor chain.
Original fixes:
UnitsSharePred was a source of randomness in the scheduler: node
priority depended on the queue data structure. I rewrote the recent
VRegCycle heuristics to completely replace the old heuristic without
any randomness. To make the ndoe latency adjustments work, I also
needed to do something a little more reasonable with TokenFactor. I
gave it zero latency to its consumers and always schedule it as low as
possible.
llvm-svn: 129421
Now that we have a first-class way to represent unaligned loads, the unaligned
load intrinsics are superfluous.
First part of <rdar://problem/8460511>.
llvm-svn: 129401
Add handling for tracking the relocations on symbols and resolving them.
Keep track of the relocations even after they are resolved so that if
the RuntimeDyld client moves the object, it can update the address and any
relocations to that object will be updated.
For our trival object file load/run test harness (llvm-rtdyld), this enables
relocations between functions located in the same object module. It should
be trivially extendable to load multiple objects with mutual references.
As a simple example, the following now works (running on x86_64 Darwin 10.6):
$ cat t.c
int bar() {
return 65;
}
int main() {
return bar();
}
$ clang t.c -fno-asynchronous-unwind-tables -o t.o -c
$ otool -vt t.o
t.o:
(__TEXT,__text) section
_bar:
0000000000000000 pushq %rbp
0000000000000001 movq %rsp,%rbp
0000000000000004 movl $0x00000041,%eax
0000000000000009 popq %rbp
000000000000000a ret
000000000000000b nopl 0x00(%rax,%rax)
_main:
0000000000000010 pushq %rbp
0000000000000011 movq %rsp,%rbp
0000000000000014 subq $0x10,%rsp
0000000000000018 movl $0x00000000,0xfc(%rbp)
000000000000001f callq 0x00000024
0000000000000024 addq $0x10,%rsp
0000000000000028 popq %rbp
0000000000000029 ret
$ llvm-rtdyld t.o -debug-only=dyld ; echo $?
Function sym: '_bar' @ 0
Function sym: '_main' @ 16
Extracting function: _bar from [0, 15]
allocated to 0x100153000
Extracting function: _main from [16, 41]
allocated to 0x100154000
Relocation at '_main' + 16 from '_bar(Word1: 0x2d000000)
Resolving relocation at '_main' + 16 (0x100154010) from '_bar (0x100153000)(pcrel, type: 2, Size: 4).
loaded '_main' at: 0x100154000
65
$
llvm-svn: 129388
Use debug info in the IR to find the directory/file:line:col. Each time that location changes, bump a counter.
Unlike the existing profiling system, we don't try to look at argv[], and thusly don't require main() to be present in the IR. This matches GCC's technique where you specify the profiling flag when producing each .o file.
The runtime library is minimal, currently just calling printf at program shutdown time. The API is designed to make it possible to emit GCOV data later on.
llvm-svn: 129340
disassembler API. Hooked this up to the ARM target so such tools as Darwin's
otool(1) can now print things like branch targets for example this:
blx _puts
instead of this:
blx #-36
And even print the expression encoded in the Mach-O relocation entried for
things like this:
movt r0, :upper16:((_foo-_bar)+1234)
llvm-svn: 129284
--- Reverse-merging r129235 into '.':
D test/Feature/bb_attrs.ll
U include/llvm/BasicBlock.h
U include/llvm/Bitcode/LLVMBitCodes.h
U lib/VMCore/AsmWriter.cpp
U lib/VMCore/BasicBlock.cpp
U lib/AsmParser/LLParser.cpp
U lib/AsmParser/LLLexer.cpp
U lib/AsmParser/LLToken.h
U lib/Bitcode/Reader/BitcodeReader.cpp
U lib/Bitcode/Writer/BitcodeWriter.cpp
llvm-svn: 129259
* Add a "landing pad" attribute to the BasicBlock.
* Modify the bitcode reader and writer to handle said attribute.
Later: The verifier will ensure that the landing pad attribute is used in the
appropriate manner. I.e., not applied to the entry block, and applied only to
basic blocks that are branched to via a `dispatch' instruction.
