Shiny new DBG_PHI instruction usually have physical registers as operands
-- however, the machine verifier checks to see whether they're live, and
occasionally this fails. There's a filter for DBG_VALUE instructions to not
get verified in this way: expand it to exempt all debug instructions from
liveness checking, which means DBG_PHIs get treated like DBG_VALUEs.
This also future proofs against us adding new debug instructions.
Differential Revision: https://reviews.llvm.org/D117891
This patch updates how splat loads handled and is an extension of D106555.
Particularly, for v2i64/v4f32/v4i32 types, they are updated to handle only
non-extending loads. For v8i16/v16i8 types, they are updated to handle extending
loads only if the memory VT is the same vector element VT type.
A test case has been added to illustrate a scenario where a PPCISD::LD_SPLAT
node should not be produced. In this test, it depicts the following f64
extending load used in a v2f64 build vector, but the extending load is actually
used in more places other than the build vector (such as in t12 and t16).
```
Type-legalized selection DAG: %bb.0 'test:entry'
SelectionDAG has 20 nodes:
t0: ch = EntryToken
t4: i64,ch = CopyFromReg t0, Register:i64 %1
t6: i64,ch = CopyFromReg t0, Register:i64 %2
t11: f64,ch = load<(load (s64) from %ir.b, !tbaa !7)> t0, t4, undef:i64
t16: f64 = fadd t31, t37
t34: ch = store<(store (s64) into %ir.c, !tbaa !7)> t31:1, t16, t6, undef:i64
t36: ch = TokenFactor t34, t37:1
t27: v2f64 = BUILD_VECTOR t37, t37
t22: ch,glue = CopyToReg t36, Register:v2f64 $v2, t27
t12: f64 = fadd t11, t37
t28: ch = store<(store (s64) into %ir.b, !tbaa !7)> t11:1, t12, t4, undef:i64
t31: f64,ch = load<(load (s64) from %ir.c, !tbaa !7)> t28, t6, undef:i64
t2: i64,ch = CopyFromReg t0, Register:i64 %0
t37: f64,ch = load<(load (s32) from %ir.a, !tbaa !3), anyext from f32> t0, t2, undef:i64
t23: ch = PPCISD::RET_FLAG t22, Register:v2f64 $v2, t22:1
```
Differential Revision: https://reviews.llvm.org/D117803
When a comparison is extended and it would be free to extend the
arguments to that comparison, we can propagate the extend into those arguments.
This prevents extra instructions being generated to extend the result of the
comparison, which is not free to extend.
This is a resubmission of D116812 with fixes that need another review.
Differential Revision: https://reviews.llvm.org/D118139
Fixes a false positive that occurs when a user-implemented memmove is
instrumented by HWASan.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D118180
Using a _BitInt (or _ExtInt) type as a block parameter or block return
type hits an "unreachable" when trying to determine the encoding for
the block. Instead of crashing, this patch handles it like some of the
other types for which we don't yet have an encoding. The test case
verifies we no longer crash, but does not verify that we provide any
particular encoding (it can be updated once someone more familiar with
ObjC steps in to define the encoding).
Fixes PR50503.
Adds patterns of the form "(and a, (not b)) -> bic".
NOTE: With this support I'm inclined to remove AArch64ISD::BIC,
but will leave that investigation for another time.
Differential Revision: https://reviews.llvm.org/D118365
This updates NewGVN test cases that were running
"opt -basic-aa -newgvn ..."
to run
"opt -passes=newgvn ..."
instead.
The pipeline will be more similar to what we used to have with
legacy PM by doing it this way. The compatility mode that we've
been using for awhile during transition from legacy PM to new PM,
i.e. using the legacy syntax together with new PM, has resulted in
a pipeline such as
-passes='function(require<basic-aa>),function(newgvn)'
but running the analysis in a separate function pass manager seem
overly complicated for these tests.
Another difference is that we will get the default aa-pipeline instead
of only running basic-aa. That might be a bit questioned (given that
the tests originally specified basic-aa). The output is however
identival for all the test cases modified here regardless of using
basic-aa or the default aa-pipeline.
This is also another small step towards removal of the support for
using the legacy PM syntax in opt.
Differential Revision: https://reviews.llvm.org/D118340
Use the new PM syntax when specifying the pipeline in regression
tests previously running
"opt -newgvn ..."
Instead we now do
"opt -passes=newgvn ..."
Notice that this also changes the aa-pipeline to become the default
aa-pipeline instead of just basic-aa. Since these tests haven't been
explicitly requesting basic-aa in the past (compared to the test cases
updated in a separate patch involving "-basic-aa -newgvn") it is
assumed that the exact aa-pipeline isn't important for the validity
of the test cases. An alternative could have been to add
-aa-pipeline=basic-aa as well to the run lines, but that might just
add clutter in case the test cases do not care about the aa-pipeline.
