We currently support them only in AArch64. The NEON Reference,
however, says they are 'ARMv7, ARMv8' intrinsics.
Differential Revision: https://reviews.llvm.org/D47446
llvm-svn: 334362
I'd like to make the select builtins require an avx512f, avx512bw, or avx512vl fature to match what is normally required to get masking. Truncate is special in that there are instructions with a 128/256-bit masked result even without avx512vl.
By using special buitlins we can emit a select without using the 128/256-bit select builtins.
llvm-svn: 334331
I'm looking into making the select builtins require avx512f, avx512bw, or avx512vl since masking operations generally require those features.
The extract builtins are funny because the 512-bit versions return a 128 or 256 bit vector with masking even when avx512vl is not supported.
llvm-svn: 334330
Test changes are due to differences in how we generate undef elements now. We also changed the types used for extractf128_si256/insertf128_si256 to match the signature of the builtin that previously existed which this patch resurrects. This also matches gcc.
llvm-svn: 334261
Adds support for these intrinsics, which are ARM and ARM64 only:
_interlockedbittestandreset_acq
_interlockedbittestandreset_rel
_interlockedbittestandreset_nf
_interlockedbittestandset_acq
_interlockedbittestandset_rel
_interlockedbittestandset_nf
Refactor the bittest intrinsic handling to decompose each intrinsic into
its action, its width, and its atomicity.
llvm-svn: 334239
We still emit shufflevector instructions we just do it from CGBuiltin.cpp now. This ensures the intrinsics that use this are only available on CPUs that support the feature.
I also added range checking to the immediate, but only checked it is 8 bits or smaller. We should maybe be stricter since we never use all 8 bits, but gcc doesn't seem to do that.
llvm-svn: 334237
We still lower them to native shuffle IR, but we do it in CGBuiltin.cpp now. This allows us to check the target feature and ensure the immediate fits in 8 bits.
This also improves our -O0 codegen slightly because we're able to see the zeroinitializer in the shuffle. It looks like it got lost behind a store+load previously.
llvm-svn: 334208
Summary:
We recently switch to using a selects in the intrinsics header files for FMA instructions. But the 512-bit versions support flavors with rounding mode which must be an Integer Constant Expression. This has forced those intrinsics to be implemented as macros. As it stands now the mask and mask3 intrinsics evaluate one of their macro arguments twice. If that argument itself is another intrinsic macro, we can end up over expanding macros. Or if its something we can CSE later it would show up multiple times when it shouldn't.
I tried adding __extension__ around the macro and making it an expression statement and declaring a local variable. But whatever name you choose for the local variable can never be used as the name of an input to the macro in user code. If that happens you would end up with the same name on the LHS and RHS of an assignment after expansion. We might be safe if we use __ in front of the variable names because those names are reserved and user code shouldn't use that, but I wasn't sure I wanted to make that claim.
The other option which I've chosen here, is to add back _mask, _maskz, and _mask3 flavors of the builtin which we will expand in CGBuiltin.cpp to replicate the argument as needed and insert any fneg needed on the third operand to make a subtract. The _maskz isn't truly necessary if we have an unmasked version or if we use the masked version with a -1 mask and wrap a select around it. But I've chosen to make things more uniform.
I separated out the scalar builtin handling to avoid too many things going on in EmitX86FMAExpr. It was different enough due to the extract and insert that the minor duplication of the CreateCall was probably worth it.
Reviewers: tkrupa, RKSimon, spatel, GBuella
Reviewed By: tkrupa
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D47724
llvm-svn: 334159
Factor out the common setjmp call emission code.
Based on a patch by Chris January
Differential Revision: https://reviews.llvm.org/D47784
llvm-svn: 334112
I tested these locally on an x86 machine by disabling the inline asm
codepath and confirming that it does the same bitflips as we do with the
inline asm.
Addresses code review feedback.
llvm-svn: 334059
Previously we were just using extended vector operations in the header file.
This unfortunately allowed non-constant indices to be used with the intrinsics. This is incompatible with gcc, icc, and MSVC. It also introduces a different performance characteristic because non-constant index gets lowered to a vector store and an element sized load.
By adding the builtins we can check for the index to be a constant and ensure its in range of the vector element count.