(This is a work-in-progress.)
llvm-svn: 129235
It is common for large live ranges to have few basic blocks with register uses
and many live-through blocks without any uses. This approach grows the Hopfield
network incrementally around the use blocks, completely avoiding checking
interference for some through blocks.
llvm-svn: 129188
Teach 32-bit section loading to use the Memory Manager interface, just like
the 64-bit loading does. Tidy up a few other things here and there.
llvm-svn: 129138
with the newer, cleaner model. It uses the IAPrinter class to hold the
information that is needed to match an instruction with its alias. This also
takes into account the available features of the platform.
There is one bit of ugliness. The way the logic determines if a pattern is
unique is O(N**2), which is gross. But in reality, the number of items it's
checking against isn't large. So while it's N**2, it shouldn't be a massive time
sink.
llvm-svn: 129110
induction variable. The preRA scheduler is unaware of induction vars,
so we look for potential "virtual register cycles" instead.
Fixes <rdar://problem/8946719> Bad scheduling prevents coalescing
llvm-svn: 129100
Start teaching the runtime Dyld interface to use the memory manager API
for allocating space. Rather than mapping directly into the MachO object,
we extract the payload for each object and copy it into a dedicated buffer
allocated via the memory manager. For now, just do Segment64, so this works
on x86_64, but not yet on ARM.
llvm-svn: 128973
developers can see if their driver changed any cl::Option's. The
current implementation isn't perfect but handles most kinds of
options. This is nice to have when decomposing the stages of
compilation and moving between different drivers. It's also a good
sanity check when comparing results produced by different command line
invocations that are expected to produce the comparable results.
Note: This is not an attempt to prolong the life of cl::Option. On the
contrary, it's a placeholder for a feature that must exist when
cl::Option is replaced by a more appropriate framework. A new
framework needs: a central option registry, dynamic name lookup,
non-global containers of option values (e.g. per-module,
per-function), *and* the ability to print options values and their defaults at
any point during compilation.
llvm-svn: 128910
This allows us to always keep the smaller slot for an instruction which is what
we want when a register has early clobber defines.
Drop the UsingInstrs set and the UsingBlocks map. They are no longer needed.
llvm-svn: 128886
inlined path for the common case.
Most basic blocks don't contain a call that may throw, so the last split point
os simply the first terminator.
llvm-svn: 128874
It needed to be moved closer to the setjmp statement, because the code directly
after the setjmp needs to know about values that are on the stack. Also, the
'bitcast' of the function context was causing a dead load. This wouldn't be too
horrible, except that at -O0 it wasn't optimized out, and because it wasn't
using the correct base pointer (if there is a VLA), it would try to access a
value from a garbage address.
<rdar://problem/9130540>
llvm-svn: 128873
The JITMemory manager references LLVM IR constructs directly, while the
runtime Dyld works at a lower level and can handle objects which may not
originate from LLVM IR. Introduce a new layer for the memory manager to
handle the interface between them. For the MCJIT, this layer will be almost
entirely simply a call-through w/ translation between the IR objects and
symbol names.
llvm-svn: 128851
after the given instruction; make sure to handle that case correctly.
(It's difficult to trigger; the included testcase involves a dead
block, but I don't think that's a requirement.)
While I'm here, get rid of the unnecessary warning about
SimplifyInstructionsInBlock, since it should work correctly as far as I know.
llvm-svn: 128782
When the greedy register allocator is splitting multiple global live ranges, it
tends to look at the same interference data many times. The InterferenceCache
class caches queries for unaltered LiveIntervalUnions.
llvm-svn: 128764
transformations in target-specific DAG combines without causing DAGCombiner to
delete the same node twice. If you know of a better way to avoid this (see my
next patch for an example), please let me know.
llvm-svn: 128758
StringMap was not properly updating NumTombstones after a clear or rehash.
This was not fatal until now because the table was growing faster than
NumTombstones could, but with the previous change of preventing infinite
growth of the table the invariant (NumItems + NumTombstones <= NumBuckets)
stopped being observed, causing infinite loops in certain situations.
Patch by José Fonseca!
llvm-svn: 128567
When the hash function uses object pointers all free entries eventually
become tombstones as they are used at least once, regardless of the size.
DenseMap cannot function with zero empty keys, so it double size to get
get ridof the tombstones.
However DenseMap never shrinks automatically unless it is cleared, so
the net result is that certain tables grow infinitely.
The solution is to make a fresh copy of the table without tombstones
instead of doubling size, by simply calling grow with the current size.
