When we use mask compare intrinsics under strict FP option, the masked
elements shouldn't raise any exception. So, we cann't replace the
intrinsic with a full compare + "and" operation.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D85385
This patch simplified IR generation for __builtin_btf_type_id().
For __builtin_btf_type_id(obj, flag), previously IR builtin
looks like
if (obj is a lvalue)
llvm.bpf.btf.type.id(obj.ptr, 1, flag) !type
else
llvm.bpf.btf.type.id(obj, 0, flag) !type
The purpose of the 2nd argument is to differentiate
__builtin_btf_type_id(obj, flag) where obj is a lvalue
vs.
__builtin_btf_type_id(obj.ptr, flag)
Note that obj or obj.ptr is never used by the backend
and the `obj` argument is only used to derive the type.
This code sequence is subject to potential llvm CSE when
- obj is the same .e.g., nullptr
- flag is the same
- metadata type is different, e.g., typedef of struct "s"
and strust "s".
In the above, we don't want CSE since their metadata is different.
This patch change IR builtin to
llvm.bpf.btf.type.id(seq_num, flag) !type
and seq_num is always increasing. This will prevent potential
llvm CSE.
Also report an error if the type name is empty for
remote relocation since remote relocation needs non-empty
type name to do relocation against vmlinux.
Differential Revision: https://reviews.llvm.org/D85174
This patch added the following additional compile-once
run-everywhere (CO-RE) relocations:
- existence/size of typedef, struct/union or enum type
- enum value and enum value existence
These additional relocations will make CO-RE bpf programs more
adaptive for potential kernel internal data structure changes.
For existence/size relocations, the following two code patterns
are supported:
1. uint32_t __builtin_preserve_type_info(*(<type> *)0, flag);
2. <type> var;
uint32_t __builtin_preserve_field_info(var, flag);
flag = 0 for existence relocation and flag = 1 for size relocation.
For enum value existence and enum value relocations, the following code
pattern is supported:
uint64_t __builtin_preserve_enum_value(*(<enum_type> *)<enum_value>,
flag);
flag = 0 means existence relocation and flag = 1 for enum value.
relocation. In the above <enum_type> can be an enum type or
a typedef to enum type. The <enum_value> needs to be an enumerator
value from the same enum type. The return type is uint64_t to
permit potential 64bit enumerator values.
Differential Revision: https://reviews.llvm.org/D83242
Specified in https://github.com/WebAssembly/simd/pull/237, these
instructions load the first vector lane from memory and zero the other
lanes. Since these instructions are not officially part of the SIMD
proposal, they are only available on an opt-in basis via LLVM
intrinsics and clang builtin functions. If these instructions are
merged to the proposal, this implementation will change so that the
instructions will be generated from normal IR. At that point the
intrinsics and builtin functions would be removed.
This PR also changes the opcodes for the experimental f32x4.qfm{a,s}
instructions because their opcodes conflicted with those of the
v128.load{32,64}_zero instructions. The new opcodes were chosen to
match those used in V8.
Differential Revision: https://reviews.llvm.org/D84820
fptosi/fptoui have similar, but not identical, semantics. In
particular, the behavior on overflow is different.
Fixes https://bugs.llvm.org/show_bug.cgi?id=46844 for 64-bit. (The
corresponding patch for 32-bit is more involved because the equivalent
intrinsics don't exist, as far as I can tell.)
Differential Revision: https://reviews.llvm.org/D84703
Instead, pattern match extends of extract_subvectors to generate
widening operations. Since extract_subvector is not a legal node, this
is implemented via a custom combine that recognizes extract_subvector
nodes before they are legalized. The combine produces custom ISD nodes
that are later pattern matched directly, just like the intrinsic was.
Also removes the clang builtins for these operations since the
instructions can now be generated from portable code sequences.
Differential Revision: https://reviews.llvm.org/D84556
Reapply 49e5f603d4
which had been reverted in c94332919b.
Originally reverted because I hadn't updated it in quite a while when I
got around to committing it, so there were a bunch of missing changes to
new code since I'd written the patch.
Reviewers: aaron.ballman
Differential Revision: https://reviews.llvm.org/D76646
Some of the system registers readable on AArch64 and ARM platforms
return different values with each read (for example a timer counter),
these shouldn't be hoisted outside loops or otherwise interfered with,
but the normal @llvm.read_register intrinsic is only considered to read
memory.
This introduces a separate @llvm.read_volatile_register intrinsic and
maps all system-registers on ARM platforms to use it for the
__builtin_arm_rsr calls. Registers declared with asm("r9") or similar
are unaffected.
There is a version that just tests (also called
isIntegerConstantExpression) & whereas this version is specifically used
when the value is of interest (a few call sites were actually refactored
to calling the test-only version) so let's make the API look more like
it.
Reviewers: aaron.ballman
Differential Revision: https://reviews.llvm.org/D76646
Remove _COMPAT. Drop the ARCHNAME. Remove the non-COMPAT versions
that are no longer needed.
We now only use these macros in places where we need compatibility
with libgcc/compiler-rt. So we don't need to call out _COMPAT
specifically.
When using __sync_nand_and_fetch with __int128, a problem is found that
the wrong value for the 'invert' value gets emitted to the xor in case
where the int size is greater than 64 bits.
This is because uses of llvm::ConstantInt::get which zero extends the
greater than 64 bits, so instead -1 that we require, it end up
getting 18446744073709551615
This patch replaces the call to llvm::ConstantInt::get with the call
to llvm::Constant::getAllOnesValue which works for all integer types.
Reviewers: jfp, erichkeane, rjmccall, hfinkel
Differential Revision: https://reviews.llvm.org/D82832
This covers both the existing memory functions as well as the new bulk memory proposal.
Added new test files since changes where also required in the inputs.
Also removes unused init/drop intrinsics rather than trying to make them work for 64-bit.
Differential Revision: https://reviews.llvm.org/D82821
This replaces the switch statement implementation in the clang's
X86.cpp with a lookup table in X86TargetParser.cpp.
I've used constexpr and copy of the FeatureBitset from
SubtargetFeature.h to store the features in a lookup table.
After the lookup the bitset is translated into strings for use
by the rest of the frontend code.
I had to modify the implementation of the FeatureBitset to avoid
bugs in gcc 5.5 constexpr handling. It seems to not like the
same array entry to be used on the left side and right hand side
of an assignment or &= or |=. I've also used uint32_t instead of
uint64_t and sized based on the X86::CPU_FEATURE_MAX.
I've initialized the features for different CPUs outside of the
table so that we can express inheritance in an adhoc way. This
was one of the big limitations of the switch and we had resorted
to labels and gotos.
