Including following 14 instructions:
4 ld1 insts: load multiple 1-element structure to sequential 1/2/3/4 registers.
ld2/ld3/ld4: load multiple N-element structure to sequential N registers (N=2,3,4).
4 st1 insts: store multiple 1-element structure from sequential 1/2/3/4 registers.
st2/st3/st4: store multiple N-element structure from sequential N registers (N = 2,3,4).
E.g. ld1(3 registers version) will load 32-bit elements {A, B, C, D, E, F} sequentially into the three 64-bit vectors list {BA, DC, FE}.
E.g. ld3 will load 32-bit elements {A, B, C, D, E, F} into the three 64-bit vectors list {DA, EB, FC}.
llvm-svn: 192351
These IR instructions are undefined when the amount is equal to operand
size, but NEON right shifts support such shifts. Work around that by
emitting a different IR in these cases.
llvm-svn: 191953
Patch by Ana Pazos.
1.Added support for v1ix and v1fx types.
2.Added Scalar Pairwise Reduce instructions.
3.Added initial implementation of Scalar Arithmetic instructions.
llvm-svn: 191264
This restores the sqrt -> llvm.sqrt mapping, but only in fast-math mode
(specifically, when the UnsafeFPMath or NoNaNsFPMath CodeGen options are
enabled). The @llvm.sqrt* intrinsics have slightly different semantics from the
libm call, specifically, they are undefined when given a non-zero negative
number (the libm calls will always return NaN for any negative number).
This mapping was removed in r100613, and replaced with a TODO, but at that time
the fast-math flags were not yet implemented. Now that we have these, restoring
this mapping is important because it will enable autovectorization of sqrt
calls in loops (at least in fast-math mode).
llvm-svn: 190646
These operations "vector add high-half narrow" actually correspond to the
sequence:
%sum = add <4 x i32> %lhs, %rhs
%high = lshr <4 x i32> %sum, <i32 16, i32 16, i32 16, i32 16>
%res = trunc <4 x i32> %high to <4 x i16>
Now that LLVM can spot this, Clang should emit the corresponding LLVM IR.
llvm-svn: 189463
The NEON intrinsics vqdmlal and vqdmlsl are really just combinations of a
saturating-doubling-multiply (vqdmull) and a saturating add/sub, so now that
LLVM can spot those patterns Clang should emit them instead of specialised
intrinsics.
Feature already tested by existing ARM NEON intrinsics tests.
llvm-svn: 189462
Patch by Ana Pazos
- Completed implementation of instruction formats:
AdvSIMD three same
AdvSIMD modified immediate
AdvSIMD scalar pairwise
- Completed implementation of instruction classes
(some of the instructions in these classes
belong to yet unfinished instruction formats):
Vector Arithmetic
Vector Immediate
Vector Pairwise Arithmetic
- Initial implementation of instruction formats:
AdvSIMD scalar two-reg misc
AdvSIMD scalar three same
- Intial implementation of instruction class:
Scalar Arithmetic
- Initial clang changes to support arm v8 intrinsics.
Note: no clang changes for scalar intrinsics function name mangling yet.
- Comprehensive test cases for added instructions
To verify auto codegen, encoding, decoding, diagnosis, intrinsics.
llvm-svn: 187568
This patch provides basic support for powerpc64le as an LLVM target.
However, use of this target will not actually generate little-endian
code. Instead, use of the target will cause the correct little-endian
built-in defines to be generated, so that code that tests for
__LITTLE_ENDIAN__, for example, will be correctly parsed for
syntax-only testing. Code generation will otherwise be the same as
powerpc64 (big-endian), for now.
The patch leaves open the possibility of creating a little-endian
PowerPC64 back end, but there is no immediate intent to create such a
thing.