Patch by José Fonseca!
llvm-svn: 128564
The idea is, that if an ieee 754 float is divided by a power of two, we can
turn the division into a cheaper multiplication. This function sees if we can
get an exact multiplicative inverse for a divisor and returns it if possible.
This is the hard part of PR9587.
I tested many inputs against llvm-gcc's frotend implementation of this
optimization and didn't find any difference. However, floating point is the
land of weird edge cases, so any review would be appreciated.
llvm-svn: 128545
was lowering them to sext / uxt + mul instructions. Unfortunately the
optimization passes may hoist the extensions out of the loop and separate them.
When that happens, the long multiplication instructions can be broken into
several scalar instructions, causing significant performance issue.
Note the vmla and vmls intrinsics are not added back. Frontend will codegen them
as intrinsics vmull* + add / sub. Also note the isel optimizations for catching
mul + sext / zext are not changed either.
First part of rdar://8832507, rdar://9203134
llvm-svn: 128502
otool(1), this time with the needed fix for case sensitive file systems :) .
This is a work in progress as the interface for producing symbolic operands is
not done. But a hacked prototype using information from the object file's
relocation entiries and replacing immediate operands with MCExpr's has been
shown to work with no changes to the instrucion printer. These APIs will be
moved into a dynamic library at some point.
llvm-svn: 128415
Correctly terminate the range of register DBG_VALUEs when the register is
clobbered or when the basic block ends.
The code is now ready to deal with variables that are sometimes in a register
and sometimes on the stack. We just need to teach emitDebugLoc to say 'stack
slot'.
llvm-svn: 128327
This is a work in progress as the interface for producing symbolic operands is
not done. But a hacked prototype using information from the object file's
relocation entiries and replacing immediate operands with MCExpr's has been
shown to work with no changes to the instrucion printer. These APIs will be
moved into a dynamic library at some point.
llvm-svn: 128308
The MC asm lexer wasn't honoring a non-default (anything but ';') statement
separator. Fix that, and generalize a bit to support multi-character
statement separators.
llvm-svn: 128227
Move the dynamic linking functionality of the llvm-rtdyld program into an
ExecutionEngine support library. Update llvm-rtdyld to just load an object
file into memory, use the library to process it, then run the _main()
function, if one is found.
llvm-svn: 128031
the alias of an InstAlias instead of the thing being aliased. Because we need to
know the features that are valid for an InstAlias.
This is part of a work-in-progress.
llvm-svn: 127986
to have single return block (at least getting there) for optimizations. This
is general goodness but it would prevent some tailcall optimizations.
One specific case is code like this:
int f1(void);
int f2(void);
int f3(void);
int f4(void);
int f5(void);
int f6(void);
int foo(int x) {
switch(x) {
case 1: return f1();
case 2: return f2();
case 3: return f3();
case 4: return f4();
case 5: return f5();
case 6: return f6();
}
}
=>
LBB0_2: ## %sw.bb
callq _f1
popq %rbp
ret
LBB0_3: ## %sw.bb1
callq _f2
popq %rbp
ret
LBB0_4: ## %sw.bb3
callq _f3
popq %rbp
ret
This patch teaches codegenprep to duplicate returns when the return value
is a phi and where the phi operands are produced by tail calls followed by
an unconditional branch:
sw.bb7: ; preds = %entry
%call8 = tail call i32 @f5() nounwind
br label %return
sw.bb9: ; preds = %entry
%call10 = tail call i32 @f6() nounwind
br label %return
return:
%retval.0 = phi i32 [ %call10, %sw.bb9 ], [ %call8, %sw.bb7 ], ... [ 0, %entry ]
ret i32 %retval.0
This allows codegen to generate better code like this:
LBB0_2: ## %sw.bb
jmp _f1 ## TAILCALL
LBB0_3: ## %sw.bb1
jmp _f2 ## TAILCALL
LBB0_4: ## %sw.bb3
jmp _f3 ## TAILCALL
rdar://9147433
llvm-svn: 127953
Proof-of-concept code that code-gens a module to an in-memory MachO object.
This will be hooked up to a run-time dynamic linker library (see: llvm-rtdyld
for similarly conceptual work for that part) which will take the compiled
object and link it together with the rest of the system, providing back to the
JIT a table of available symbols which will be used to respond to the
getPointerTo*() queries.
llvm-svn: 127916
For example, on 32-bit architecture, don't promote all uses of the IV
to 64-bits just because one use is a 64-bit cast.
