indirectly requires a function analysis.
This bug was reported by Jason Kim. He included a test case here:
http://reviews.llvm.org/D3312
llvm-svn: 205753
Before, we would have conditional operators where one side of the
operator would be of type RelocationTypeAMD64 and the other is of type
RelocationTypeI386. GCC would noisly warn with -Wenum-compare
diagnostic.
Instead, refactor the code so it is more like the X86 ELF object writer.
llvm-svn: 205752
The IO normalizer would essentially lump I386 and AMD64 relocations
together. Relocation types with the same numeric value would then get
mapped in appropriately.
For example:
IMAGE_REL_AMD64_ADDR64 and IMAGE_REL_I386_DIR16 both have a numeric
value of one. We would see IMAGE_REL_I386_DIR16 in obj2yaml conversions
of object files with a machine type of IMAGE_FILE_MACHINE_AMD64.
llvm-svn: 205746
Fixes PR16365 - Extremely slow compilation in -O1 and -O2.
The SD scheduler has a quadratic implementation of load clustering
which absolutely blows up compile time for large blocks with constant
pool loads. The MI scheduler has a better implementation of load
clustering. However, we have not done the work yet to completely
eliminate the SD scheduler. Some benchmarks still seem to benefit from
early load clustering, although maybe by chance.
As an intermediate term fix, I just put a nice limit on the number of
DAG users to search before finding a match. With this limit there are no
binary differences in the LLVM test suite, and the PR16365 test case
does not suffer any compile time impact from this routine.
llvm-svn: 205738
Moving these patterns from TableGen files to PerformDAGCombine()
should allow us to generate better code by eliminating unnecessary
shifts and extensions earlier.
This also fixes a bug where the MAD pattern was calling
SimplifyDemandedBits with a 24-bit mask on the first operand
even when the full pattern wasn't being matched. This occasionally
resulted in some instructions being incorrectly deleted from the
program.
v2:
- Fix bug with 64-bit mul
llvm-svn: 205731
into a constant size alloca by inlining.
Ran a run over the testsuite, no results out of the noise, fixes
the testcase in the PR.
PR19115.
llvm-svn: 205710
- take->release: LLVM has moved to C++11. MockWrapper became an instance of unique_ptr.
- method symbol_iterator::increment disappeared recently, in this revision:
r200442 | rafael | 2014-01-29 20:49:50 -0600 (Wed, 29 Jan 2014) | 9 lines
Simplify the handling of iterators in ObjectFile.
None of the object file formats reported error on iterator increment. In
retrospect, that is not too surprising: no object format stores symbols or
sections in a linked list or other structure that requires chasing pointers.
As a consequence, all error checking can be done on begin() and end().
This reduces the text segment of bin/llvm-readobj in my machine from 521233 to
518526 bytes.
My change mimics the change that the revision made to lib/DebugInfo/DWARFContext.cpp .
- const_cast: Shut up a warning from gcc.
I ran unittests/ExecutionEngine/JIT/Debug+Asserts/JITTests to make sure it worked.
- Arch
llvm-svn: 205689
I really should read the spec more often (and test GCC more often too).
I just assumed that namespace aliases would be the same as using
directives, except with a name. But apparently that's not how the DWARF
standards suggests they be implemented. DWARF4 provides an example and
other non-normative text suggesting that namespace aliases be
implemented by named imported declarations intsead of named imported
modules.
So be it.
llvm-svn: 205685
Also update a few null pointers in this function to be consistent with
new null pointers being added.
Patch by Robert Matusewicz!
Differential Revision: http://reviews.llvm.org/D3123
llvm-svn: 205682
This adds a warning when linker_private or linker_private_weak is provided and
we handle it in a compatible manner.
Suggested by Chris Lattner!
llvm-svn: 205681
This consolidates the duplicated MachO checks in the directive parsing for
various directives that are unsupported for Mach-O. The error message change is
unimportant as this restores the behaviour to that prior to the addition of the
new directive handling. Furthermore, use a more direct check for MachO
targeting rather than an indirect feature check of the assembler.