Differential Revision: https://reviews.llvm.org/D82731
This change enables PowerPC compiler builtins to generate constrained
floating point operations when clang is indicated to do so.
A couple of possibly unexpected backend divergences between constrained
floating point and regular behavior are highlighted under the test tag
FIXME-CHECK. This may be something for those on the PPC backend to look
at.
Patch by: Drew Wock <drew.wock@sas.com>
Differential Revision: https://reviews.llvm.org/D82020
This was orignally done so we could separate the compatibility
values and the llvm internal only features into a separate entries
in the feature array. This was needed when we explicitly had to
convert the feature into the proper 32-bit chunk at every reference
and we didn't want things moving around.
Now everything is in an array and we have helper funtions or macros
to convert encoding to index. So we renumbering is no longer an
issue.
Currently, in order to extract an element from a bf16 vector, we cast
the vector to an i16 vector, perform the extraction, and cast the result to
bfloat. This behavior was copied from the old fp16 implementation.
The goal of this patch is to achieve optimal code generation for lane
copying intrinsics in a subsequent patch (LLVM fails to fold certain
combinations of bitcast, insertelement, extractelement and
shufflevector instructions leading to the generation of suboptimal code).
Differential Revision: https://reviews.llvm.org/D82206
Add a new builtin-function __builtin_expect_with_probability and
intrinsic llvm.expect.with.probability.
The interface is __builtin_expect_with_probability(long expr, long
expected, double probability).
It is mainly the same as __builtin_expect besides one more argument
indicating the probability of expression equal to expected value. The
probability should be a constant floating-point expression and be in
range [0.0, 1.0] inclusive.
It is similar to builtin-expect-with-probability function in GCC
built-in functions.
Differential Revision: https://reviews.llvm.org/D79830
The struct store intrinsics in LLVM IR take the individual parts
as arguments, so this patch uses the intrinsics used for `svget`
to break the tuples into individual parts.
Reviewers: c-rhodes, efriedma, ctetreau, david-arm
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81466
Summary:
The new SVE builtin type __SVBFloat16_t` is used to represent scalable
vectors of bfloat elements.
Reviewers: sdesmalen, efriedma, stuij, ctetreau, shafik, rengolin
Subscribers: tschuett, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81304
This patch add __builtin_matrix_column_major_store to Clang,
as described in clang/docs/MatrixTypes.rst. In the initial version,
the stride is not optional yet.
Reviewers: rjmccall, jfb, rsmith, Bigcheese
Reviewed By: rjmccall
Differential Revision: https://reviews.llvm.org/D72782
For example:
svint32_t svget4(svint32x4_t tuple, uint64_t imm_index)
returns the subvector at `index`, which must be in range `0..3`.
svint32x3_t svset3(svint32x3_t tuple, uint64_t index, svint32_t vec)
returns a tuple vector with `vec` inserted into `tuple` at `index`,
which must be in range `0..2`.
Reviewers: c-rhodes, efriedma
Reviewed By: c-rhodes
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81464
This patch add __builtin_matrix_column_major_load to Clang,
as described in clang/docs/MatrixTypes.rst. In the initial version,
the stride is not optional yet.
Reviewers: rjmccall, rsmith, jfb, Bigcheese
Reviewed By: rjmccall
Differential Revision: https://reviews.llvm.org/D72781
This patch upstreams support for BFloat Matrix Multiplication Intrinsics
and Code Generation from __bf16 to AArch64. This includes IR intrinsics. Unittests are
provided as needed. AArch32 Intrinsics + CodeGen will come after this
patch.
This patch is part of a series implementing the Bfloat16 extension of
the
Armv8.6-a architecture, as detailed here:
https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-architecture-developments-armv8-6-a
The bfloat type, and its properties are specified in the Arm
Architecture
Reference Manual:
https://developer.arm.com/docs/ddi0487/latest/arm-architecture-reference-manual-armv8-for-armv8-a-architecture-profile
The following people contributed to this patch:
Luke Geeson
- Momchil Velikov
- Mikhail Maltsev
- Luke Cheeseman
Reviewers: SjoerdMeijer, t.p.northover, sdesmalen, labrinea, miyuki,
stuij
Reviewed By: miyuki, stuij
Subscribers: kristof.beyls, hiraditya, danielkiss, cfe-commits,
llvm-commits, miyuki, chill, pbarrio, stuij
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D80752
Change-Id: I174f0fd0f600d04e3799b06a7da88973c6c0703f
_ExtInt types
- Fix computed size for _ExtInt types passed to checked arithmetic
builtins.
- Emit diagnostic when signed _ExtInt larger than 128-bits is passed
to __builtin_mul_overflow.
- Change Sema checks for builtins to accept placeholder types.
Differential Revision: https://reviews.llvm.org/D81420
This patch adds new SVE types to Clang that describe tuples of SVE
vectors. For example `svint32x2_t` which maps to the twice-as-wide
vector `<vscale x 8 x i32>`. Similarly, `svint32x3_t` will map to
`<vscale x 12 x i32>`.
It also adds builtins to return an `undef` vector for a given
SVE type.
Reviewers: c-rhodes, david-arm, ctetreau, efriedma, rengolin
Reviewed By: c-rhodes
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81459
Summary:
As specified in https://github.com/WebAssembly/simd/pull/232. These
instructions are implemented as LLVM intrinsics for now rather than
normal ISel patterns to make these instructions opt-in. Once the
instructions are merged to the spec proposal, the intrinsics will be
replaced with proper ISel patterns.
Reviewers: aheejin
Subscribers: dschuff, sbc100, jgravelle-google, hiraditya, sunfish, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D81222
Summary:
__builtin_amdgcn_atomic_inc32(int *Ptr, int Val, unsigned MemoryOrdering, const char *SyncScope)
__builtin_amdgcn_atomic_inc64(int64_t *Ptr, int64_t Val, unsigned MemoryOrdering, const char *SyncScope)
__builtin_amdgcn_atomic_dec32(int *Ptr, int Val, unsigned MemoryOrdering, const char *SyncScope)
__builtin_amdgcn_atomic_dec64(int64_t *Ptr, int64_t Val, unsigned MemoryOrdering, const char *SyncScope)
First and second arguments gets transparently passed to the amdgcn atomic
inc/dec intrinsic. Fifth argument of the intrinsic is set as true if the
first argument of the builtin is a volatile pointer. The third argument of
this builtin is one of the memory-ordering specifiers ATOMIC_ACQUIRE,
ATOMIC_RELEASE, ATOMIC_ACQ_REL, or ATOMIC_SEQ_CST following C++11 memory
model semantics. This is mapped to corresponding LLVM atomic memory ordering
for the atomic inc/dec instruction using CLANG atomic C ABI. The fourth
argument is an AMDGPU-specific synchronization scope defined as string.