Alternate implementation of the patch by Arnaud de Grandmaison.
llvm-svn: 127884
SCEV may generate expressions composed of multiple pointers, which can
lead to invalid GEP expansion. Until we can teach SCEV to follow strict
pointer rules, make sure no bad GEPs creep into IR.
Fixes rdar://problem/9038671.
llvm-svn: 127839
I have convinced myself that it can only happen when a phi value dies. When it
happens, allocate new virtual registers for the components.
llvm-svn: 127827
rather than an int. Thankfully, this only causes LLVM to miss optimizations, not
generate incorrect code.
This just fixes the zext at the return. We still insert an i32 ZextAssert when
reading a function's arguments, but it is followed by a truncate and another i8
ZextAssert so it is not optimized.
llvm-svn: 127766
doesn't return, so just go back to using the old runtime function instead
of trying to use abort() when __builtin_unreachable (or an equivalent) isn't
supported.
llvm-svn: 127629
properties.
Added the self-wrap flag for SCEV::AddRecExpr.
A slew of temporary FIXMEs indicate the intention of the no-self-wrap flag
without changing behavior in this revision.
llvm-svn: 127590
llvm-gcc-i386-linux-selfhost and llvm-x86_64-linux-checks buildbots.
The original log entry:
Remove optimization emitting a reference insted of label difference, since
it can create more relocations. Removed isBaseAddressKnownZero method,
because it is no longer used.
llvm-svn: 127540
It generates output that lools like
8 times line number info lost by Scalar Replacement of Aggregates (SSAUp)
1 times line number info lost by Simplify well-known library calls
12 times variable info lost by Jump Threading
llvm-svn: 127381
flexible.
If it returns a register class that's different from the input, then that's the
register class used for cross-register class copies.
If it returns a register class that's the same as the input, then no cross-
register class copies are needed (normal copies would do).
If it returns null, then it's not at all possible to copy registers of the
specified register class.
llvm-svn: 127368
the value splatted into every element. Extend this to getTrue and getFalse which
by providing new overloads that take Types that are either i1 or <N x i1>. Use
it in InstCombine to add vector support to some code, fixing PR8469!
llvm-svn: 127116
This makes lookup slightly more expensive but it's worth it, unused
DenseMaps are common in LLVM code apparently.
1% speedup on clang -O3 bzip2.c
4% speedup on clang -O3 oggenc.c (Release build of clang on i386/linux)
llvm-svn: 127088
regs. This is the only change in this checkin that may affects the
default scheduler. With better register tracking and heuristics, it
doesn't make sense to artificially lower the register limit so much.
Added -sched-high-latency-cycles and X86InstrInfo::isHighLatencyDef to
give the scheduler a way to account for div and sqrt on targets that
don't have an itinerary. It is currently defaults to 10 (the actual
number doesn't matter much), but only takes effect on non-default
schedulers: list-hybrid and list-ilp.
Added several heuristics that can be individually disabled for the
non-default sched=list-ilp mode. This helps us determine how much
better we can do on a given benchmark than the default
scheduler. Certain compute intensive loops run much faster in this
mode with the right set of heuristics, and it doesn't seem to have
much negative impact elsewhere. Not all of the heuristics are needed,
but we still need to experiment to decide which should be disabled by
default for sched=list-ilp.
llvm-svn: 127067
Initially, slot indexes are quad-spaced. There is room for inserting up to 3
new instructions between the original instructions.
When we run out of indexes between two instructions, renumber locally using
double-spaced indexes. The original quad-spacing means that we catch up quickly,
and we only have to renumber a handful of instructions to get a monotonic
sequence. This is much faster than renumbering the whole function as we did
before.
llvm-svn: 127023
it. It's been assumed up til now that it would be in its immediate
successor. However, this isn't necessarily the case. It could be in one of its
successor's successors.
Modify the code to more thoroughly check for an 'eh.selector' call in
successors. It only looks at a successor if we get there as a result of an
unconditional branch.