The new test case variant ensures that correct built-in defines for
little-endian code are generated.
llvm-svn: 187180
r186899 and r187061 added a preferred way for some architectures not to get
intrinsic generation for math builtins. So the code changes in r185568 can
now be undone (the test remains).
llvm-svn: 187079
This adds three overloaded intrinsics to Clang:
T __builtin_arm_ldrex(const volatile T *addr)
int __builtin_arm_strex(T val, volatile T *addr)
void __builtin_arm_clrex()
The intent is that these do what users would expect when given most sensible
types. Currently, "sensible" translates to ints, floats and pointers.
llvm-svn: 186394
& operator (ignoring any overloaded operator& for the type). The purpose of
this builtin is for use in std::addressof, to allow it to be made constexpr;
the existing implementation technique (reinterpret_cast to some reference type,
take address, reinterpert_cast back) does not permit this because
reinterpret_cast between reference types is not permitted in a constant
expression in C++11 onwards.
llvm-svn: 186053
Without fmath-errno, Clang currently generates calls to @llvm.pow.* intrinsics
when it sees pow*(). This may not be suitable for all targets (for
example le32/PNaCl), so the attached patch adds a target hook that CodeGen
queries. The target can state its preference for having or not having the
intrinsic generated. Non-PNaCl behavior remains unchanged;
PNaCl-specific test added.
llvm-svn: 185568
This will enable users in security critical applications to perform
checked-arithmetic in a fast safe manner that is amenable to c.
Tests/an update to Language Extensions is included as well.
rdar://13421498.
llvm-svn: 184497
I have had several people ask me about why this builtin was not available in
clang (since it seems like a logical conclusion). This patch implements said
builtins.
Relevant tests are included as well. I also updated the Clang language extension reference.
rdar://14192664.
llvm-svn: 184227
Current gcc's produce an error if __clear_cache is anything but
__clear_cache(char *a, char *b);
It looks like we had just implemented a gcc bug that is now fixed.
llvm-svn: 181784
aggregate types in a profoundly wrong way that has to be
worked around in every call site, to getEvaluationKind,
which classifies and distinguishes between all of these
cases.
Also, normalize the API for loading and storing complexes.
I'm working on a larger patch and wanted to pull these
changes out, but it would have be annoying to detangle
them from each other.
llvm-svn: 176656
calls and declarations.
LLVM has a default CC determined by the target triple. This is
not always the actual default CC for the ABI we've been asked to
target, and so we sometimes find ourselves annotating all user
functions with an explicit calling convention. Since these
calling conventions usually agree for the simple set of argument
types passed to most runtime functions, using the LLVM-default CC
in principle has no effect. However, the LLVM optimizer goes
into histrionics if it sees this kind of formal CC mismatch,
since it has no concept of CC compatibility. Therefore, if this
module happens to define the "runtime" function, or got LTO'ed
with such a definition, we can miscompile; so it's quite
important to get this right.
Defining runtime functions locally is quite common in embedded
applications.
llvm-svn: 176286
In ArrayRef<T>(X), X should not be temporary value. It could be rewritten more redundantly;
llvm::Type *XTy = X->getType();
ArrayRef<llvm::Type *> Ty(XTy);
llvm::Value *Callee = CGF.CGM.getIntrinsic(IntrinsicID, Ty);
Since it is safe if both XTy and Ty are temporary value in one statement, it could be shorten;
llvm::Value *Callee = CGF.CGM.getIntrinsic(IntrinsicID, ArrayRef<llvm::Type*>(X->getType()));
ArrayRef<T> has an implicit constructor to create uni-entry of T;
llvm::Value *Callee = CGF.CGM.getIntrinsic(IntrinsicID, X->getType());
MSVC-generated clang.exe crashed.
llvm-svn: 172352
PR 14529 was opened because neither Clang or LLVM was expanding
calls to creal* or cimag* into instructions that just load the
respective complex field. After some discussion, it was not
considered realistic to do this in LLVM because of the platform
specific way complex types are expanded. Thus a way to solve
this in Clang was pursued. GCC does a similar expansion.
This patch adds the feature to Clang by making the creal* and
cimag* functions library builtins and modifying the builtin code
generator to look for the new builtin types.
llvm-svn: 170455
uncovered.
This required manually correcting all of the incorrect main-module
headers I could find, and running the new llvm/utils/sort_includes.py
script over the files.