Reviewers: arsenm, sameerds, JonChesterfield, jdoerfert
Reviewed By: arsenm, sameerds
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, jfb, kerbowa, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D80804
This patch add __builtin_matrix_transpose to Clang, as described in
clang/docs/MatrixTypes.rst.
Reviewers: rjmccall, jfb, rsmith, Bigcheese
Reviewed By: rjmccall
Differential Revision: https://reviews.llvm.org/D72778
Summary:
Add -ftrivial-auto-var-init-stop-after= to limit the number of times
stack variables are initialized when -ftrivial-auto-var-init= is used to
initialize stack variables to zero or a pattern. This flag can be used
to bisect uninitialized uses of a stack variable exposed by automatic
variable initialization, such as http://crrev.com/c/2020401.
Reviewers: jfb, vitalybuka, kcc, glider, rsmith, rjmccall, pcc, eugenis, vlad.tsyrklevich
Reviewed By: jfb
Subscribers: phosek, hubert.reinterpretcast, srhines, MaskRay, george.burgess.iv, dexonsmith, inglorion, gbiv, llozano, manojgupta, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77168
Summary:
This patch upstreams support for a new storage only bfloat16 C type.
This type is used to implement primitive support for bfloat16 data, in
line with the Bfloat16 extension of the Armv8.6-a architecture, as
detailed here:
https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-architecture-developments-armv8-6-a
The bfloat type, and its properties are specified in the Arm Architecture
Reference Manual:
https://developer.arm.com/docs/ddi0487/latest/arm-architecture-reference-manual-armv8-for-armv8-a-architecture-profile
In detail this patch:
- introduces an opaque, storage-only C-type __bf16, which introduces a new bfloat IR type.
This is part of a patch series, starting with command-line and Bfloat16
assembly support. The subsequent patches will upstream intrinsics
support for BFloat16, followed by Matrix Multiplication and the
remaining Virtualization features of the armv8.6-a architecture.
The following people contributed to this patch:
- Luke Cheeseman
- Momchil Velikov
- Alexandros Lamprineas
- Luke Geeson
- Simon Tatham
- Ties Stuij
Reviewers: SjoerdMeijer, rjmccall, rsmith, liutianle, RKSimon, craig.topper, jfb, LukeGeeson, fpetrogalli
Reviewed By: SjoerdMeijer
Subscribers: labrinea, majnemer, asmith, dexonsmith, kristof.beyls, arphaman, danielkiss, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76077
This is a re-revert with a corrected test.
This patch adds a test for the PowerPC fma compiler builtins, some variations
of which negate inputs and outputs. The code to generate IR for these
builtins was untested before this patch.
Originally, the code used the outdated method of subtracting floating point
values from -0.0 as floating point negation. This patch remedies that.
Patch by: Drew Wock <drew.wock@sas.com>
Differential Revision: https://reviews.llvm.org/D76949
Summary:
During CodeGen for AArch64 Neon intrinsics, Clang was incorrectly
assuming all the pointers from which loads were being generated for vld1
intrinsics were aligned according to the intrinsics result type, causing
alignment faults on the code generated by the backend.
This patch updates vld1 intrinsics' CodeGen to properly capture the
correct load alignment based on the type of the pointer provided as
input for the intrinsic.
Reviewers: t.p.northover, ostannard, pcc, efriedma
Reviewed By: ostannard, efriedma
Subscribers: echristo, plotfi, nickdesaulniers, efriedma, kristof.beyls, danielkiss, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79721
The os_log helper functions are linkonce_odr and supposed to be
uniqued across TUs, so attachine a DW_AT_decl_line on it is highly
misleading. By setting the function decl to implicit, CGDebugInfo
properly marks the functions as artificial and uses a default file /
line 0 location for the function.
rdar://problem/63450824
Differential Revision: https://reviews.llvm.org/D80463
Summary:
During CodeGen for AArch64 Neon intrinsics, Clang was incorrectly
assuming all the pointers from which loads were being generated for vld1
intrinsics were aligned according to the intrinsics result type, causing
alignment faults on the code generated by the backend.
This patch updates vld1 intrinsics' CodeGen to properly capture the
correct load alignment based on the type of the pointer provided as
input for the intrinsic.
Reviewers: t.p.northover, ostannard, pcc
Reviewed By: ostannard
Subscribers: kristof.beyls, danielkiss, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79721
If we're going to assume references are dereferenceable, we should also
assume they're aligned: otherwise, we can't actually dereference them.
See also D80072.
Differential Revision: https://reviews.llvm.org/D80166
Such a builtin function is mostly useful to preserve btf type id
for non-global data. For example,
extern void foo(..., void *data, int size);
int test(...) {
struct t { int a; int b; int c; } d;
d.a = ...; d.b = ...; d.c = ...;
foo(..., &d, sizeof(d));
}
The function "foo" in the above only see raw data and does not
know what type of the data is. In certain cases, e.g., logging,
the additional type information will help pretty print.
This patch implemented a BPF specific builtin
u32 btf_type_id = __builtin_btf_type_id(param, flag)
which will return a btf type id for the "param".
flag == 0 will indicate a BTF local relocation,
which means btf type_id only adjusted when bpf program BTF changes.
flag == 1 will indicate a BTF remote relocation,
which means btf type_id is adjusted against linux kernel or
future other entities.
Differential Revision: https://reviews.llvm.org/D74668
Summary:
As proposed in https://github.com/WebAssembly/simd/pull/122. Since
these instructions are not yet merged to the SIMD spec proposal, this
patch makes them entirely opt-in by surfacing them only through LLVM
intrinsics and clang builtins. If these instructions are made
official, these intrinsics and builtins should be replaced with simple
instruction patterns.
Reviewers: aheejin
Subscribers: dschuff, sbc100, jgravelle-google, hiraditya, sunfish, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D79742
Summary:
Although using `__builtin_shufflevector` and the `shufflevector`
instruction works fine, they are not opaque to the optimizer. As a
result, DAGCombine can potentially reduce the number of shuffles and
change the shuffle masks. This is unexpected behavior for users of the
WebAssembly SIMD intrinsics who have crafted their shuffles to
optimize the code generated by engines. This patch solves the problem
by adding a new shuffle intrinsic that is opaque to the optimizers in
line with the decision of the WebAssembly SIMD contributors at
https://github.com/WebAssembly/simd/issues/196#issuecomment-622494748. In
the future we may implement custom DAG combines to properly optimize
shuffles and replace this solution.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, hiraditya, sunfish, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D66983
These builtins are expanded in CGBuiltin to use intrinsics
for (signed/unsigned) shift left long top/bottom.