Testcase ObjC/exceptions-4.m in r126968.
llvm-svn: 126969
This is much faster than using a pointer to a ManagedStatic object accessed with
a function call. The greedy register allocator is 5% faster overall just from
the SlotIndex default constructor savings.
llvm-svn: 126925
The SlotIndex created by the default construction does not represent a position
in the function, and it doesn't make sense to compare it to other indexes.
llvm-svn: 126924
uses.
The result produced by the streamer is used to give the linker more accurate
information and to add to llvm.compiler.used. The second improvement removes
the need for the user to add __attribute__((used)) to functions only used in
inline asm. The first one lets us build firefox with LTO on Darwin :-)
llvm-svn: 126830
This method could probably be used by LiveIntervalAnalysis::shrinkToUses, and
now it can use extendIntervalEndTo() which coalesces ranges.
llvm-svn: 126803
Use 8 bits from line number field to keep track of argument ordering while encoding debug info for an argument. That leaves 24 bit for line no, DebugLoc also allocates 24 bit for line numbers. If a function has more than 255 arguments then rest of the arguments will be ordered by llvm.dbg.* intrinsics' ordering in IR.
llvm-svn: 126793
a StringRef, for the benefit of clients that want the result as a
nul-terminated string. Clients that expect a StringRef will get one via
the implicit conversion.
llvm-svn: 126784
Introduce a variable in the AsmParserExtension whether [] is valid in an
expression. If it is true, parse them like (). Enable this for ELF only.
llvm-svn: 126443
registers at phis. This enables us to eliminate a lot of pointless zexts during
the DAGCombine phase. This fixes <rdar://problem/8760114>.
llvm-svn: 126380
template <class T1, class T2> pair<T1,T2> make_pair(const T1&, const T2&);
to
template <class T1, class T2> pair<V1, V2> make_pair(T1&&, T2&&);
so explicitly specifying the template arguments to make_pair<> is going to break
when C++0x rolls through. Replace them with equivalent std::pair<>. Patch by
James Dennett!
llvm-svn: 126256
share entries. Add a DenseSet to MachineConstantPool for the MachineCPVs that
it owns.
This will hopefully fix the MC/ARM/elf-reloc-01.ll failure on the leaks bots.
llvm-svn: 126218
In other words, do not keep track of argument's location. The debugger (gdb) is not prepared to see line table entries for arguments. For the debugger, "second" line table entry marks beginning of function body.
This requires some coordination with debugger to get this working.
- The debugger needs to be aware of prolog_end attribute attached with line table entries.
- The compiler needs to accurately mark prolog_end in line table entries (at -O0 and at -O1+)
llvm-svn: 126155
itself without going via a phi node then we could return false here in
spite of making a change. Also, tweak the comment because this method
can (and always could) return true without deleting the original phi node.
For example, if the phi node was used by a read-only invoke instruction
which is used by another phi node phi2 which is only used by and only uses
the invoke, then phi2 would be deleted but not the invoke instruction and
not the original phi node.
llvm-svn: 126129
need to be pulled out of the pass manager when the user specifies
-fno-builtin. It can intelligently determine which libcalls to
optimize based on what is enabled in TargetLibraryInfo. This
allows -fno-builtin-foo to work someday.
llvm-svn: 125981
A major part of its (eventual) goal is to support a much cleaner separation between disassembly callbacks
provided by the target and the disassembler emitter itself, i.e. not requiring hardcoding of knowledge in tblgen
like the existing disassembly emitters do.
The hope is that some day this will allow us to replace the existing non-Thumb ARM disassembler and remove
some of the hacks the old one introduced to tblgen.
llvm-svn: 125966
query about available library functions. For now this just has
memset_pattern16, which exists on darwin, but it can be extended for a
bunch of other things in the future.
llvm-svn: 125965
No one uses *-mingw64. mingw-w64 is represented as {i686|x86_64}-w64-mingw32. In llvm side, i686 and x64 can be treated as similar way.
llvm-svn: 125747
Simplify the spill weight calculation a bit by bypassing
getApproximateInstructionCount() and using LiveInterval::getSize() directly.