I also manually added quite a few missing headers that were uncovered by
shuffling the order or moving headers up to be main-module-headers.
llvm-svn: 169237
checks to enable. Remove frontend support for -fcatch-undefined-behavior,
-faddress-sanitizer and -fthread-sanitizer now that they don't do anything.
llvm-svn: 167413
GCC has always supported this on PowerPC and 4.8 supports it on all platforms,
so it's a good idea to expose it in clang too. LLVM supports this on all targets.
llvm-svn: 165362
The expression based expansion too often results in IR level optimizations
splitting the intermediate values into separate basic blocks, preventing
the formation of the VBSL instruction as the code author intended. In
particular, LICM would often hoist part of the computation out of a loop.
rdar://11011471
llvm-svn: 164342
the trap BB out of the individual checks and into a common function, to prepare
for making this code call into a runtime library. Rename the existing EmitCheck
to EmitTypeCheck to clarify it and to move it out of the way of the new
EmitCheck.
llvm-svn: 163451
The backend has to legalize i64 types by splitting them into two 32-bit pieces,
which leads to poor quality code. If we produce code for these intrinsics that
uses one-element vector types, which can live in Neon vector registers without
getting split up, then the generated code is much better. Radar 11998303.
llvm-svn: 161879
intrinsics. The second instruction(s) to be handled are the vector versions
of count set bits (ctpop).
The changes here are to clang so that it generates a target independent
vector ctpop when it sees an ARM dependent vector bits set count. The changes
in llvm are to match the target independent vector ctpop and in
VMCore/AutoUpgrade.cpp to update any existing bc files containing ARM
dependent vector pop counts with target-independent ctpops. There are also
changes to an existing test case in llvm for ARM vector count instructions and
to a test for the bitcode upgrade.
<rdar://problem/11892519>
There is deliberately no test for the change to clang, as so far as I know, no
consensus has been reached regarding how to test neon instructions in clang;
q.v. <rdar://problem/8762292>
llvm-svn: 160409
intrinsics with target-indepdent intrinsics. The first instruction(s) to be
handled are the vector versions of count leading zeros (ctlz).
The changes here are to clang so that it generates a target independent
vector ctlz when it sees an ARM dependent vector ctlz. The changes in llvm
are to match the target independent vector ctlz and in VMCore/AutoUpgrade.cpp
to update any existing bc files containing ARM dependent vector ctlzs with
target-independent ctlzs. There are also changes to an existing test case in
llvm for ARM vector count instructions and a new test for the bitcode upgrade.
<rdar://problem/11831778>
There is deliberately no test for the change to clang, as so far as I know, no
consensus has been reached regarding how to test neon instructions in clang;
q.v. <rdar://problem/8762292>
llvm-svn: 160201
in the ABI arrangement, and leave a hook behind so that we can easily
tweak CCs on platforms that use different CCs by default for C++
instance methods.
llvm-svn: 159894
The tablegen'd code does the same thing without this egregious duplication.
In my limited testing everything seems to work, however there can be
differences if the clang and llvm builtin definitions don't match.
llvm-svn: 159371
pointer, but such folding encounters side-effects, ignore the side-effects
rather than performing them at runtime: CodeGen generates wrong code for
__builtin_object_size in that case.
llvm-svn: 157310
r155047. See the LLVM log for the primary motivation:
http://llvm.org/viewvc/llvm-project?rev=155047&view=rev
Primary commit r154828:
- Several issues were raised in review, and fixed in subsequent
commits.
- Follow-up commits also reverted, and which should be folded into the
original before reposting:
- r154837: Re-add the 'undef BUILTIN' thing to fix the build.
- r154928: Fix build warnings, re-add (and correct) header and
license
- r154937: Typo fix.
Please resubmit this patch with the relevant LLVM resubmission.
llvm-svn: 155048
__atomic_test_and_set, __atomic_clear, plus a pile of undocumented __GCC_*
predefined macros.
Implement library fallback for __atomic_is_lock_free and
__c11_atomic_is_lock_free, and implement __atomic_always_lock_free.
Contrary to their documentation, GCC's __atomic_fetch_add family don't
multiply the operand by sizeof(T) when operating on a pointer type.
libstdc++ relies on this quirk. Remove this handling for all but the
__c11_atomic_fetch_add and __c11_atomic_fetch_sub builtins.
Contrary to their documentation, __atomic_test_and_set and __atomic_clear
take a first argument of type 'volatile void *', not 'void *' or 'bool *',
and __atomic_is_lock_free and __atomic_always_lock_free have an argument
of type 'const volatile void *', not 'void *'.