This is another step to move away from the legacy PM syntax when
specifying passes in opt.
Differential Revision: https://reviews.llvm.org/D118341
These splats -- whether BUILD_VECTOR or SPLAT_VECTOR -- are formed by
first extracting a value from a vector and splatting it to all elements
of the destination vector. These could be performed more optimally,
avoiding the drop to scalar, using RVV's vrgather, for example.
The LMULMAX check names didn't match the options we were passing to llc
(they were swapped around) and we were silently missing coverage for one
test which differs between RV32 and RV64.
This patch introduces a new options for script llvm-mca-compare.py
(-plot-resource-pressure, -plot) to draw plots for llvm-mca tool
statistics and option (--plot-path) to specify relative path where
you want to save the plots.
Differential Revision: https://reviews.llvm.org/D115718
Previous folds by combineSetCCMOVMSK might have converted these to CMP when changing the bitwidth, and the CMP->SUB fold might not have happened (or will happen)
SortJavaScriptImports attempts to set its currently parsed token to an
invalid token when it reaches the end of the line. However in doing so,
it used a `FormatToken`, which contains a `Token Tok`. `Token` does not
have a constructor, so its fields start out as uninitialized memory.
`Token::startToken()` initializes all fields. Calling it in
`JavaScriptImportSorter`'s constructor thus fixes the problem.
Differential Revision: https://reviews.llvm.org/D118448
Currently, basic AA has special support for llvm.memcpy.* intrinsics. This change extends this support for any memory trancsfer opration and in particular llvm.memmove.* intrinsic.
Reviewed By: reames, nikic
Differential Revision: https://reviews.llvm.org/D117095
It causes builds to fail with
llvm/include/llvm/Support/Casting.h:269:
typename llvm::cast_retty<X, Y*>::ret_type llvm::cast(Y*)
[with X = llvm::IntegerType; Y = const llvm::Type; typename llvm::cast_retty<X, Y*>::ret_type = const llvm::IntegerType*]:
Assertion `isa<X>(Val) && "cast<Ty>() argument of incompatible type!"' failed.
See the code review for link to a reproducer.
> This patch introduces folding of and-reduce idiom and generates code
> that is easier to read and which is lest costly in terms of icmp operations.
> The folding is
> ```
> icmp eq (bitcast(icmp ne (lhs, rhs)), 0)
> ```
> into
> ```
> icmp eq(bitcast(lhs), bitcast(rhs))
> ```
>
> See PR53419.
>
> Differential Revision: https://reviews.llvm.org/D118317
> Reviewed By: lebedev.ri, spatel
This reverts commit 8599bb0f26.
This also revertes the dependent change:
"[Test] Add 'ne' tests for and-reduce pattern folding"
This reverts commit a4aaa59953.
On the level of the generated object files, both symbols (both
original and alias) are generally indistinguishable - both are
regular defined symbols. But previously, only the original
function had the COFF ComplexType set to IMAGE_SYM_DTYPE_FUNCTION,
while the symbol created via an alias had the type set to
IMAGE_SYM_DTYPE_NULL.
This matches what GCC does, which emits directives for setting the
COFF symbol type for this kind of alias symbol too.
This makes a difference when GNU ld.bfd exports symbols without
dllexport directives or a def file - it seems to decide between
function or data exports based on the COFF symbol type. This means
that functions created via aliases, like some C++ constructors,
are exported as data symbols (missing the thunk for calling without
dllimport).
The hasnt been an issue when doing the same with LLD, as LLD decides
between function or data export based on the flags of the section
that the symbol points at.
This should fix the root cause of
https://github.com/msys2/MINGW-packages/issues/10547.
Differential Revision: https://reviews.llvm.org/D118328
rG9103b73fe052 was assuming that we could OR/AND with the source vector, but that will fail on float/double vectors without bitcasting - it also missed the case that any_of checks might be testing less than all the source elements
import X = A.B.C;
Previously, these were unhandled and would terminate import sorting.
With this change, aliases sort as their own group, coming last after all
other imports.
Aliases are not sorted within their group, as they may reference each
other, so order is significant.
This reverts commit f750c3d95a. It fixes
the msan issue by not parsing past the end of the line when handling
import aliases.
Differential Revision: https://reviews.llvm.org/D118446
Based on the output of include-what-you-use. No big deal here, it's a utils
library and it doesn't seem to be used a lot across the codebase.
$ clang++ -E -Iinclude -I../llvm/include ../llvm/utils/TableGen/GlobalISel/*.cpp -std=c++14 -fno-rtti -fno-exceptions | wc -l
before: 573143
after: 568908
Related Discourse thread: https://llvm.discourse.group/t/include-what-you-use-include-cleanup
Differential Revision: https://reviews.llvm.org/D118375
Based on the output of include-what-you-use. No other library seems affected by
the new forward declaration.
