Summary:
I initially encountered those assertions when trying to create
this IR `alignment` attribute from clang's `__attribute__((assume_aligned(imm)))`,
because until D72994 there is no sanity checking for the value of `imm`.
But even then, we have `llvm::Value::MaximumAlignment` constant (which is `536870912`),
which is enforced for clang attributes, and then there are some other magical constant
(`0x40000000` i.e. `1073741824` i.e. `2 * 536870912`) in
`Attribute::getWithAlignment()`/`AttrBuilder::addAlignmentAttr()`.
I strongly suspect that `0x40000000` is incorrect,
and that also should be `llvm::Value::MaximumAlignment`.
Reviewers: erichkeane, hfinkel, jdoerfert, gchatelet, courbet
Reviewed By: erichkeane
Subscribers: hiraditya, cfe-commits, llvm-commits
Tags: #llvm, #clang
Differential Revision: https://reviews.llvm.org/D72998
This patch implements P1141R2 "Yet another approach for constrained declarations".
General strategy for this patch was:
- Expand AutoType to include optional type-constraint, reflecting the wording and easing the integration of constraints.
- Replace autos in parameter type specifiers with invented parameters in GetTypeSpecTypeForDeclarator, using the same logic
previously used for generic lambdas, now unified with abbreviated templates, by:
- Tracking the template parameter lists in the Declarator object
- Tracking the template parameter depth before parsing function declarators (at which point we can match template
parameters against scope specifiers to know if we have an explicit template parameter list to append invented parameters
to or not).
- When encountering an AutoType in a parameter context we check a stack of InventedTemplateParameterInfo structures that
contain the info required to create and accumulate invented template parameters (fields that were already present in
LambdaScopeInfo, which now inherits from this class and is looked up when an auto is encountered in a lambda context).
Resubmit after fixing MSAN failures caused by incomplete initialization of AutoTypeLocs in TypeSpecLocFiller.
Differential Revision: https://reviews.llvm.org/D65042
Summary:
Immediate vmvnq is code-generated as a simple vector constant in IR,
and left to the backend to recognize that it can be created with an
MVE VMVN instruction. The predicated version is represented as a
select between the input and the same constant, and I've added a
Tablegen isel rule to turn that into a predicated VMVN. (That should
be better than the previous VMVN + VPSEL: it's the same number of
instructions but now it can fold into an adjacent VPT block.)
The unpredicated forms of VBIC and VORR are done by enabling the same
isel lowering as for NEON, recognizing appropriate immediates and
rewriting them as ARMISD::VBICIMM / ARMISD::VORRIMM SDNodes, which I
then instruction-select into the right MVE instructions (now that I've
also reworked those instructions to use the same MC operand encoding).
In order to do that, I had to promote the Tablegen SDNode instance
`NEONvorrImm` to a general `ARMvorrImm` available in MVE as well, and
similarly for `NEONvbicImm`.
The predicated forms of VBIC and VORR are represented as a vector
select between the original input vector and the output of the
unpredicated operation. The main convenience of this is that it still
lets me use the existing isel lowering for VBICIMM/VORRIMM, and not
have to write another copy of the operand encoding translation code.
This intrinsic family is the first to use the `imm_simd` system I put
into the MveEmitter tablegen backend. So, naturally, it showed up a
bug or two (emitting bogus range checks and the like). Fixed those,
and added a full set of tests for the permissible immediates in the
existing Sema test.
Also adjusted the isel pattern for `vmovlb.u8`, which stopped matching
because lowering started turning its input into a VBICIMM. Now it
recognizes the VBICIMM instead.
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/D72934
After rGb4a99a061f517e60985667e39519f60186cbb469, passing a response file such as -Wp,@a.rsp wasn't working anymore because .rsp expansion happens inside clang's main() function.
This patch adds response file expansion in the -cc1 tool.