Also simplify the test execution command to avoid temporary files. Further more,
perform the check in both object and assembly emission.
Whether all non-applicable directives are handled is another question. .fnstart
is marked as being unsupported, however, the complementary .fnend is not. The
additional unwinding directives are also still honoured. This change does not
change that, though, it would be good to validate and mark them as being
unsupported if they are unsupported for the MachO emission.
llvm-svn: 205678
This avoids an extra copy during decompression and avoids the use of
MemoryBuffer which is a weirdly esoteric device that includes unrelated
concepts like "file name" (its rather generic name is a bit misleading).
Similar refactoring of zlib::compress coming up.
llvm-svn: 205676
This restores the linker_private and linker_private_weak lexemes to permit
translation of the deprecated lexmes. The behaviour is identical to the bitcode
handling: linker_private and linker_private_weak are handled as if private had
been specified. This enables compatibility with IR generated by LLVM 3.4.
Reported on IRC by ki9a!
llvm-svn: 205675
This provides more realistic costs for the insert/extractelement instructions
(which are load/store pairs), accounts for the cheap unaligned Altivec load
sequence, and for unaligned VSX load/stores.
Bad news:
MultiSource/Applications/sgefa/sgefa - 35% slowdown (this will require more investigation)
SingleSource/Benchmarks/McGill/queens - 20% slowdown (we no longer vectorize this, but it was a constant store that was scalarized)
MultiSource/Benchmarks/FreeBench/pcompress2/pcompress2 - 2% slowdown
Good news:
SingleSource/Benchmarks/Shootout/ary3 - 54% speedup
SingleSource/Benchmarks/Shootout-C++/ary - 40% speedup
MultiSource/Benchmarks/Ptrdist/ks/ks - 35% speedup
MultiSource/Benchmarks/FreeBench/neural/neural - 30% speedup
MultiSource/Benchmarks/TSVC/Symbolics-flt/Symbolics-flt - 20% speedup
Unfortunately, estimating the costs of the stack-based scalarization sequences
is hard, and adjusting these costs is like a game of whac-a-mole :( I'll
revisit this again after we have better codegen for vector extloads and
truncstores and unaligned load/stores.
llvm-svn: 205658
This patch adds the Octeon cnMips instructions seqi/snei and v3mulu/vmm0/vmulu.
It is only for the assembler. Test case is included.
Reviewed by: Daniel.Sanders@imgtec.com
llvm-svn: 205631
gcc inline asm supports specifying "cc" as a clobber of all condition
registers. Add just enough modeling of the full register to make this work.
Fixed PR19326.
llvm-svn: 205630
Summary:
They behave in accordance with the Has2008 and ABS2008 configuration bits of the
processor which are used to select between the 1985 and 2008 versions of IEEE
754. In 1985 mode, these instructions are arithmetic (i.e. they raise invalid
operation exceptions when given NaN), in 2008 mode they are non-arithmetic
(i.e. they are copies).
nmadd.[ds], and nmsub.[ds] are still subject to -enable-no-nans-fp-math because
the ISA spec does not explicitly state that they obey Has2008 and ABS2008.
Reviewers: matheusalmeida
Reviewed By: matheusalmeida
Differential Revision: http://llvm-reviews.chandlerc.com/D3274
llvm-svn: 205628
When LLVM sees something like (v1iN (vselect v1i1, v1iN, v1iN)) it can
decide that the result is OK (v1i64 is legal on AArch64, for example)
but it still need scalarising because of that v1i1. There was no code
to do this though.
AArch64 and ARM64 have DAG combines to produce efficient code and
prevent that occuring in *most* such situations, but there are edge
cases that they miss. This adds a legalization to cope with that.
llvm-svn: 205626
There were several overlapping problems here, and this solution is
closely inspired by the one adopted in AArch64 in r201381.