User code still has the option to use extended vector operations themselves if they need non-constant indexing.
llvm-svn: 334057
This builtin takes an index as its second operand, but the codegen hardcodes an index of 0 and doesn't use the operand. The only use of the builtin in the header file passes 0 to the operand so this works for that usage. But its more correct to use the real operand.
llvm-svn: 334054
We need to implement _interlockedbittestandset as a builtin for
windows.h, so we might as well do the whole family. It reduces code
duplication anyway.
Fixes PR33188, a long standing bug in our bittest implementation
encountered by Chakra.
llvm-svn: 333978
Adding __attribute__((aligned(32))) to __m256 breaks the implementation
of _mm256_loadu_ps on Windows. On Windows, alignment attributes have
higher precedence than packing attributes.
We also might want to carefully consider the consequences of changing
our vector typedefs, since many users copy them and invent their own
new, non-Intel specific vector type names.
llvm-svn: 333958
This seems like a premature optimization. It's unlikely a user would pass something the frontend can tell is all ones to the masked load/store intrinsics.
We do this optimization for emitting select for masking because we have builtin calls in header files that pass an all ones mask in. Though at this point we may not longer have any builtins that emit some IR and a select. We may only have the select builtins so maybe we can remove that optimization too.
llvm-svn: 333847
We currently support them only in AArch64. The NEON Reference,
however, says they are 'ARMv7, ARMv8' intrinsics.
Differential Revision: https://reviews.llvm.org/D47121
llvm-svn: 333829
This fixes two major problems:
- We were not capping vector alignment as desired on 32-bit ARM.
- We were using different alignments based on the AVX settings on
Intel, so we did not have a consistent ABI.
This is an ABI break, but we think we can get away with it because
vectors tend to be used mostly in inline code (which is why not having
a consistent ABI has not proven disastrous on Intel).
Intel's AVX types are specified as having 32-byte / 64-byte alignment,
so align them explicitly instead of relying on the base ABI rule.
Note that this sort of attribute is stripped from template arguments
in template substitution, so there's a possibility that code templated
over vectors will produce inadequately-aligned objects. The right
long-term solution for this is for alignment attributes to be
interpreted as true qualifiers and thus preserved in the canonical type.
llvm-svn: 333791
The WebAssembly committee has decided on the names `memory.size` and
`memory.grow` for the memory intrinsics, so update the clang builtin
functions to follow those names, keeping both sets of old names in place
for compatibility.
llvm-svn: 333712
This patch replaces all packed (and scalar without rounding
mode) fused intrinsics with fmadd/fmaddsub variations.
Then fmadd/fmaddsub are lowered to native IR.
Patch by tkrupa
Reviewers: craig.topper, sroland, spatel, RKSimon
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D47444
llvm-svn: 333555
These intrinsics are used by MSVC's header files on AArch64 Windows as
well as AArch32, so we should support them for both targets. I've
factored them out of CodeGenFunction::EmitARMBuiltinExpr into separate
functions that EmitAArch64BuiltinExpr can call as well.
Reviewers: javed.absar, mstorsjo
Reviewed By: mstorsjo
Subscribers: kristof.beyls, cfe-commits
Differential Revision: https://reviews.llvm.org/D47476
llvm-svn: 333513
The clang builtins have the same semantics as the stdlib functions.
The stdlib functions are defined in section 7.20.6.1 of the C standard with:
"If the result cannot be represented, the behavior is undefined."
That lets us mark the negation with 'nsw' because "sub i32 0, INT_MIN" would
be UB/poison.
Differential Revision: https://reviews.llvm.org/D47202
llvm-svn: 333038
Because the intrinsics in the headers are implemented as macros, we can't just use a select builtin and pternlog builtin. This would require one of the macro arguments to be used twice. Depending on what was passed to the macro we could expand an expression twice leading to weird behavior. We could maybe declare our local variable in the macro, but that would need to worry about name collisions.
To avoid that just generate IR directly in CGBuiltin.cpp.