Reviewers: efriedma, SjoerdMeijer
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D79579
passed to __builtin_os_log_format to extend its lifetime to the end of
its enclosing block
Extend only lifetimes of pointers returned by function calls or message
sends instead. In the long term, we should lifetime-extend pointers in
more complex expressions and non-ARC objects (e.g., C++ temporaries)
too.
rdar://problem/61846261
The reinterpret builtins are generated separately because they
need the cross product of all types, 121 functions in total,
which is inconvenient to specify in the arm_sve.td file.
Reviewers: SjoerdMeijer, efriedma, ctetreau, rengolin
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78756
* svdupq builtins that duplicate scalars to every quadword of a vector
are defined using builtins for svld1rq (load and replicate quadword).
* svdupq builtins that duplicate boolean values to fill a predicate vector
are defined using `svcmpne`.
Reviewers: SjoerdMeijer, efriedma, ctetreau
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78750
Summary:
As described in https://github.com/WebAssembly/simd/pull/209. This is
the final reorganization of the SIMD opcode space before
standardization. It has been landed in concert with corresponding
changes in other projects in the WebAssembly SIMD ecosystem.
Reviewers: aheejin
Subscribers: dschuff, sbc100, jgravelle-google, hiraditya, sunfish, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79224
Summary:
Removes usage of VectorType::getNumElements identified by test located
at CodeGen/aarch64-sve-intrinsics/acle_sve_abs.c. Since the type is an
SVE predicate vector, it makes sense to specialize the code for scalable
vectors only.
Reviewers: rengolin, efriedma
Reviewed By: efriedma
Subscribers: tschuett, kristof.beyls, rkruppe, psnobl, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78958
The svlen builtins return the number of elements in a vector
and are implemented using `llvm.vscale`.
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D78755
Some ACLE builtins leave out the argument to specify the predicate
pattern, which is expected to be expanded to an SV_ALL pattern.
This patch adds the flag IsInsertOp1SVALL to insert SV_ALL as the
second operand.
Reviewers: efriedma, SjoerdMeijer
Reviewed By: SjoerdMeijer
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78401
Expose llvm fence instruction as clang builtin for AMDGPU target
__builtin_amdgcn_fence(unsigned int memoryOrdering, const char *syncScope)
The first argument of this builtin is one of the memory-ordering specifiers
__ATOMIC_ACQUIRE, __ATOMIC_RELEASE, __ATOMIC_ACQ_REL, or __ATOMIC_SEQ_CST
following C++11 memory model semantics. This is mapped to corresponding
LLVM atomic memory ordering for the fence instruction using LLVM atomic C
ABI. The second argument is an AMDGPU-specific synchronization scope
defined as string.
Reviewed By: sameerds
Differential Revision: https://reviews.llvm.org/D75917
Some ACLE builtins leave out the argument to specify the predicate
pattern, which is expected to be expanded to an SV_ALL pattern.
This patch adds the flag IsAppendSVALL to append SV_ALL as the final
operand.
Reviewers: SjoerdMeijer, efriedma, rovka, rengolin
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77597
This intrinsic isn't overloaded so we should query with types.
Doing so causes the backend to miss the intrinsic and not codegen it.
This eventually leads to a linker error.
This patch upstreams support for the Armv8.6-a Matrix Multiplication
Extension. A summary of the features can be found here:
https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-architecture-developments-armv8-6-a
This patch includes:
- Assembly support for AArch32
- Intrinsics Support for AArch32 Neon Intrinsics for Matrix
Multiplication
Note: these extensions are optional in the 8.6a architecture and so have
to be enabled by default
No additional IR types or C Types are needed for this extension.
This is part of a patch series, starting with BFloat16 support and
the other components in the armv8.6a extension (in previous patches
linked in phabricator)
Based on work by:
- Luke Geeson
- Oliver Stannard
- Luke Cheeseman
Reviewers: t.p.northover, miyuki
Reviewed By: miyuki
Subscribers: miyuki, ostannard, kristof.beyls, hiraditya, danielkiss,
cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77872
This patch upstreams support for the Armv8.6-a Matrix Multiplication
Extension. A summary of the features can be found here:
https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-architecture-developments-armv8-6-a
This patch includes:
- Assembly support for AArch64 only (no SVE or Neon)
- Intrinsics Support for AArch64 Armv8.6a Matrix Multiplication Instructions (No bfloat16 matrix multiplication)
No IR types or C Types are needed for this extension.
This is part of a patch series, starting with BFloat16 support and
the other components in the armv8.6a extension (in previous patches
linked in phabricator)
Based on work by:
- Luke Geeson
- Oliver Stannard
- Luke Cheeseman
Reviewers: ostannard, t.p.northover, rengolin, kmclaughlin
Reviewed By: kmclaughlin
Subscribers: kmclaughlin, kristof.beyls, hiraditya, danielkiss,
cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77871
The IsReverseCompare flag tells CGBuiltin to swap the operands,
so that a LT/LE intrinsics can be expressed in terms of GE/GT
intrinsics.
This patch also adds builtins for the wide-variants of the compares.
Reviewers: SjoerdMeijer, efriedma, ctetreau
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78747
This patch also adds the enum `sv_prfop` for the prefetch operation specifier
and checks to ensure the passed enum values are valid.
Reviewers: SjoerdMeijer, efriedma, ctetreau
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78674
This patch changes the codegen of the builtins for contiguous loads
to map onto the SVE specific IR intrinsics llvm.aarch64.sve.ld1/st1.
Reviewers: SjoerdMeijer, efriedma, kmclaughlin, rengolin
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78673
This patch changes the FP conversion intrinsics to take a predicate
that matches the number of lanes for the vector with the widest element
type as opposed to using <vscale x 16 x i1>.
For example:
```<vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f16(<vscale x 4 x float>, <vscale x 4 x i1>, <vscale x 8 x half>)```
now uses <vscale x 4 x i1> instead of <vscale x 16 x i1>
And similar for:
```<vscale x 4 x float> @llvm.aarch64.sve.fcvt.f32f64(<vscale x 4 x float>, <vscale x 2 x i1>, <vscale x 2 x double>)```
where the predicate now matches the wider type, so <vscale x 2 x i1>.
Reviewers: efriedma, SjoerdMeijer, paulwalker-arm, rengolin
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78402
This adds the flag IsOverloadCvt which tells CGBulitin to use
the result type and the type of the last operand as the
overloaded types for the LLVM IR intrinsic.
This also adds the flag IsFPConvert, which is needed to avoid
converting the predicate of the operation from svbool_t to
a predicate with fewer lanes, as the LLVM IR intrinsics use
the <vscale x 16 x i1> as the predicate.