$ clang++ -E -Iinclude -I../llvm/include ../llvm/lib/TableGen/*.cpp -std=c++14 -fno-rtti -fno-exceptions | wc -l
before: 795231
after: 750654
Related Discourse thread: https://llvm.discourse.group/t/include-what-you-use-include-cleanup
Differential Revision: https://reviews.llvm.org/D118374
This moves the dependency of several files on include/llvm/ADT/STLExtras.h to
the much shorter llvm/ADT/STLArrayExtras.h
Differential Revision: https://reviews.llvm.org/D118342
When creating an alloca to copy a matrix due to memory conflicts, those
allocas used to use VectorTypes, which forced them to have huge
alignments for large vectors.
This patch updates LowerMatrixIntrinsics to use a corresponding array
type, like Clang already does, to get more manageable alignments.
Reviewed By: anemet, thegameg
Differential Revision: https://reviews.llvm.org/D118239
During the upstreaming process from fir-dev some
new builder have been introduced in the `flang/Optimizer/Builder`
directory. This patch removes the obsolete DoLoopHelper still present
in the lowering directories and makes use of the new one where needed.
This patch is part of the upstreaming effort from fir-dev branch.
Reviewed By: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D118442
This patch removes some files made obsolete by newer version
of them available in the Optimizer directory.
`flang/include/flang/Lower/FIRBuilder.h` and `flang/lib/Lower/FIRBuilder.cpp` are
removed and replace by the newer version present in
`flang/include/flang/Optimizer/Builder/FIRBuilder.h` and
`flang/lib/Optimizer/Builder/FIRBuilder.cpp`.
`flang/include/flang/Lower/Support/BoxValue.h` and `flang/lib/Lower/ConvertExpr.cpp` are removed and replace by the newer
version present in `flang/include/flang/Optimizer/Builder/BoxValue.h`
This patch is a preparation to be able to upstream the lowering from
fir-dev.
Reviewed By: jeanPerier, kiranchandramohan
Differential Revision: https://reviews.llvm.org/D118404
The named address space overloads of builtins that take a pointer
argument are conditionalized on the `__opencl_c_generic_address_space`
feature macro (in a `#else` body). Introduce an internal feature
macro instead, such that their availability can be controlled in a
single place and independently of the generic address space feature
macro.
This commit does not change the available builtins.
Differential Revision: https://reviews.llvm.org/D118158
This change is to fix a link time error when building llvm with msvc.
MSVC's implementation does not support weak hook or lsan so this change
disables lsan's weak hook definition.
Only GCC supports LSan.
Tested with visual studio 2019 v16.9.6
Reviewed By: aaron.ballman, rnk
Differential Revision: https://reviews.llvm.org/D118162
Branch protection in M-class is supported by
- Armv8.1-M.Main
- Armv8-M.Main
- Armv7-M
Attempting to enable this for other architectures, either by
command-line (e.g -mbranch-protection=bti) or by target attribute
in source code (e.g. __attribute__((target("branch-protection=..."))) )
will generate a warning.
In both cases function attributes related to branch protection will not
be emitted. Regardless of the warning, module level attributes related to
branch protection will be emitted when it is enabled via the command-line.
The following people also contributed to this patch:
- Victor Campos
Reviewed By: chill
Differential Revision: https://reviews.llvm.org/D115501
`instrprof-icall-promo.test` `FAIL`s on Solaris/sparcv9:
Profile-sparc :: instrprof-icall-promo.test
Profile-sparcv9 :: instrprof-icall-promo.test
when compiling `compiler-rt/test/profile/Inputs/instrprof-icall-promo_2.cpp` with
fatal error: error in backend: Relocation for CG Profile could not be created: unknown relocation name
This happens because the Sparc backend doesn't implement `BFD_RELOC_NONE`.
This patch fixes that, following what X86 does.
Tested on `sparcv9-sun-solaris2.11`.
Differential Revision: https://reviews.llvm.org/D118136
This explicitly records whether a scalar IV is needed in the
VPWidenIntOrFpInductionRecipe, to remove a dependence on the cost-model
during its ::execute.
It will also be used in D116123 to determine if a vector phi will be
generated.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D118167
These were omitted in all Windows configurations, but it turns out
that they work just fine in MinGW mode.
This allows converting a couple cases of "XFAIL: LIBCXX-WINDOWS-FIXME"
into "XFAIL: msvc" as the bug is specific to MSVC mode (clang-cl).
Differential Revision: https://reviews.llvm.org/D118192
Nikita Popov