Differential Revision: https://reviews.llvm.org/D47125
llvm-svn: 332891
1. added restrictions to memory scope, order and volatile parameters
2. added custom processing for these builtins - currently is not used code,
needed to switch off GCCBuiltin link to the builtins (ongoing change to llvm
tree)
3. builtins renamed as requested
Differential Revision: https://reviews.llvm.org/D43281
llvm-svn: 332848
This is unnecessary for AVX512VL supporting CPUs like SKX. We can just emit a 128-bit masked load/store here no matter what. The backend will widen it to 512-bits on KNL CPUs.
Fixes the frontend portion of PR37386. Need to fix the backend to optimize the new sequences well.
llvm-svn: 331958
Previously we emitted something like
rotl(x, n) {
n &= bitwidth-1;
return n != 0 ? ((x << n) | (x >> (bitwidth - n)) : x;
}
We use a select to avoid the undefined behavior on the (bitwidth - n) shift.
The middle and backend don't really recognize this as a rotate and end up emitting a cmov or control flow because of the select.
A better pattern is (x << (n & mask)) | (x << (-n & mask)) where mask is bitwidth - 1.
Fixes the main complaint in PR37387. There's still some work to be done if the user writes that sequence directly on a short or char where type promotion rules can prevent it from being recognized. The builtin is emitting direct IR with unpromoted types so that isn't a problem for it.
Differential Revision: https://reviews.llvm.org/D46656
llvm-svn: 331943
This is similar to the LLVM change https://reviews.llvm.org/D46290.
We've been running doxygen with the autobrief option for a couple of
years now. This makes the \brief markers into our comments
redundant. Since they are a visual distraction and we don't want to
encourage more \brief markers in new code either, this patch removes
them all.
Patch produced by
for i in $(git grep -l '\@brief'); do perl -pi -e 's/\@brief //g' $i & done
for i in $(git grep -l '\\brief'); do perl -pi -e 's/\\brief //g' $i & done
Differential Revision: https://reviews.llvm.org/D46320
llvm-svn: 331834
The ACLE spec which describes these intrinsics hasn't been published yet, but
this is based on the final draft which will be published soon, and these have
already been implemented by GCC.
Differential revision: https://reviews.llvm.org/D46109
llvm-svn: 331039
SPIR-V encodes the read_only and write_only access qualifiers of pipes,
so separate LLVM IR types are required to target SPIR-V. Other backends
may also find this useful.
These new types are `opencl.pipe_ro_t` and `opencl.pipe_wo_t`, which
replace `opencl.pipe_t`.
This replaces __get_pipe_num_packets(...) and __get_pipe_max_packets(...)
which took a read_only pipe with separate versions for read_only and
write_only pipes, namely:
* __get_pipe_num_packets_ro(...)
* __get_pipe_num_packets_wo(...)
* __get_pipe_max_packets_ro(...)
* __get_pipe_max_packets_wo(...)
These separate versions exist to avoid needing a bitcast to one of the
two qualified pipe types.
Patch by Stuart Brady.
Differential Revision: https://reviews.llvm.org/D46015
llvm-svn: 331026
This is the patch that lowers x86 intrinsics to native IR
in order to enable optimizations.
Patch by tkrupa
Differential Revision: https://reviews.llvm.org/D44786
llvm-svn: 330323
Summary:
A clang builtin for xray typed events. Differs from
__xray_customevent(...) by the presence of a type tag that is vended by
compiler-rt in typical usage. This allows xray handlers to expand logged
events with their type description and plugins to process traced events
based on type.
This change depends on D45633 for the intrinsic definition.
Reviewers: dberris, pelikan, rnk, eizan
Subscribers: cfe-commits, llvm-commits
Differential Revision: https://reviews.llvm.org/D45716
llvm-svn: 330220
Summary:
This change addresses http://llvm.org/PR36926 by allowing users to pick
which instrumentation bundles to use, when instrumenting with XRay. In
particular, the flag `-fxray-instrumentation-bundle=` has four valid
values:
- `all`: the default, emits all instrumentation kinds
- `none`: equivalent to -fnoxray-instrument
- `function`: emits the entry/exit instrumentation
- `custom`: emits the custom event instrumentation
These can be combined either as comma-separated values, or as
repeated flag values.