Reviewers: SjoerdMeijer, efriedma
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78239
This also adds the IsOverloadWhileRW flag which tells CGBuiltin to use
the result predicate type and the first pointer type as the
overloaded types for the LLVM IR intrinsic.
Reviewers: SjoerdMeijer, efriedma
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78238
This also adds the IsOverloadWhile flag which tells CGBuiltin to use
both the default type (predicate) and the type of the second operand
(scalar) as the overloaded types for the LLMV IR intrinsic.
Reviewers: SjoerdMeijer, efriedma, rovka
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77595
Add the IsOverloadNone flag to tell CGBuiltin that it does not have
an overloaded type. This is used for e.g. svpfalse which does
not take any arguments and always returns a svbool_t.
This patch also adds builtins for svcntb_pat, svcnth_pat, svcntw_pat
and svcntd_pat, as those don't require custom codegen.
Reviewers: SjoerdMeijer, efriedma, rovka
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77596
The ACLE has builtins that take a scalar value that is to be expanded
into a vector by the operation. While the ISA may have an instruction
that takes an immediate or a scalar to represent this, the LLVM IR
intrinsic may not, so Clang will have to splat the scalar value.
This patch also adds the _n forms for svabd, svadd, svdiv, svdivr,
svmax, svmin, svmul, svmulh, svub and svsubr.
Reviewers: SjoerdMeijer, efriedma, rovka
Reviewed By: SjoerdMeijer
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77594
This implements zeroing of false lanes for binary operations,
where instead of merging into the first operand vector (_m)
a `select` is placed on the first input vector. This approach
easily translates to the use of the `zeroing movprfx` instruction.
This patch also adds builtins for svabd, svadd, svdiv, svdivr,
svmax, svmin, svmul, svmulh, svub and svsubr.
Reviewers: SjoerdMeijer, efriedma, rovka
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77593
Builtins that have the merge type MergeAnyExp or MergeZeroExp,
merge into a 'undef' or 'zero' vector respectively, which enables the
_x and _z behaviour for unary operations.
This patch also adds builtins for svabs and svneg.
Reviewers: SjoerdMeijer, efriedma, rovka
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77591
This patch adds a number of intrinsics that take immediates with
varying ranges based on the element size one of the operands.
svext: immediate ranging 0 to (2048/sizeinbits(elt) - 1)
svasrd: immediate ranging 1..sizeinbits(elt)
svqshlu: immediate ranging 1..sizeinbits(elt)/2
ftmad: immediate ranging 0..(sizeinbits(elt) - 1)
Reviewers: efriedma, SjoerdMeijer, rovka, rengolin
Reviewed By: SjoerdMeijer
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76679
Summary:
Remove asserting vector getters from Type in preparation for the
VectorType refactor. The existence of these functions complicates the
refactor while adding little value.
Reviewers: dexonsmith, sdesmalen, efriedma
Reviewed By: efriedma
Subscribers: cfe-commits, hiraditya, llvm-commits
Tags: #llvm, #clang
Differential Revision: https://reviews.llvm.org/D77278
Summary:
This patch adds a mechanism to easily add range checks for a builtin's
immediate operands. This patch is tested with the qdech intrinsic, which takes
both an enum for the predicate pattern, as well as an immediate for the
multiplier.
Reviewers: efriedma, SjoerdMeijer, rovka
Reviewed By: efriedma, SjoerdMeijer
Subscribers: mgorny, tschuett, mgrang, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76678
This adds builtins for all contiguous loads/stores, including
non-temporal, first-faulting and non-faulting.
Reviewers: efriedma, SjoerdMeijer
Reviewed By: SjoerdMeijer
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76238
It can be used to avoid passing the begin and end of a range.
This makes the code shorter and it is consistent with another
wrappers we already have.
Differential revision: https://reviews.llvm.org/D78016
Summary:
Remove usages of asserting vector getters in Type in preparation for the
VectorType refactor. The existence of these functions complicates the
refactor while adding little value.
Reviewers: sdesmalen, efriedma, krememek
Reviewed By: sdesmalen, efriedma
Subscribers: dexonsmith, Charusso, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77257
When constrained floating point is enabled the AArch64-specific builtins don't use constrained intrinsics in some cases. Fix that.
Neon is part of this patch, so ARM is affected as well.
Differential Revision: https://reviews.llvm.org/D77074
This patch adds a test for the PowerPC fma compiler builtins, some variations
of which negate inputs and outputs. The code to generate IR for these
builtins was untested before this patch.
Originally, the code used the outdated method of subtracting floating point
values from -0.0 as floating point negation. This patch remedies that.
Patch by: Drew Wock <drew.wock@sas.com>
Differential Revision: https://reviews.llvm.org/D76949
The main purpose of introducing these builtins is to add a range
metadata [1, 1025) on the work group size loaded from dispatch
ptr, which cannot be done by source code.
Differential Revision: https://reviews.llvm.org/D76772
Summary:
Since the conditional operator cannot be used with vector conditions
in C, we need a builtin to be able to express this operation in C
source.
Reviewers: aheejin
Subscribers: dschuff, sbc100, jgravelle-google, sunfish, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76538
Summary:
The range checks performed for the vqrdmulh_lane and vqrdmulh_lane Neon
intrinsics were incorrectly using their return type as the base type for
the range check performed on their 'lane' argument.
This patch updates those intrisics to use the type of the proper reference
argument to perform the range checks.
Reviewers: jmolloy, t.p.northover, rsmith, olista01, dnsampaio
Reviewed By: dnsampaio
Subscribers: dnsampaio, kristof.beyls, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D74766
Summary:
Range checks were not properly performed in the lane arguments of Neon
intrinsics implemented based on splat operations. Calls to those
intrinsics where translated to `__builtin__shufflevector` calls directly
by the pre-processor through the arm_neon.h macros, missing the chance
for the proper range checks.
This patch enables the range check by introducing an auxiliary splat
instruction in arm_neon.td, delaying the translation to shufflevector
calls to CGBuiltin.cpp in clang after the checks were performed.
Reviewers: jmolloy, t.p.northover, rsmith, olista01, ostannard
Reviewed By: ostannard
Subscribers: ostannard, dnsampaio, danielkiss, kristof.beyls, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D74619
Summary:
Some of the `*_laneq` intrinsics defined in arm_neon.td were missing the
setting of the `isLaneQ` attribute. This patch sets the attribute on the
related definitions, as they will be required to properly perform range
checks on their lane arguments.