This changes the computed spill weights, but only by a constant factor in each
function. It should not affect how spill weights compare against each other, and
so it shouldn't affect code generation.
llvm-svn: 125530
use in many places where we pass a pointer and size to abstract APIs
that can take C arrays, std::vector, SmallVector, etc. It is to arrays
what StringRef is to strings.
llvm-svn: 125486
generating i8 shift amounts for things like i1024 types. Add
an assert in getNode to prevent this from occuring in the future,
fix the buggy transformation, revert my previous patch, and
document this gotcha in ISDOpcodes.h
llvm-svn: 125465
name of a path, after resolving symbolic links and eliminating excess
path elements such as "foo/../" and "./".
This routine still needs a Windows implementation, but I don't have a
Windows machine available. Help? Please?
llvm-svn: 125228
versions of creation functions. Eventually, the "insertion point" versions
of these should just be removed, we do have IRBuilder afterall.
Do a massive rewrite of much of pattern match. It is now shorter and less
redundant and has several other widgets I will be using in other patches.
Among other changes, m_Div is renamed to m_IDiv (since it only matches
integer divides) and m_Shift is gone (it used to match all binops!!) and
we now have m_LogicalShift for the one client to use.
Enhance IRBuilder to have "isExact" arguments to things like CreateUDiv
and reduce redundancy within IRbuilder by having these methods chain to
each other more instead of duplicating code.
llvm-svn: 125194
AC_CHECK_FUNCS seeks a symbol only in libs. We should check the declaration in string.h.
FIXME: I have never seen mingw(s) have strerror_s() (not _strerror_s()).
FIXME: Autoconf/CMake may seek strerror_s() with the definition MINGW_HAS_SECURE_API in future.
llvm-svn: 125172
After uses of a live range are removed, recompute the live range to only cover
the remaining uses. This is necessary after rematerializing the value before
some (but not all) uses.
llvm-svn: 125058
Motivation: Improve the parsing of not usual (different from registers or
immediates) operand forms.
This commit implements only the generic support. The ARM specific modifications
will come next.
A table like the one below is autogenerated for every instruction
containing a 'ParserMethod' in its AsmOperandClass
static const OperandMatchEntry OperandMatchTable[20] = {
/* Mnemonic, Operand List Mask, Operand Class, Features */
{ "cdp", 29 /* 0, 2, 3, 4 */, MCK_Coproc, Feature_IsThumb|Feature_HasV6 },
{ "cdp", 58 /* 1, 3, 4, 5 */, MCK_Coproc, Feature_IsARM },
A matcher function very similar (but lot more naive) to
MatchInstructionImpl scans the table. After the mnemonic match, the
features are checked and if the "to be parsed" operand index is
present in the mask, there's a real match. Then, a switch like the one
below dispatch the parsing to the custom method provided in
'ParseMethod':
case MCK_Coproc:
return TryParseCoprocessorOperandName(Operands);
llvm-svn: 125030
a) Making it a per call site bonus for functions that we can move from
indirect to direct calls.
b) Reduces the bonus from 500 to 100 per call site.
c) Subtracts the size of the possible newly inlineable call from the
bonus to only add a bonus if we can inline a small function to devirtualize
it.
Also changes the bonus from a positive that's subtracted to a negative
that's added.
Fixes the remainder of rdar://8546196 by reducing the object file size
after inlining by 84%.
llvm-svn: 124916
A live range cannot be split everywhere in a basic block. A split must go before
the first terminator, and if the variable is live into a landing pad, the split
must happen before the call that can throw.
llvm-svn: 124894
precisely track pressure on a selection DAG, but we can at least keep
it balanced. This design accounts for various interesting aspects of
selection DAGS: register and subregister copies, glued nodes, dead
nodes, unused registers, etc.
Added SUnit::NumRegDefsLeft and ScheduleDAGSDNodes::RegDefIter.
Note: I disabled PrescheduleNodesWithMultipleUses when register
pressure is enabled, based on no evidence other than I don't think it
makes sense to have both enabled.
llvm-svn: 124853
The greedy register allocator revealed some problems with the value mapping in
SplitKit. We would sometimes start mapping values before all defs were known,
and that could change a value from a simple 1-1 mapping to a multi-def mapping
that requires ssa update.
The new approach collects all defs and register assignments first without
filling in any live intervals. Only when finish() is called, do we compute
liveness and mapped values. At this time we know with certainty which values map
to multiple values in a split range.
This also has the advantage that we can compute live ranges based on the
remaining uses after rematerializing at split points.