With this change, libstdc++4.7's <atomic> passes libc++'s atomic test suite,
except for a couple of libstdc++ bugs and some cases where libc++'s test
suite tests for properties which implementations have latitude to vary.
llvm-svn: 154640
<stdatomic.h> header.
In passing, fix LanguageExtensions to note that C11 and C++11 are no longer
"upcoming standards" but are now actually standardized.
llvm-svn: 154513
match the behavior of GCC. Also add a test for these intrinsics, which
apparently have *zero* tests. =[ Not surprisingly, Clang crashed when
compiling these.
Fix the bug in CodeGen where we failed to bitcast the argument type to
x86mmx prior to calling the LLVM intrinsic. This fixes an assert on the
new 3dnow-builtins.c test.
This is one issue impacting the efforts to get Clang to emulate the
Microsoft intrinsics headers -- 3dnow intrinsics are implictitly made
available there.
llvm-svn: 150948
We had been generating load/store instructions with the default alignment
for the vector element type, even when the pointer argument had less alignment.
<rdar://problem/10538555>
llvm-svn: 149794
ARM supports clz and ctz directly and both operations have well-defined
results for zero. There is no disadvantage in performance to using the
defined-at-zero versions of llvm.ctlz/cttz intrinsics. We're running into
ARM-specific code written with the assumption that __builtin_clz(0) == 32,
even though that value is technically undefined. The code is failing now
because of llvm optimizations that are taking advantage of the undef
behavior (specifically svn r147255). There's nothing wrong with that
optimization on x86 where any incorrect assumptions about __builtin_clz(0)
will quickly be exposed. For ARM, though, optimizations based on that undef
behavior are likely to cause subtle bugs. Other targets with defined-at-zero
clz/ctz support may want to override the default behavior as well.
llvm-svn: 149086
"use the new ConstantVector::getSplat method where it makes sense."
Also simplify a bunch of code to use the Builder->getInt32 instead
of doing it the hard and ugly way. Much more progress could be made
here, but I don't plan to do it.
llvm-svn: 148926
setting the is_zero_undef flag appropriately to true as that matches the
semantics of these GCC builtins.
This is the Clang side of r146357 in LLVM.
llvm-svn: 146358
This patch just adds a simple NeonTypeFlags class to replace the various
hardcoded constants that had been used until now. Unfortunately I couldn't
figure out a good way to avoid duplicating that class between clang and
TableGen, but since it's small and rarely changes, that's not so bad.
llvm-svn: 144054
implicitly perform an lvalue-to-rvalue conversion if used on an lvalue
expression. Also improve the documentation of Expr::Evaluate* to indicate which
of them will accept expressions with side-effects.
llvm-svn: 143263
- Remodel Expr::EvaluateAsInt to behave like the other EvaluateAs* functions,
and add Expr::EvaluateKnownConstInt to capture the current fold-or-assert
behaviour.
- Factor out evaluation of bitfield bit widths.
- Fix a few places which would evaluate an expression twice: once to determine
whether it is a constant expression, then again to get the value.
llvm-svn: 141561
There are still a few issues which need to be resolved with code generation for atomic load and store, so I'm not converting the places which need those for now.
I'm not entirely sure what to do about __builtin_llvm_memory_barrier: the fence instruction doesn't expose all the possibilities which can be expressed by __builtin_llvm_memory_barrier. I would appreciate hearing from anyone who is using this intrinsic.
llvm-svn: 139216
Unlike most of the other Neon intrinsics, these are not overloaded and do not
have the extra argument that specifies the vector type. This has not been
fatal because the lane number operand is supposed to be an ICE and so that
value has harmlessly been used as the type identifier. Radar 9901281.
llvm-svn: 137550
Sorry, this was a bad idea. Within clang these builtins are in a separate
"ARM" namespace, but the actual builtin names should clearly distinguish tha
they are target specific.
llvm-svn: 133833
(__m128){ p[0], p[1], p[2], p[3] }
which produces really bad code. This could be done in instcombine, but it's
probably better to do it in the front-end instead.
<rdar://problem/9424836>
llvm-svn: 131237
are trivial. This exposes opportunities earlier, and allows fastisel
to do good things with these at -O0.
This addresses rdar://9289468 - clang doesn't fold memset_chk at -O0
llvm-svn: 129651
clang's -Wuninitialized-experimental warning.