Reviewers: jmolloy, t.p.northover, rsmith, olista01, dnsampaio
Reviewed By: dnsampaio
Subscribers: dnsampaio, kristof.beyls, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D74616
Reworked the patch to avoid sharing a header (SVETypeFlags.h) between
include/clang/Basic and utils/TableGen/SveEmitter.cpp. Now the patch
generates the enum/flags which is included in TargetBuiltins.h.
Also renamed one of the SveEmitter options to be in line with MVE.
Summary:
This is a first patch in a series for the SveEmitter to generate the arm_sve.h
header file and builtins.
I've tried my best to strip down this patch as best as I could, but there
are still a few changes that are not necessarily exercised by the load intrinsics
in this patch, mostly around the SVEType class which has some common logic to
represent types from a type and prototype string. I thought it didn't make
much sense to remove that from this patch and split it up.
This is a first patch in a series for the SveEmitter to generate the arm_sve.h
header file and builtins.
I've tried my best to strip down this patch as best as I could, but there
are still a few changes that are not necessarily exercised by the load intrinsics
in this patch, mostly around the SVEType class which has some common logic to
represent types from a type and prototype string. I thought it didn't make
much sense to remove that from this patch and split it up.
Reviewers: efriedma, rovka, SjoerdMeijer, rsandifo-arm, rengolin
Reviewed By: SjoerdMeijer
Tags: #clang
Differential Revision: https://reviews.llvm.org/D75470
This patch adds 'q' to mean 'scalable vector' in the builtin
type string, and for SVE will return the matching builtin
type as defined in the C/C++ language extensions for SVE.
This patch also adds some scaffolding to generate the arm_sve.h
header file, and some builtin definitions (+CodeGen) to be able
to implement some simple masked load intrinsics that use the
ACLE types, such as:
svint8_t test_svld1_s8(svbool_t pg, const int8_t *base) {
return svld1_s8(pg, base);
}
Reviewers: efriedma, rjmccall, rovka, rsandifo-arm, rengolin
Reviewed By: efriedma
Tags: #clang
Differential Revision: https://reviews.llvm.org/D75298
Summary:
Support ConstantInt::get() and Constant::getAllOnesValue() for scalable
vector type, this requires ConstantVector::getSplat() to take in 'ElementCount',
instead of 'unsigned' number of element count.
This change is needed for D73753.
Reviewers: sdesmalen, efriedma, apazos, spatel, huntergr, willlovett
Reviewed By: efriedma
Subscribers: tschuett, hiraditya, rkruppe, psnobl, cfe-commits, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D74386
Summary:
This patch generalizes the existing code to support CDE intrinsics
which will share some properties with existing MVE intrinsics
(some of the intrinsics will be polymorphic and accept/return values
of MVE vector types).
Specifically the patch:
* Adds new tablegen backends -gen-arm-cde-builtin-def,
-gen-arm-cde-builtin-codegen, -gen-arm-cde-builtin-sema,
-gen-arm-cde-builtin-aliases, -gen-arm-cde-builtin-header based on
existing MVE backends.
* Renames the '__clang_arm_mve_alias' attribute into
'__clang_arm_builtin_alias' (it will be used with CDE intrinsics as
well as MVE intrinsics)
* Implements semantic checks for the coprocessor argument of the CDE
intrinsics as well as the existing coprocessor intrinsics.
* Adds one CDE intrinsic __arm_cx1 to test the above changes
Reviewers: simon_tatham, MarkMurrayARM, ostannard, dmgreen
Reviewed By: simon_tatham
Subscribers: sdesmalen, mgorny, kristof.beyls, danielkiss, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D75850
__builtin_os_log_format
This is needed to keep all the objects, including temporaries returned
by function calls, written to the buffer alive until os_log_pack_send is
called.
rdar://problem/60105410
Summary:
Although SIMD integer min/max operations can be expressed using the ?:
operator in C++, that operator is disallowed for vectors in C. As a
workaround, this change introduces new WebAssembly-specific builtin
functions that lower to the desired vector icmp/select sequences.
Reviewers: aheejin, dschuff, kripken
Subscribers: sbc100, jgravelle-google, sunfish, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D75770
This removes everything but int_x86_avx512_mask_vcvtph2ps_512 which provides the SAE variant, but even this can use the fpext generic if the rounding control is the default.
Differential Revision: https://reviews.llvm.org/D75162
Previously we emitted an fmadd and a fmadd+fneg and combined them with a shufflevector. But this doesn't follow the correct exception behavior for unselected elements so the backend can't merge them into the fmaddsub/fmsubadd instructions.
This patch restores the the fmaddsub intrinsics so we don't have two arithmetic operations. We lose out on optimization opportunity in the non-strict FP case, but I don't think this is a big loss. If someone gives us a test case we can look into adding instcombine/dagcombine improvements. I'd rather not have the frontend do completely different things for strict and non-strict.
This still has problems because target specific intrinsics don't support strict semantics yet. We also still have all of the problems with masking. But we at least generate the right instruction in constrained mode now.
Differential Revision: https://reviews.llvm.org/D74268
This commit removes the artificial types <512 x i1> and <1024 x i1>
from HVX intrinsics, and makes v512i1 and v1024i1 no longer legal on
Hexagon.
It may cause existing bitcode files to become invalid.
* Converting between vector predicates and vector registers must be
done explicitly via vandvrt/vandqrt instructions (their intrinsics),
i.e. (for 64-byte mode):
%Q = call <64 x i1> @llvm.hexagon.V6.vandvrt(<16 x i32> %V, i32 -1)
%V = call <16 x i32> @llvm.hexagon.V6.vandqrt(<64 x i1> %Q, i32 -1)
The conversion intrinsics are:
declare <64 x i1> @llvm.hexagon.V6.vandvrt(<16 x i32>, i32)
declare <128 x i1> @llvm.hexagon.V6.vandvrt.128B(<32 x i32>, i32)
declare <16 x i32> @llvm.hexagon.V6.vandqrt(<64 x i1>, i32)
declare <32 x i32> @llvm.hexagon.V6.vandqrt.128B(<128 x i1>, i32)
They are all pure.
* Vector predicate values cannot be loaded/stored directly. This directly
reflects the architecture restriction. Loading and storing or vector
predicates must be done indirectly via vector registers and explicit
conversions via vandvrt/vandqrt instructions.
Summary:
These are in some sense the inverse of vmovl[bt]q: they take a vector
of n wide elements and truncate each to half its width. So they only
write half a vector's worth of output data, and therefore they also
take an 'inactive' parameter to provide the other half of the data in
the output vector. So vmovnb overwrites the even lanes of 'inactive'
with the narrowed values from the main input, and vmovnt overwrites
the odd lanes.