Reviewers: echristo, kpw, eizan, pelikan
Reviewed By: pelikan
Subscribers: mgorny, cfe-commits
Differential Revision: https://reviews.llvm.org/D44970
llvm-svn: 329985
I believe all the pieces are now in place in the backend to make this work correctly. We can either mask the input to 32 bits for pmuludg or shl/ashr for pmuldq and use a regular mul instruction. The backend should combine this to PMULUDQ/PMULDQ and then SimplifyDemandedBits will remove the and/shifts.
Differential Revision: https://reviews.llvm.org/D45421
llvm-svn: 329605
Found via codespell -q 3 -I ../clang-whitelist.txt
Where whitelist consists of:
archtype
cas
classs
checkk
compres
definit
frome
iff
inteval
ith
lod
methode
nd
optin
ot
pres
statics
te
thru
Patch by luzpaz! (This is a subset of D44188 that applies cleanly with a few
files that have dubious fixes reverted.)
Differential revision: https://reviews.llvm.org/D44188
llvm-svn: 329399
A recent addition to Coroutines TS (https://wg21.link/p0913) adds a pre-defined
coroutine noop_coroutine that does nothing. To implement this feature, we implemented
an llvm.coro.noop intrinsic that returns a coroutine handle to a coroutine that
does nothing when resumed or destroyed.
This patch adds a builtin __builtin_coro_noop() that maps to llvm.coro.noop intrinsic.
Related llvm change: https://reviews.llvm.org/D45114
llvm-svn: 328993
The conversion of operatios to bitcode helps to eliminate an additional
store in certain cases. We used to lower these load intrinsics in DAG to
DAG conversion by which time, the "Dead Store Elimination" pass is
already run. There is an associated LLVM patch.
Patch by Sumanth Gundapaneni.
llvm-svn: 328776
These instructions have been around for a long time, but we
haven't supported intrinsics for them. The "new" vesrions use
the CSx register for the start of the buffer instead of the K
field in the Mx register.
There is a related llvm patch.
Patch by Brendon Cahoon.
llvm-svn: 328725
Putting back the code in commit r327189 that was reverted in r322737. The code is being committed in three stages and this one is the last stage: 1) r327455 fp16 feature flags, 2) r327836 pass half type or i16 based on FullFP16, and 3) the code here which the front-end fp16 vector intrinsic for ARM.
Differential revision https://reviews.llvm.org/D43650
llvm-svn: 328277
This is needed for the upcoming implementation of the
new 8x32x16 and 32x8x16 variants of WMMA instructions
introduced in CUDA 9.1.
Differential Revision: https://reviews.llvm.org/D44719
llvm-svn: 328158
Summary:
Libc++'s default allocator uses `__builtin_operator_new` and `__builtin_operator_delete` in order to allow the calls to new/delete to be ellided. However, libc++ now needs to support over-aligned types in the default allocator. In order to support this without disabling the existing optimization Clang needs to support calling the aligned new overloads from the builtins.
See llvm.org/PR22634 for more information about the libc++ bug.
This patch changes `__builtin_operator_new`/`__builtin_operator_delete` to call any usual `operator new`/`operator delete` function. It does this by performing overload resolution with the arguments passed to the builtin to determine which allocation function to call. If the selected function is not a usual allocation function a diagnostic is issued.
One open issue is if the `align_val_t` overloads should be considered "usual" when `LangOpts::AlignedAllocation` is disabled.
In order to allow libc++ to detect this new behavior the value for `__has_builtin(__builtin_operator_new)` has been updated to `201802`.
Reviewers: rsmith, majnemer, aaron.ballman, erik.pilkington, bogner, ahatanak
Reviewed By: rsmith
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D43047
llvm-svn: 328134
This way we can support address-space specific variants without explicitly
encoding the space in the name of the intrinsic. Less intrinsics to deal with ->
less boilerplate.
Added a bit of tablegen magic to match/replace an intrinsics with a pointer
argument in particular address space with the space-specific instruction
variant.
Updated tests to use non-default address spaces.
Differential Revision: https://reviews.llvm.org/D43268
llvm-svn: 328006
For generating NEON intrinsics, this determines the NEON data type, and whether
it should be a half type or an i16 type. I.e., we always pass a half type for
AArch64, this hasn't changed, but now also for ARM but only when FullFP16 is
enabled, and i16 otherwise.