Differential Revision: https://reviews.llvm.org/D73120
We previously had to guard against older MSVC and GCC versions which had rvalue
references but not support for marking functions with ref qualifiers. However,
having bumped our minimum required version to MSVC 2017 and GCC 5.1 mean we can
unconditionally enable this feature. Rather than keeping the macro around, this
replaces use of the macro with the actual ref qualifier.
This patch implements P1141R2 "Yet another approach for constrained declarations".
General strategy for this patch was:
- Expand AutoType to include optional type-constraint, reflecting the wording and easing the integration of constraints.
- Replace autos in parameter type specifiers with invented parameters in GetTypeSpecTypeForDeclarator, using the same logic
previously used for generic lambdas, now unified with abbreviated templates, by:
- Tracking the template parameter lists in the Declarator object
- Tracking the template parameter depth before parsing function declarators (at which point we can match template
parameters against scope specifiers to know if we have an explicit template parameter list to append invented parameters
to or not).
- When encountering an AutoType in a parameter context we check a stack of InventedTemplateParameterInfo structures that
contain the info required to create and accumulate invented template parameters (fields that were already present in
LambdaScopeInfo, which now inherits from this class and is looked up when an auto is encountered in a lambda context).
Resubmit after incorrect check in NonTypeTemplateParmDecl broke lldb.
Differential Revision: https://reviews.llvm.org/D65042
Add a simple cache for constraint satisfaction results. Whether or not this simple caching
would be permitted in final C++2a is currently being discussed but it is required for
acceptable performance so we use it in the meantime, with the possibility of adding some
cache invalidation mechanisms later.
Differential Revision: https://reviews.llvm.org/D72552
This patch implements P1141R2 "Yet another approach for constrained declarations".
General strategy for this patch was:
- Expand AutoType to include optional type-constraint, reflecting the wording and easing the integration of constraints.
- Replace autos in parameter type specifiers with invented parameters in GetTypeSpecTypeForDeclarator, using the same logic
previously used for generic lambdas, now unified with abbreviated templates, by:
- Tracking the template parameter lists in the Declarator object
- Tracking the template parameter depth before parsing function declarators (at which point we can match template
parameters against scope specifiers to know if we have an explicit template parameter list to append invented parameters
to or not).
- When encountering an AutoType in a parameter context we check a stack of InventedTemplateParameterInfo structures that
contain the info required to create and accumulate invented template parameters (fields that were already present in
LambdaScopeInfo, which now inherits from this class and is looked up when an auto is encountered in a lambda context).
Differential Revision: https://reviews.llvm.org/D65042
This patch broke the Sanitizer buildbots. Please see the commit's
differential revision for more information
(https://reviews.llvm.org/D67678).
This reverts commit b72a8c65e4e34779b6bc9e466203f553f5294486.
Target regions have implicit outer region which may erroneously capture
some globals when it should not. It may lead to a compiler crash at the
compile time.
Summary:
The documentation for IndentCaseLabels claimed that the "Switch
statement body is always indented one level more than case labels". This
is technically false for the code block immediately following the label.
Its closing bracket aligns with the start of the label.
If the case label are not indented, it leads to a style where the
closing bracket of the block aligns with the closing bracket of the
switch statement, which can be hard to parse.
This change introduces a new option, IndentCaseBlocks, which when true
treats the block as a scope block (which it technically is).
(Note: regenerated ClangFormatStyleOptions.rst using tools/dump_style.py)
Reviewed By: MyDeveloperDay
Patch By: capn
Tags: #clang-format, #clang
Differential Revision: https://reviews.llvm.org/D72276
Implement support for C++2a requires-expressions.
Re-commit after compilation failure on some platforms due to alignment issues with PointerIntPair.
Differential Revision: https://reviews.llvm.org/D50360
Currently there are 4 different mechanisms for controlling denormal
flushing behavior, and about as many equivalent frontend controls.