LLVM had existing codegen which generates these MVE instructions in
response to IR that takes two vectors of wide elements, or two vectors
of narrow ones. But in this case, we have one vector of each. So my
clang codegen strategy is to narrow the input vector of wide elements
by simply reinterpreting it as the output type, and then we have two
narrow vectors and can represent the operation as a vector shuffle
that interleaves lanes from both of them.
Even so, not all the cases I needed ended up being selected as a
single MVE instruction, so I've added a couple more patterns that spot
combinations of the 'MVEvmovn' and 'ARMvrev32' SDNodes which can be
generated as a VMOVN instruction with operands swapped.
This commit adds the unpredicated forms only.
Reviewers: dmgreen, miyuki, MarkMurrayARM, ostannard
Reviewed By: dmgreen
Subscribers: kristof.beyls, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D74337
Summary:
These intrinsics take a vector of 2n elements, and return a vector of
n wider elements obtained by sign- or zero-extending every other
element of the input vector. They're represented in IR as a
shufflevector that extracts the odd or even elements of the input,
followed by a sext or zext.
Existing LLVM codegen already matches this pattern and generates the
VMOVLB instruction (which widens the even-index input lanes). But no
existing isel rule was generating VMOVLT, so I've added some. However,
the new rules currently only work in little-endian MVE, because the
pattern they expect from isel lowering includes a bitconvert which
doesn't have the right semantics in big-endian.
The output of one existing codegen test is improved by those new
rules.
This commit adds the unpredicated forms only.
Reviewers: dmgreen, miyuki, MarkMurrayARM, ostannard
Reviewed By: dmgreen
Subscribers: kristof.beyls, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D74336
Summary:
These intrinsics just reorder the lanes of a vector, so the natural IR
representation is as a shufflevector operation. Existing LLVM codegen
already recognizes those particular shufflevectors and generates the
MVE VREV instruction.
This commit adds the unpredicated forms only.
Reviewers: dmgreen, miyuki, MarkMurrayARM, ostannard
Reviewed By: dmgreen
Subscribers: kristof.beyls, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D74334
Summary:
This commit adds the unpredicated intrinsics for the unary operations
vabsq (absolute value), vnegq (arithmetic negation), vmvnq (bitwise
complement), vqabsq and vqnegq (saturating versions of abs and neg for
signed integers, in the sense that they give INT_MAX if an input lane
is INT_MIN).
This is done entirely in clang: all of these operations have existing
isel patterns and existing tests for them on the LLVM side, so I've
just made clang emit the same IR that those patterns already match.
Reviewers: dmgreen, miyuki, MarkMurrayARM, ostannard
Reviewed By: MarkMurrayARM
Subscribers: kristof.beyls, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D74331
With REQUIRES: x86-register-target added to the tests.
Also remove some unneeded FIXMEs
But add a FIXME for bad IR generation for FMADDSUB/FMSUBADD with
constrained FP.
Original patch by Kevin P. Neal
This reverts commit 208470dd5d.
Tests fail:
error: unable to create target: 'No available targets are compatible with triple "x86_64-apple-darwin"'
This happens on clang-hexagon-elf, clang-cmake-armv7-quick, and
clang-cmake-armv7-quick bots.
If anyone has any suggestions on why then I'm all ears.
Differential Revision: https://reviews.llvm.org/D73570
Revert "[FPEnv][X86] Speculative fix for failures introduced by eda495426."
This reverts commit 80e17e5fcc.
The speculative fix didn't solve the test failures on Hexagon, ARMv6, and
MSVC AArch64.
When constrained floating point is enabled the X86-specific builtins don't
use constrained intrinsics in some cases. Fix that.
Differential Revision: https://reviews.llvm.org/D73570
Summary:
In big-endian MVE, the simple vector load/store instructions (i.e.
both contiguous and non-widening) don't all store the bytes of a
register to memory in the same order: it matters whether you did a
VSTRB.8, VSTRH.16 or VSTRW.32. Put another way, the in-register
formats of different vector types relate to each other in a different
way from the in-memory formats.
So, if you want to 'bitcast' or 'reinterpret' one vector type as
another, you have to carefully specify which you mean: did you want to
reinterpret the //register// format of one type as that of the other,
or the //memory// format?
The ACLE `vreinterpretq` intrinsics are specified to reinterpret the
register format. But I had implemented them as LLVM IR bitcast, which
is specified for all types as a reinterpretation of the memory format.
So a `vreinterpretq` intrinsic, applied to values already in registers,
would code-generate incorrectly if compiled big-endian: instead of
emitting no code, it would emit a `vrev`.
To fix this, I've introduced a new IR intrinsic to perform a
register-format reinterpretation: `@llvm.arm.mve.vreinterpretq`. It's
implemented by a trivial isel pattern that expects the input in an
MQPR register, and just returns it unchanged.
In the clang codegen, I only emit this new intrinsic where it's
actually needed: I prefer a bitcast wherever it will have the right
effect, because LLVM understands bitcasts better. So we still generate
bitcasts in little-endian mode, and even in big-endian when you're
casting between two vector types with the same lane size.
For testing, I've moved all the codegen tests of vreinterpretq out
into their own file, so that they can have a different set of RUN
lines to check both big- and little-endian.
Reviewers: dmgreen, MarkMurrayARM, miyuki, ostannard
Reviewed By: dmgreen
Subscribers: kristof.beyls, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D73786
The constrained fcmp intrinsics don't allow the TRUE/FALSE predicates.
Using them will assert. To workaround this I'm emitting the old X86 specific intrinsics that were never removed from the backend when we switched to using fcmp in IR. We have no way to mark them as being strict, but that's true of all target specific intrinsics so doesn't seem like we need to solve that here.
I've also added support for selecting between signaling and quiet.
Still need to support SAE which will require using a target specific
intrinsic. Also need to fix masking to not use an AND instruction
after the compare.
Differential Revision: https://reviews.llvm.org/D72906
Summary:
Currently, sqdmulh_lane and friends from the ACLE (implemented in arm_neon.h),
are represented in LLVM IR as a (by vector) sqdmulh and a vector of (repeated)
indices, like so:
%shuffle = shufflevector <4 x i16> %v, <4 x i16> undef, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
%vqdmulh2.i = tail call <4 x i16> @llvm.aarch64.neon.sqdmulh.v4i16(<4 x i16> %a, <4 x i16> %shuffle)
When %v's values are known, the shufflevector is optimized away and we are no
longer able to select the lane variant of sqdmulh in the backend.
This defeats a (hand-coded) optimization that packs several constants into a
single vector and uses the lane intrinsics to reduce register pressure and
trade-off materialising several constants for a single vector load from the
constant pool, like so:
int16x8_t v = {2,3,4,5,6,7,8,9};
a = vqdmulh_laneq_s16(a, v, 0);
b = vqdmulh_laneq_s16(b, v, 1);
c = vqdmulh_laneq_s16(c, v, 2);
d = vqdmulh_laneq_s16(d, v, 3);
[...]