Firstly, scalarisation of v1i1 setcc operations simply fails if the
input types are legal. This is fixed in LegalizeVectorTypes.cpp this
time, and allows AArch64 code to be simplified slightly.
Second, vselect with such a setcc feeding into it ends up in
ScalarizeVectorOperand, where it's not handled. I experimented with an
implementation, but found that whatever DAG came out was rather
horrific. I think Hao's DAG combine approach is a good one for
quality, though there are edge cases it won't catch (to be fixed
separately).
Should fix PR19335.
llvm-svn: 205625
Removed "GNU Assembler extension (compatibility)" definitions from ARMInstrInfo.td
Fixed ARMAsmParser::ParseInstruction GNU compatability branch, so it also works for thumb mode from now.
Added new tests.
llvm-svn: 205622
The previous patterns directly inserted FMOV or INS instructions into
the DAG for scalar_to_vector & bitconvert patterns. This is horribly
inefficient and can generated lots more GPR <-> FPR register traffic
than necessary.
It's much better to emit instructions the register allocator
understands so it can coalesce the copies when appropriate.
It led to at least one ISelLowering hack to avoid the problems, which
was incorrect for v1i64 (FPR64 has no dsub). It can now be removed
entirely.
This should also fix PR19331.
llvm-svn: 205616
Without this change, the llvm_unreachable kicked in. The code pattern
being spotted is rather non-canonical for 128-bit MLAs, but it can
happen and there's no point in generating sub-optimal code for it just
because it looks odd.
Should fix PR19332.
llvm-svn: 205615
recoloring cut-offs are encountered and register allocation failed.
This is related to PR18747
Patch by MAYUR PANDEY <mayur.p@samsung.com>.
llvm-svn: 205601
Removes unnecessary casts from non-generic address spaces to the generic address
space for certain code patterns.
Patch by Jingyue Wu.
llvm-svn: 205571
When rematerializing through truncates, the coalescer may produce instructions
with dead defs, but live implicit-defs of subregs:
E.g.
%X1<def,dead> = MOVi64imm 2, %W1<imp-def>; %X1:GPR64, %W1:GPR32
These instructions are live, and their definitions should not be rewritten.
Fixes <rdar://problem/16492408>
llvm-svn: 205565
Acording to AMD documentation, the correct opcode for
BFE_INT is 0x5, not 0x4
Fixes Arithm/Absdiff.Mat/3 OpenCV test
Patch by: Bruno Jiménez
llvm-svn: 205562
Implementing this via ComputeMaskedBits has two advantages:
+ It actually works. DAGISel doesn't deal with the chains properly
in the previous pattern-based solution, so they never trigger.
+ The information can be used in other DAG combines, as well as the
trivial "get rid of truncs". For example if the trunc is in a
different basic block.
rdar://problem/16227836
llvm-svn: 205540
The terminal barrier of a cmpxchg expansion will be either Acquire or
SequentiallyConsistent. In either case it can be skipped if the
operation has Monotonic requirements on failure.
rdar://problem/15996804
llvm-svn: 205535
Summary:
Adds the 'mips4' processor and a simple test of the ELF e_flags.
Patch by David Chisnall
His work was sponsored by: DARPA, AFRL
I made one small change to the testcase so that it uses
mips64-unknown-linux instead of mips4-unknown-linux.
This patch indirectly adds FeatureCondMov to FeatureMips64. This is ok
because it's supposed to be there anyway and it turns out that
FeatureCondMov is not a predicate of any instructions at the moment
(this is a bug that hasn't been noticed because there are no targets
without the conditional move instructions yet).
CC: theraven
Differential Revision: http://llvm-reviews.chandlerc.com/D3244
llvm-svn: 205530
llc doesn't generate nodes for unconditional fall-through branches for targets
without FastISel implementation (X86 has it, but can be disabled by
"-fast-isel=false") in SelectionDAGBuilder::visitBr().
So for line 4 in the following testcase
1: void foo(int i){
2: switch(i){
3: default:
4: break;
5: }
6: return;
7: }
there is no corresponding line in .debug_line section, and a debugger
cannot set a breakpoint at line 4.