- AMDGPU uses the fp32-denormals and fp64-f16-denormals subtarget features
- NVPTX uses the nvptx-f32ftz attribute
- ARM directly uses the denormal-fp-math attribute
- Other targets indirectly use denormal-fp-math in one DAGCombine
- cl-denorms-are-zero has a corresponding denorms-are-zero attribute
AMDGPU wants a distinct control for f32 flushing from f16/f64, and as
far as I can tell the same is true for NVPTX (based on the attribute
name).
Work on consolidating these into the denormal-fp-math attribute, and a
new type specific denormal-fp-math-f32 variant. Only ARM seems to
support the two different flush modes, so this is overkill for the
other use cases. Ideally we would error on the unsupported
positive-zero mode on other targets from somewhere.
Move the logic for selecting the flush mode into the compiler driver,
instead of handling it in cc1. denormal-fp-math/denormal-fp-math-f32
are now both cc1 flags, but denormal-fp-math-f32 is not yet exposed as
a user flag.
-cl-denorms-are-zero, -fcuda-flush-denormals-to-zero and
-fno-cuda-flush-denormals-to-zero will be mapped to
-fp-denormal-math-f32=ieee or preserve-sign rather than the old
attributes.
Stop emitting the denorms-are-zero attribute for the OpenCL flag. It
has no in-tree users. The meaning would also be target dependent, such
as the AMDGPU choice to treat this as only meaning allow flushing of
f32 and not f16 or f64. The naming is also potentially confusing,
since DAZ in other contexts refers to instructions implicitly treating
input denormals as zero, not necessarily flushing output denormals to
zero.
This also does not attempt to change the behavior for the current
attribute. The LangRef now states that the default is ieee behavior,
but this is inaccurate for the current implementation. The clang
handling is slightly hacky to avoid touching the existing
denormal-fp-math uses. Fixing this will be left for a future patch.
AMDGPU is still using the subtarget feature to control the denormal
mode, but the new attribute are now emitted. A future change will
switch this and remove the subtarget features.
A TemplateIdAnnotation represents only a template-id, not a
nested-name-specifier plus a template-id. Don't make a redundant copy of
the CXXScopeSpec and store it on the template-id annotation.
This slightly improves error recovery by more properly handling the case
where we would form an invalid CXXScopeSpec while parsing a typename
specifier, instead of accidentally putting the token stream into a
broken "annot_template_id with a scope specifier, but with no preceding
annot_cxxscope token" state.
Extend -fxray-instrumentation-bundle to split function-entry and
function-exit into two separate options, so that it is possible to
instrument only function entry or only function exit. For use cases
that only care about one or the other this will save significant overhead
and code size.
Differential Revision: https://reviews.llvm.org/D72890
XRay allows tuning by minimum function size, but also always instruments
functions with loops in them. If the minimum function size is set to a
large value the loop instrumention ends up causing most functions to be
instrumented anyway. This adds a new flag, -fxray-ignore-loops, to disable
the loop detection logic.
Differential Revision: https://reviews.llvm.org/D72873
Summary: Currently, an attempt to rewrite source code inside a macro expansion succeeds, but results in empty text, rather than failing with an error. This patch restructures to the code to explicitly validate ranges before attempting to edit them.
Reviewers: gribozavr
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D72274
Right now every dataflow algorithm uses its own worklist implementation.
This is a first step to reduce this duplication. Some upcoming
algorithms such as the lifetime analysis is going to use the factored
out implementations.
Differential Revision: https://reviews.llvm.org/D72380
This reverts commit 3d210ed3d1880c615776b07d1916edb400c245a6.
See https://reviews.llvm.org/D71082 for the patch and discussion that make it
possible to reapply this patch.
Partially reverts 0a2be46cfdb698fefcc860a56b47dde0884d5335 as it turned
out to cause redundant module rebuilds in multi-process incremental builds.
When a module was getting out of date, all compilation processes started at the
same time were marking it as `ToBuild`. So each process was building the same
module instead of checking if it was built by someone else and using that
result. In addition to the work duplication, contention on the same .pcm file
wasn't making builds faster.