While these don't look like real bugs, clang's
-Wuninitialized-experimental analysis is stricter
than GCC's, and these fixes have the benefit
of being general nice cleanups.
llvm-svn: 124072
delete the block we began emitting into if it had no predecessors. We never
want to do this, because there are several valid cases during statement
emission where an existing block has no known predecessors but will acquire
some later. The case in my test case doesn't inherently fall into this
category, because we could safely emit the case-range code before the statement
body, but there are examples with labels that can't be fallen into
that would also demonstrate this bug.
rdar://problem/8837067
llvm-svn: 123303
Also tweak the VCVT_F32_F16 entry in arm_neon.td to be more consistent with
the other floating-point conversion builtins. Radar 8068427.
llvm-svn: 121916
__builtin_ia32_vec_init_v8qi
__builtin_ia32_vec_init_v4hi
__builtin_ia32_vec_init_v2si
They are lowered to bitcasts. (These are all ready tested by the gcc testsuite.)
<rdar://problem/8529957>
llvm-svn: 116147
force it to be a constant instead of emitting with EmitScalarExpr. In
-ftrapv mode, they are not the same.
This fixes rdar://8478728 + PR8221
llvm-svn: 115388
one of them) was causing a series of failures:
http://google1.osuosl.org:8011/builders/clang-x86_64-darwin10-selfhost/builds/4518
svn merge -c -114929 https://llvm.org/svn/llvm-project/cfe/trunk
--- Reverse-merging r114929 into '.':
U include/clang/Sema/Sema.h
U include/clang/AST/DeclCXX.h
U lib/Sema/SemaDeclCXX.cpp
U lib/Sema/SemaTemplateInstantiateDecl.cpp
U lib/Sema/SemaDecl.cpp
U lib/Sema/SemaTemplateInstantiate.cpp
U lib/AST/DeclCXX.cpp
svn merge -c -114925 https://llvm.org/svn/llvm-project/cfe/trunk
--- Reverse-merging r114925 into '.':
G include/clang/AST/DeclCXX.h
G lib/Sema/SemaDeclCXX.cpp
G lib/AST/DeclCXX.cpp
svn merge -c -114924 https://llvm.org/svn/llvm-project/cfe/trunk
--- Reverse-merging r114924 into '.':
G include/clang/AST/DeclCXX.h
G lib/Sema/SemaDeclCXX.cpp
G lib/Sema/SemaDecl.cpp
G lib/AST/DeclCXX.cpp
U lib/AST/ASTContext.cpp
svn merge -c -114921 https://llvm.org/svn/llvm-project/cfe/trunk
--- Reverse-merging r114921 into '.':
G include/clang/AST/DeclCXX.h
G lib/Sema/SemaDeclCXX.cpp
G lib/Sema/SemaDecl.cpp
G lib/AST/DeclCXX.cpp
llvm-svn: 114933
with zext/sext operations, instead of to llvm intrinsics. I have a plan to
avoid the clang builtins for these, but it is going to take a little longer
and I want to get the NEON intrinsics updated before the 2.8 release.
llvm-svn: 112764
with zext/sext operations, instead of to llvm intrinsics. (We can also
get rid of the clang builtins and handle these entirely in the arm_neon.h
header if there is a way to express vector sext/zext in C.)
llvm-svn: 112413
their call expressions synthetically have the "deduced" types based on their
first argument. We only insert conversions in the AST for arguments whose
values require conversion to match the value type expected. This keeps PR7600
closed by maintaining the return type, but avoids assertions due to unexpected
implicit casts making the type unsigned (test case added from Daniel).
The magic is moved into the codegen for the atomic builtin which inserts the
casts as needed at the IR level to raise the type to an integer suitable for
the LLVM intrinsic. This shouldn't cause any real change in functionality, but
now we can make the builtin be more truly polymorphic.
llvm-svn: 108638
have CGF create and make accessible standard int32,int64 and
intptr types. This fixes a ton of 80 column violations
introduced by LLVMContextification and cleans up stuff a lot.
llvm-svn: 106977
objective-c++ class objects which have GC'able objc object
pointers and need to use ObjC's objc_memmove_collectable
API (radar 8070772).
llvm-svn: 106061
Tim Northover