Fix this by always emitting a branch when we're not optimizing and add a
testcase to ensure that there's code on every line we'd want to break.
Patch by Daniil Fukalov.
llvm-svn: 205529
The previous situation where ATOMIC_LOAD_WHATEVER nodes were expanded
at MachineInstr emission time had grown to be extremely large and
involved, to account for the subtly different code needed for the
various flavours (8/16/32/64 bit, cmpxchg/add/minmax).
Moving this transformation into the IR clears up the code
substantially, and makes future optimisations much easier:
1. an atomicrmw followed by using the *new* value can be more
efficient. As an IR pass, simple CSE could handle this
efficiently.
2. Making use of cmpxchg success/failure orderings only has to be done
in one (simpler) place.
3. The common "cmpxchg; did we store?" idiom can be exposed to
optimisation.
I intend to gradually improve this situation within the ARM backend
and make sure there are no hidden issues before moving the code out
into CodeGen to be shared with (at least ARM64/AArch64, though I think
PPC & Mips could benefit too).
llvm-svn: 205525
The trouble as in ARMAsmParser, in ParseInstruction method. It assumes that ARM::R12 + 1 == ARM::SP.
It is wrong, since ARM::<Register> codes are generated by tablegen and actually could be any random numbers.
llvm-svn: 205524
add operation since extract_vector_elt can perform an extend operation. Get the input lane
type from the vector on which we're performing the vpaddl operation on and extend or
truncate it to the output type of the original add node.
llvm-svn: 205523
%highest(sym1 - sym2 + const) relocations. Remove "ABS_" from VK_Mips_HI
and VK_Mips_LO enums in MipsMCExpr, to be consistent with VK_Mips_HIGHER
and VK_Mips_HIGHEST.
This change also deletes test file test/MC/Mips/higher_highest.ll and moves
its CHECK's to the new test file test/MC/Mips/higher-highest-addressing.s.
The deleted file tests that R_MIPS_HIGHER and R_MIPS_HIGHEST relocations are
emitted in the .o file. Since it uses -force-mips-long-branch option, it was
created when MipsLongBranch's implementation was emitting R_MIPS_HIGHER and
R_MIPS_HIGHEST relocations in the .o file. It was disabled when MipsLongBranch
started to directly calculate offsets.
Differential Revision: http://llvm-reviews.chandlerc.com/D3230
llvm-svn: 205522
Switching between i32 and i64 based on the LHS type is a good idea in
theory, but pre-legalisation uses i64 regardless of our choice,
leading to potential ISel errors.
Should fix PR19294.
llvm-svn: 205519
While we were encoding 64 bit values (data8) in the subrange itself,
using a 32 bit type for the subrange was still confusing the gdb. Oh,
and make it unsigned too.
As the comment points out, this could be pushed into the frontend so
that it would be 32 or 64 bit as appropriate, etc.
llvm-svn: 205512
I should have read that comment a little more carefully. ;)
Regression test in the works, committing in the mean time to un-break people.
llvm-svn: 205511
This has the following advantages:
* Less code.
* The old ELF implementation was wrong for non-relocatable objects.
* The old ELF implementation (and I think MachO) was wrong for thumb.
No current testcase since this is only used from MCJIT and it only uses
relocatable objects and I don't think it supports thumb yet.
llvm-svn: 205508
When a vector type legalizes to a larger vector type, and the target does not
support the associated extending load (or truncating store), then legalization
will scalarize the load (or store) resulting in an associated scalarization
cost. BasicTTI::getMemoryOpCost needs to account for this.
Between this, and r205487, PowerPC on the P7 with VSX enabled shows:
MultiSource/Benchmarks/PAQ8p/paq8p: 43% speedup
SingleSource/Benchmarks/BenchmarkGame/puzzle: 51% speedup
SingleSource/UnitTests/Vectorizer/gcc-loops 28% speedup
(some of these are new; some of these, such as PAQ8p, just reverse regressions
that VSX support would trigger)
llvm-svn: 205495
This reverts commit r205479.