The current implementation has many opportunities for compile time optimization.
llvm-svn: 124765
may be useful to understand "none", this is not the place for it. Tweak
the fix to Normalize while there: the fix added in 123990 works correctly,
but I like this way better. Finally, now that Triple understands some
non-trivial environment values, teach the unittests about them.
llvm-svn: 124720
the load, then it may be legal to transform the load and store to integer
load and store of the same width.
This is done if the target specified the transformation as profitable. e.g.
On arm, this can transform:
vldr.32 s0, []
vstr.32 s0, []
to
ldr r12, []
str r12, []
rdar://8944252
llvm-svn: 124708
Modified patch by Adam Preuss.
This builds on the existing framework for block tracing, edge profiling and optimal edge profiling.
See -help-hidden for new flags.
For documentation, see the technical report "Implementation of Path Profiling..." in llvm.org/pubs.
llvm-svn: 124515
benchmarks, and that it can be simplified to X/Y. (In general you can only
simplify (Z*Y)/Y to Z if the multiplication did not overflow; if Z has the
form "X/Y" then this is the case). This patch implements that transform and
moves some Div logic out of instcombine and into InstructionSimplify.
Unfortunately instcombine gets in the way somewhat, since it likes to change
(X/Y)*Y into X-(X rem Y), so I had to teach instcombine about this too.
Finally, thanks to the NSW/NUW flags, sometimes we know directly that "Z*Y"
does not overflow, because the flag says so, so I added that logic too. This
eliminates a bunch of divisions and subtractions in 447.dealII, and has good
effects on some other benchmarks too. It seems to have quite an effect on
tramp3d-v4 but it's hard to say if it's good or bad because inlining decisions
changed, resulting in massive changes all over.
llvm-svn: 124487
default implementation for x86, going through the stack in a similr
fashion to how the codegen implements BUILD_VECTOR. Eventually this
will get matched to VINSERTF128 if AVX is available.
llvm-svn: 124307
implementation of EXTRACT_SUBVECTOR for x86, going through the stack
in a similr fashion to how the codegen implements BUILD_VECTOR.
Eventually this will get matched to VEXTRACTF128 if AVX is available.
llvm-svn: 124292
a few loops accordingly. Should be no functional change.
This is a step for more accurate cost/benefit analysis of devirt/inlining
bonuses.
llvm-svn: 124275
This will be used to check patterns referencing a forthcoming
INSERT_SUBVECTOR SDNode and will also be used to check
EXTRACT_SUBVECTOR nodes.
llvm-svn: 124191
optimized code are:
(non-negative number)+(power-of-two) != 0 -> true
and
(x | 1) != 0 -> true
Instcombine knows about the second one of course, but only does it if X|1
has only one use. These fire thousands of times in the testsuite.
llvm-svn: 124183
with BasicAA's DecomposeGEPExpression, which recently began
using a TargetData. This fixes PR8968, though the testcase
is awkward to reduce.
Also, update several off GetUnderlyingObject's users
which happen to have a TargetData handy to pass it in.
llvm-svn: 124134
computation, the Ancestor field is always set to the Parent, so we can remove
the explicit link entirely and merge the Parent and Ancestor fields. Instead of
checking for whether an ancestor exists for a node or not, we simply check
whether the node has already been processed. This is simpler if Compress is
inlined into Eval, so I did that as well.
This is about a 3% speedup running -domtree on test-suite + SPEC2000 & SPEC2006,
but it also opens up some opportunities for further improvement.
llvm-svn: 124061
flags. They are still not enable in this revision.
Added TargetInstrInfo::isZeroCost() to fix a fundamental problem with
the scheduler's model of operand latency in the selection DAG.
Generalized unit tests to work with sched-cycles.
llvm-svn: 123969
TargetInstrInfo:
Change produceSameValue() to take MachineRegisterInfo as an optional argument.
When in SSA form, targets can use it to make more aggressive equality analysis.
Machine LICM:
1. Eliminate isLoadFromConstantMemory, use MI.isInvariantLoad instead.
2. Fix a bug which prevent CSE of instructions which are not re-materializable.
3. Use improved form of produceSameValue.
ARM:
1. Teach ARM produceSameValue to look pass some PIC labels.
2. Look for operands from different loads of different constant pool entries
which have same values.