This is intended to be non-functional change, but together with the backend
work in D44538 which adds support for f16 vectors, this enables adding the
AArch32 FP16 (vector) intrinsics.
Differential Revision: https://reviews.llvm.org/D44561
llvm-svn: 327836
The MSVC runtime library does not provide a definition of wmemcmp,
so we need an inline implementation.
Differential Revision: https://reviews.llvm.org/D42441
llvm-svn: 323362
Summary:
kunpck intrinsics were removed in favor of native IR a few months ago. The implementation lowers them as by operation on the integer types passed to the intrinsic and then just shifting, masking, and oring them together. A special X86 DAG combine was added to recognize this patter and turn it into a concat_vector operation.
I think it makes more sense to keep the IR implementation closer to vector operations on vXi1. Given that we expect these builtins to be used around other builtins that operate on k-registers which we try to represent in IR with vXi1. InstCombine should be able to get rid of the bitcasts between integers and vXi1 leaving only the vector operations.
Reviewers: RKSimon, spatel, zvi, jina.nahias
Reviewed By: RKSimon
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D42016
llvm-svn: 322461
These just overloads for _Float128. They're supported by GCC 7 and used
by glibc. APFloat support is already there so just add the overloads.
__builtin_copysignf128
__builtin_fabsf128
__builtin_huge_valf128
__builtin_inff128
__builtin_nanf128
__builtin_nansf128
This is the same support that GCC has, according to the documentation,
but limited to _Float128.
llvm-svn: 321948
r320902 fixed the IRGen for some types of checked multiplications. It
did not handle unsigned overflow correctly in the case where the signed
operand is negative (PR35750).
Eli pointed out that on overflow, the result must be equal to the unique
value that is equivalent to the mathematically-correct result modulo two
raised to the k power, where k is the number of bits in the result type.
This patch fixes the specialized IRGen from r320902 accordingly.
Testing: Apart from check-clang, I modified the test harness from
r320902 to validate the results of all multiplications -- not just the
ones which don't overflow:
https://gist.github.com/vedantk/3eb9c88f82e5c32f2e590555b4af5081
llvm.org/PR35750, rdar://34963321
Differential Revision: https://reviews.llvm.org/D41717
llvm-svn: 321771
Diagnose 'unreachable' UB when a noreturn function returns.
1. Insert a check at the end of functions marked noreturn.
2. A decl may be marked noreturn in the caller TU, but not marked in
the TU where it's defined. To diagnose this scenario, strip away the
noreturn attribute on the callee and insert check after calls to it.
Testing: check-clang, check-ubsan, check-ubsan-minimal, D40700
rdar://33660464
Differential Revision: https://reviews.llvm.org/D40698
llvm-svn: 321231
This patch introduces a specialized way to lower overflow-checked
multiplications with mixed-sign operands. This fixes link failures and
ICEs on code like this:
void mul(int64_t a, uint64_t b) {
int64_t res;
__builtin_mul_overflow(a, b, &res);
}
The generic checked-binop irgen would use a 65-bit multiplication
intrinsic here, which requires runtime support for _muloti4 (128-bit
multiplication), and therefore fails to link on i386. To get an ICE
on x86_64, change the example to use __int128_t / __uint128_t.
Adding runtime and backend support for 65-bit or 129-bit checked
multiplication on all of our supported targets is infeasible.
This patch solves the problem by using simpler, specialized irgen for
the mixed-sign case.
llvm.org/PR34920, rdar://34963321
Testing: Apart from check-clang, I compared the output from this fairly
comprehensive test driver using unpatched & patched clangs:
https://gist.github.com/vedantk/3eb9c88f82e5c32f2e590555b4af5081
Differential Revision: https://reviews.llvm.org/D41149
llvm-svn: 320902
Summary:
InterlockedCompareExchange128 is a bit more complicated than the other
InterlockedCompareExchange functions, so it requires a bit more work. It
doesn't directly refer to 128bit ints, instead it takes pointers to
64bit ints for Destination and ComparandResult, and exchange is taken as
two 64bit ints (high & low). The previous value is written to
ComparandResult, and success is returned. This implementation does the
following in order to produce a cmpxchg instruction:
1. Cast everything to 128bit ints or int pointers, and glues together
the Exchange values
2. Reads from CompareandResult to get the comparand
3. Calls cmpxchg volatile (on X86 this will produce a lock cmpxchg16b
instruction)
1. Result 0 (previous value) is written back to ComparandResult
2. Result 1 (success bool) is zext'ed to a uchar and returned
Resolves bug https://llvm.org/PR35251
Patch by Colden Cullen!