In one microbenchmark from libjpeg-turbo this accounts for a 2.5% to 4%
performance difference.
We could teach the compiler to recover the lane variants, but this would likely
require its own pass. (Alternatively, "volatile" could be used on the constants
vector, but this is a bit ugly.)
This patch instead implements the following LLVM IR intrinsics for AArch64 to
maintain the original structure through IR optmization and into instruction
selection:
- sqdmulh_lane
- sqdmulh_laneq
- sqrdmulh_lane
- sqrdmulh_laneq.
These 'lane' variants need an additional register class. The second argument
must be in the lower half of the 64-bit NEON register file, but only when
operating on i16 elements.
Note that the existing patterns for shufflevector and sqdmulh into sqdmulh_lane
(etc.) remain, so code that does not rely on NEON intrinsics to generate these
instructions is not affected.
This patch also changes clang to emit these IR intrinsics for the corresponding
NEON intrinsics (AArch64 only).
Reviewers: SjoerdMeijer, dmgreen, t.p.northover, rovka, rengolin, efriedma
Reviewed By: efriedma
Subscribers: kristof.beyls, hiraditya, jdoerfert, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D71469
Summary:
This is a follow up on https://reviews.llvm.org/D71473#inline-647262.
There's a caveat here that `Align(1)` relies on the compiler understanding of `Log2_64` implementation to produce good code. One could use `Align()` as a replacement but I believe it is less clear that the alignment is one in that case.
Reviewers: xbolva00, courbet, bollu
Subscribers: arsenm, dylanmckay, sdardis, nemanjai, jvesely, nhaehnle, hiraditya, kbarton, jrtc27, atanasyan, jsji, Jim, kerbowa, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D73099
When constrained floating point is enabled the SystemZ-specific builtins
don't use constrained intrinsics in some cases. Fix that.
Differential Revision: https://reviews.llvm.org/D72722
Summary:
This change implements the expansion in two parts:
- Add a utility function emitAMDGPUPrintfCall() in LLVM.
- Invoke the above function from Clang CodeGen, when processing a HIP
program for the AMDGPU target.
The printf expansion has undefined behaviour if the format string is
not a compile-time constant. As a sufficient condition, the HIP
ToolChain now emits -Werror=format-nonliteral.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D71365
This change introduces three new builtins (which work on both pointers
and integers) that can be used instead of common bitwise arithmetic:
__builtin_align_up(x, alignment), __builtin_align_down(x, alignment) and
__builtin_is_aligned(x, alignment).
I originally added these builtins to the CHERI fork of LLVM a few years ago
to handle the slightly different C semantics that we use for CHERI [1].
Until recently these builtins (or sequences of other builtins) were
required to generate correct code. I have since made changes to the default
C semantics so that they are no longer strictly necessary (but using them
does generate slightly more efficient code). However, based on our experience
using them in various projects over the past few years, I believe that adding
these builtins to clang would be useful.
These builtins have the following benefit over bit-manipulation and casts
via uintptr_t:
- The named builtins clearly convey the semantics of the operation. While
checking alignment using __builtin_is_aligned(x, 16) versus
((x & 15) == 0) is probably not a huge win in readably, I personally find
__builtin_align_up(x, N) a lot easier to read than (x+(N-1))&~(N-1).
- They preserve the type of the argument (including const qualifiers). When
using casts via uintptr_t, it is easy to cast to the wrong type or strip
qualifiers such as const.
- If the alignment argument is a constant value, clang can check that it is
a power-of-two and within the range of the type. Since the semantics of
these builtins is well defined compared to arbitrary bit-manipulation,
it is possible to add a UBSAN checker that the run-time value is a valid
power-of-two. I intend to add this as a follow-up to this change.
- The builtins avoids int-to-pointer casts both in C and LLVM IR.
In the future (i.e. once most optimizations handle it), we could use the new
llvm.ptrmask intrinsic to avoid the ptrtoint instruction that would normally
be generated.
- They can be used to round up/down to the next aligned value for both
integers and pointers without requiring two separate macros.
- In many projects the alignment operations are already wrapped in macros (e.g.
roundup2 and rounddown2 in FreeBSD), so by replacing the macro implementation
with a builtin call, we get improved diagnostics for many call-sites while
only having to change a few lines.
- Finally, the builtins also emit assume_aligned metadata when used on pointers.
This can improve code generation compared to the uintptr_t casts.
[1] In our CHERI compiler we have compilation mode where all pointers are
implemented as capabilities (essentially unforgeable 128-bit fat pointers).
In our original model, casts from uintptr_t (which is a 128-bit capability)
to an integer value returned the "offset" of the capability (i.e. the
difference between the virtual address and the base of the allocation).
This causes problems for cases such as checking the alignment: for example, the
expression `if ((uintptr_t)ptr & 63) == 0` is generally used to check if the
pointer is aligned to a multiple of 64 bytes. The problem with offsets is that
any pointer to the beginning of an allocation will have an offset of zero, so
this check always succeeds in that case (even if the address is not correctly
aligned). The same issues also exist when aligning up or down. Using the
alignment builtins ensures that the address is used instead of the offset. While
I have since changed the default C semantics to return the address instead of
the offset when casting, this offset compilation mode can still be used by
passing a command-line flag.
Reviewers: rsmith, aaron.ballman, theraven, fhahn, lebedev.ri, nlopes, aqjune
Reviewed By: aaron.ballman, lebedev.ri
Differential Revision: https://reviews.llvm.org/D71499
Summary:
A few of the ARM MVE builtins directly return a structure type. This
causes an assertion failure at code-gen time if you try to assign the
result of the builtin to a variable, because the `RValue` created in
`EmitBuiltinExpr` from the `llvm::Value` produced by codegen is always
made by `RValue::get()`, which creates a non-aggregate `RValue` that
will fail an assertion when `AggExprEmitter::withReturnValueSlot` calls
`Src.getAggregatePointer()`. A similar failure occurs if you try to use
the struct return value directly to extract one field, e.g.
`vld2q(address).val[0]`.
The existing code-gen tests for those MVE builtins pass the returned
structure type directly to the C `return` statement, which apparently
managed to avoid that particular code path, so we didn't notice the
crash.
Now `EmitBuiltinExpr` checks the evaluation kind of the builtin's return
value, and does the necessary handling for aggregate returns. I've added
two extra test cases, both of which crashed before this change.
Reviewers: dmgreen, rjmccall
Reviewed By: rjmccall
Subscribers: kristof.beyls, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D72271
Christopher Tetreault