It turns out that nm does use addresses, it is just that every reasonable
relocatable ELF object has sections with address 0. I have no idea if those
exist in reality, but it at least it shows that llvm-nm should use the name
address.
The added test was includes an unusual .o file with non 0 section addresses. I
created it by hacking ELFObjectWriter.cpp.
Really sorry for the churn.
llvm-svn: 205493
TargetInstrInfo::findCommutedOpIndices to enable VFMA*231 commutation, rather
than abusing commuteInstruction.
Thanks very much for the suggestion guys!
llvm-svn: 205489
For an cast (extension, etc.), the currently logic predicts a low cost if the
associated operation (keyed on the destination type) is legal (or promoted).
This is not true when the number of values required to legalize the type is
changing. For example, <8 x i16> being sign extended by <8 x i32> is not
generically cheap on PPC with VSX, even though sign extension to v4i32 is
legal, because two output v4i32 values are required compared to the single
v8i16 input value, and without custom logic in the target, this conversion will
scalarize.
llvm-svn: 205487
What llvm-nm prints depends on the file format. On ELF for example, if the
file is relocatable, it prints offsets. If it is not, it prints addresses.
Since it doesn't really need to care what it is that it is printing, use the
generic term value.
Fix or implement getSymbolValue to keep llvm-nm working.
llvm-svn: 205479
PPCTTI::getMemoryOpCost will now make use of BasicTTI::getMemoryOpCost to
calculate the base cost of the memory access, and then adjust on top of that.
There is no functionality change from this modification, but it will become
important so that PPCTTI can take advantage of scalarization information for which
BasicTTI::getMemoryOpCost will account in the near future.
llvm-svn: 205476
on FMA3 memory operands. FMA3 instructions are VEX encoded, so they can load
from unaligned memory.
Testcase to follow, along with related patch.
<rdar://problem/16478629>
llvm-svn: 205472
Just pass a MachineInstr reference rather than an MBB iterator.
Creating a MachineInstr& is the first thing every implementation did
anyway.
llvm-svn: 205453
Unlike other v6+ processors, cortex-m0 never supports unaligned accesses.
From the v6m ARM ARM:
"A3.2 Alignment support: ARMv6-M always generates a fault when an unaligned
access occurs."
rdar://16491560
llvm-svn: 205452
Adds the instructions ext/ext32/cins/cins32.
It also changes pop/dpop to accept the two operand version and
adds a simple pattern to generate baddu.
Tests for the two operand versions (including baddu/dmul/dpop/pop)
and the code generation pattern for baddu are included.
Reviewed by: Daniel.Sanders@imgtec.com
llvm-svn: 205449
Weak symbols cannot use the small code model's usual ADRP sequences since the
instruction simply may not be able to encode a value of 0.
This redirects them to use the GOT, which hopefully linkers are able to cope
with even in the static relocation model.
llvm-svn: 205426
Some Intrinsics are overloaded to the extent that return type equality (all
that's been checked up to now) does not guarantee that the arguments are the
same. In these cases SLP vectorizer should not recurse into the operands, which
can be achieved by comparing them as "Function *" rather than simply the ID.
llvm-svn: 205424
Again, coalescing and other optimisations swiftly made the MachineInstrs
consistent again, but when compiled at -O0 a bad INSERT_SUBREGISTER was
produced.
llvm-svn: 205423
The previous attempt was fine with optimisations, but was actually rather
cavalier with its types. When compiled at -O0, it produced invalid COPY
MachineInstrs.
llvm-svn: 205422
ARM specific optimiztion, finding places in ARM machine code where 2 dmbs
follow one another, and eliminating one of them.
Patch by Reinoud Elhorst.
llvm-svn: 205409
and isTargetCygwin() to isTargetWindowsCygwin() to be consistent with the
four Windows environments in Triple.h.