Reviewers: rnk, agutowski
Reviewed By: rnk
Subscribers: majnemer, cfe-commits
Differential Revision: https://reviews.llvm.org/D41032
llvm-svn: 320730
There are 20 LLVM math intrinsics that correspond to mathlib calls according to the LangRef:
http://llvm.org/docs/LangRef.html#standard-c-library-intrinsics
We were only converting 3 mathlib calls (sqrt, fma, pow) and 12 builtin calls (ceil, copysign,
fabs, floor, fma, fmax, fmin, nearbyint, pow, rint, round, trunc) to their intrinsic-equivalents.
This patch pulls the transforms together and handles all 20 cases. The switch is guarded by a
check for const-ness to make sure we're not doing the transform if errno could possibly be set by
the libcall or builtin.
Differential Revision: https://reviews.llvm.org/D40044
llvm-svn: 319593
Summary:
The -fxray-always-emit-customevents flag instructs clang to always emit
the LLVM IR for calls to the `__xray_customevent(...)` built-in
function. The default behaviour currently respects whether the function
has an `[[clang::xray_never_instrument]]` attribute, and thus not lower
the appropriate IR code for the custom event built-in.
This change allows users calling through to the
`__xray_customevent(...)` built-in to always see those calls lowered to
the corresponding LLVM IR to lay down instrumentation points for these
custom event calls.
Using this flag enables us to emit even just the user-provided custom
events even while never instrumenting the start/end of the function
where they appear. This is useful in cases where "phase markers" using
__xray_customevent(...) can have very few instructions, must never be
instrumented when entered/exited.
Reviewers: rnk, dblaikie, kpw
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D40601
llvm-svn: 319388
LLVM exposes a file in the backend (X86TargetParser.def) that
contains information about the correct list of CpuIs values.
This patch removes 2 of the copied and pasted versions of this
list from clang and instead includes the data from the .def file.
Differential Revision: https://reviews.llvm.org/D40054
llvm-svn: 318234
cbrt() is always constant because it can't overflow or underflow. Therefore, it can't set errno.
fma() is not always constant because it can overflow or underflow. Therefore, it can set errno.
But we know that it never sets errno on GNU / MSVC, so make it constant in those environments.
Differential Revision: https://reviews.llvm.org/D39641
llvm-svn: 318093
Summary:
This just seems to have been an oversight. We already supported the f64
atomic add with an explicit scope (e.g. "cta"), but not the scopeless
version.
Reviewers: tra
Subscribers: jholewinski, sanjoy, cfe-commits, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D39638
llvm-svn: 317623
The LLVM sqrt intrinsic definition changed with:
D28797
...so we don't have to use any relaxed FP settings other than errno handling.
This patch sidesteps a question raised in PR27435:
https://bugs.llvm.org/show_bug.cgi?id=27435
Is a programmer using __builtin_sqrt() invoking the compiler's intrinsic definition of sqrt or the mathlib definition of sqrt?
But we have an answer now: the builtin should match the behavior of the libm function including errno handling.
Differential Revision: https://reviews.llvm.org/D39204
llvm-svn: 317031
In OpenCL the kernel function and non-kernel function has different calling conventions.
For certain targets they have different argument ABIs. Also kernels have special function
attributes and metadata for runtime to launch them.
The blocks passed to enqueue_kernel is supposed to be executed as kernels. As such,
the block invoke function should be emitted as kernel with proper calling convention and
argument ABI.
This patch emits enqueued block as kernel. If a block is both called directly and passed
to enqueue_kernel, separate functions will be generated.
Differential Revision: https://reviews.llvm.org/D38134
llvm-svn: 315804
Ivan A. Kosarev