Note that for a single-process build this change would cause redundant module
reads and validations. But reading a module is faster than building it and
multi-process builds are more common than single-process. So I'm willing to
make such a trade-off.
rdar://problem/54395127
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D72860
This is (more?) usable by GDB pretty printers and seems nicer to write.
There's one tricky caveat that in C++14 (LLVM's codebase today) the
static constexpr member declaration is not a definition - so odr use of
this constant requires an out of line definition, which won't be
provided (that'd make all these trait classes more annoyidng/expensive
to maintain). But the use of this constant in the library implementation
is/should always be in a non-odr context - only two unit tests needed to
be touched to cope with this/avoid odr using these constants.
Based on/expanded from D72590 by Christian Sigg.
This needs somewhat careful disambiguation, as C++2a explicit(bool) is a
breaking change. We only enable it in cases where the source construct
could not possibly be anything else.
This reverts commit 921f871ac438175ca8fcfcafdfcfac4d7ddf3905 because it
causes libc++ code to trigger __warn_memset_zero_len.
See https://reviews.llvm.org/D71082.
Flags are clang's default UI is flags.
We can have an env var in addition to that, but in D69825 nobody has yet
mentioned why this needs an env var, so omit it for now. If someone
needs to set the flag via env var, the existing CCC_OVERRIDE_OPTIONS
mechanism works for it (set CCC_OVERRIDE_OPTIONS=+-fno-integrated-cc1
for example).
Also mention the cc1-in-process change in the release notes.
Also spruce up the test a bit so it actually tests something :)
Differential Revision: https://reviews.llvm.org/D72769
This is applied to the vector types defined in <arm_mve.h> for use
with the intrinsics for the ARM MVE vector architecture.
Its purpose is to inhibit lax vector conversions, but only in the
context of overload resolution of the MVE polymorphic intrinsic
functions. This solves an ambiguity problem with polymorphic MVE
intrinsics that take a vector and a scalar argument: the scalar
argument can often have the wrong integer type due to default integer
promotions or unsuffixed literals, and therefore, the type of the
vector argument should be considered trustworthy when resolving MVE
polymorphism.
As part of the same change, I've added the new attribute to the
declarations generated by the MveEmitter Tablegen backend (and
corrected a namespace issue with the other attribute while I was
there).
Reviewers: aaron.ballman, dmgreen
Reviewed By: aaron.ballman
Subscribers: kristof.beyls, JDevlieghere, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D72518
These driver options perform some checking and delegate to MC options -x86-align-branch* and -x86-branches-within-32B-boundaries.
Reviewed By: skan
Differential Revision: https://reviews.llvm.org/D72463
Add support for type-constraints in template type parameters.
Also add support for template type parameters as pack expansions (where the type constraint can now contain an unexpanded parameter pack).
Differential Revision: https://reviews.llvm.org/D44352
Summary:
Before this change, X86_32ABIInfo::classifyArgument would be called
twice on vector arguments to vectorcall functions. This function has
side effects to track GPR register usage, and this would lead to
incorrect GPR usage in some cases. The specific case I noticed is from
running out of XMM registers with mixed FP and vector arguments and no
aggregates of any kind. Consider this prototype:
void __vectorcall vectorcall_indirect_vec(
double xmm0, double xmm1, double xmm2, double xmm3, double xmm4,
__m128 xmm5,
__m128 ecx,
int edx,
__m128 mem);
classifyArgument has no effects when called on a plain FP type, but when
called on a vector type, it modifies FreeRegs to model GPR consumption.
However, this should not happen during the vector call first pass.
I refactored the code to unify vectorcall HVA logic with regcall HVA
logic. The conventions pass HVAs in registers differently (expanded vs.
not expanded), but if they do not fit in registers, they both pass them
indirectly by address.
Reviewers: erichkeane, craig.topper
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D72110