Suggestion by Saleem Abdulrasool!
llvm-svn: 205393
For the purpose of calculating the cost of the loop at various vectorization
factors, we need to count dependencies of consecutive pointers as uniforms
(which means that the VF = 1 cost is used for all overall VF values).
For example, the TSVC benchmark function s173 has:
...
%3 = add nsw i64 %indvars.iv, 16000
%arrayidx8 = getelementptr inbounds %struct.GlobalData* @global_data, i64 0, i32 0, i64 %3
...
and we must realize that the add will be a scalar in order to correctly deduce
it to be profitable to vectorize this on PowerPC with VSX enabled. In fact, all
dependencies of a consecutive pointer must be a scalar (uniform), and so we
simply need to add all consecutive pointers to the worklist that currently
detects collects uniforms.
Fixes PR19296.
llvm-svn: 205387
I'm not sure the comment in the implementation really adds a lot of
value (it's clear that we emit zero when no symbol is provided, but it
doesn't explain why we would do that). Happy to iterate.
llvm-svn: 205386
This removes the magic-number-esque code creating/retrieving the same
label for a debug_loc entry from two places and removes the last small
piece of reusable logic from emitDebugLoc so that there will be less
duplication when refactoring it into two functions (one for debug_loc,
the other for debug_loc.dwo).
llvm-svn: 205382
Seems we didn't have any test coverage for merging... awesome. So I
added some - but hit an llvm-objdump bug while I was there. I'm choosing
not to shave that yak right now.
Code review feedback/bug catch by Adrian Prantl in r205360.
llvm-svn: 205373
No test case (this would invoke UB by examining uninitialized members,
etc, at best - and this code is apparently untested anyway - I'm about
to fix that)
Code review feedback from Adrian Prantl on r205360.
llvm-svn: 205367
This provides an initial implementation of getUnrollingPreferences for x86.
getUnrollingPreferences is used by the generic (concatenation) unroller, which
is distinct from the unrolling done by the loop vectorizer. Many modern x86
cores have some kind of uop cache and loop-stream detector (LSD) used to
efficiently dispatch small loops, and taking full advantage of this requires
unrolling small loops (small here means 10s of uops).
These caches also have limits on the number of taken branches in the loop, and
so we also cap the loop unrolling factor based on the maximum "depth" of the
loop. This is currently calculated with a partial DFS traversal (partial
because it will stop early if the path length grows too much). This is still an
approximation, and one that is both conservative (because it does not account
for branches eliminated via block placement) and optimistic (because it is only
recording the maximum depth over minimum paths). Nevertheless, because the
loops that fit in these uop caches are so small, it is not clear how much the
details matter.
The original set of patches posted for review produced the following test-suite
performance results (from the TSVC benchmark) at that time:
ControlLoops-dbl - 13% speedup
ControlLoops-flt - 15% speedup
Reductions-dbl - 7.5% speedup
llvm-svn: 205348
In preparation for an upcoming commit implementing unrolling preferences for
x86, this adds additional fields to the UnrollingPreferences structure:
- PartialThreshold and PartialOptSizeThreshold - Like Threshold and
OptSizeThreshold, but used when not fully unrolling. These are necessary
because we need different thresholds for full unrolling from those used when
partially unrolling (the full unrolling thresholds are generally going to be
larger).
- MaxCount - A cap on the unrolling factor when partially unrolling. This can
be used by a target to prevent the unrolled loop from exceeding some
resource limit independent of the loop size (such as number of branches).
There should be no functionality change for any in-tree targets.
llvm-svn: 205347
The implementation of getUserCost had duplicated (and hard-coded) the default
logic in getGEPCost. Instead, it is better to use getGEPCost directly, which
limits the default logic to the implementation of one function, and allows
targets to override the behavior.
No functionality change intended.
llvm-svn: 205346
Identical to Win32 method except the GS segment register is used for TLS
instead of FS and pvArbitrary is at TEB offset 0x28 instead of 0x14.
llvm-svn: 205342
Elena Demikhovsky