3. Re-implement PIC GA materialization using movw + movt. Combine the pair with
a "add pc" or "ldr [pc]" to form pseudo instructions. This makes it possible
to re-materialize the instruction, allow machine LICM to hoist the set of
instructions out of the loop and make it possible to CSE them. It's a bit
hacky, but it significantly improve code quality.
4. Some minor bug fixes as well.
With the fixes, using movw + movt to materialize GAs significantly outperform the
load from constantpool method. 186.crafty and 255.vortex improved > 20%, 254.gap
and 176.gcc ~10%.
llvm-svn: 123905
checks enabled:
1) Use '<' to compare integers in a comparison function rather than '<='.
2) Use the uniqued set DefBlocks rather than Info.DefiningBlocks to initialize
the priority queue.
The speedup of scalarrepl on test-suite + SPEC2000 + SPEC2006 is a bit less, at
just under 16% rather than 17%.
llvm-svn: 123662
eliminating a potentially quadratic data structure, this also gives a 17%
speedup when running -scalarrepl on test-suite + SPEC2000 + SPEC2006. My initial
experiment gave a greater speedup around 25%, but I moved the dominator tree
level computation from dominator tree construction to PromoteMemToReg.
Since this approach to computing IDFs has a much lower overhead than the old
code using precomputed DFs, it is worth looking at using this new code for the
second scalarrepl pass as well.
llvm-svn: 123609
half a million non-local queries, each of which would otherwise have triggered a
linear scan over a basic block.
Also fix a fixme for memory intrinsics which dereference pointers. With this,
we prove that a pointer is non-null because it was dereferenced by an intrinsic
112 times in llvm-test.
llvm-svn: 123533
these would try hard to match constants by inverting the bits
and recursively matching. There are two problems with this:
1) some patterns would match when we didn't want them to (theoretical)
2) this is insanely expensive to do, and most often pointless.
This was apparently useful in just 2 instcombine cases, which I
added code to handle explicitly. This change speeds up 'opt'
time on 176.gcc by 1% and produces bitwise identical code.
llvm-svn: 123518
early in the cleanup code and one late interlaced with the inliner. The second one is
important because inlining and other scalar optzns can unpin allocas, allowing them to
be split up and promoted. While important for performance, this is also relatively
rare, and we would previously force a (non-lazy) computation of DomFrontiers, which
happened even if nothing became unpinned.
With this patch, the first pass of scalarrepl still promotes the vast bulk of allocas
in programs, but hte second pass has changed to use SSAUpdater, which is more "sparse"
and lazy. This speeds up opt -O3 time on kimwitu++ (a c++ app) by about 1%. The
numbers are interesting: the first pass promotes ~17500 allocas. The second pass
promotes about 1600. For non-C++ codes, the compile time win should be greater,
because the second pass of scalarrepl does less.
llvm-svn: 123437
- Fixed :upper16: fix up routine. It should be shifting down the top 16 bits first.
- Added support for Thumb2 :lower16: and :upper16: fix up.
- Added :upper16: and :lower16: relocation support to mach-o object writer.
llvm-svn: 123424
most important simplifications, as well as resolving phase ordering issues where instcombine
would inhibit important CSE'ing opportunities, for instance on BitBench/drop3.
llvm-svn: 123418
While there, I noticed that the transform "undef >>a X -> undef" was wrong.
For example if X is 2 then the top two bits must be equal, so the result can
not be anything. I fixed this in the constant folder as well. Also, I made
the transform for "X << undef" stronger: it now folds to undef always, even
though X might be zero. This is in accordance with the LangRef, but I must
admit that it is fairly aggressive. Also, I added "i32 X << 32 -> undef"
following the LangRef and the constant folder, likewise fairly aggressive.
llvm-svn: 123417
Add methods for accessing the (single) entry / exit edge of a region. If no such
edge exists, null is returned. Both accessors return the start block of the
corresponding edge. The edge can finally be formed by utilizing
Region::getEntry() or Region::getExit();
Contributed by: Andreas Simbuerger <simbuerg@fim.uni-passau.de>
llvm-svn: 123410
in the right direction. It eliminated some hacks and will unblock codegen
work. But it's far from being done. It doesn't reject illegal expressions,
e.g. (FOO - :lower16:BAR). It also doesn't work in Thumb2 mode at all.
llvm-svn: 123369
