Roman Lebedev [Wed, 3 Jan 2018 08:45:19 +0000 (08:45 +0000)]
[Sema] -Wtautological-constant-compare is too good. Cripple it.
Summary:
The diagnostic was mostly introduced in D38101 by me, as a reaction to wasting a lot of time, see [[ https://lists.llvm.org/pipermail/cfe-commits/Week-of-Mon-20171009/206427.html | mail ]].
However, the diagnostic is pretty dumb. While it works with no false-positives,
there are some questionable cases that are diagnosed when one would argue that they should not be.
The common complaint is that it diagnoses the comparisons between an `int` and
`long` when compiling for a 32-bit target as tautological, but not when
compiling for 64-bit targets. The underlying problem is obvious: data model.
In most cases, 64-bit target is `LP64` (`int` is 32-bit, `long` and pointer are
64-bit), and the 32-bit target is `ILP32` (`int`, `long`, and pointer are 32-bit).
I.e. the common pattern is: (pseudocode)
```
#include <limits>
#include <cstdint>
int main() {
using T1 = long;
using T2 = int;
T1 r;
if (r < std::numeric_limits<T2>::min()) {}
if (r > std::numeric_limits<T2>::max()) {}
}
```
As an example, D39149 was trying to fix this diagnostic in libc++, and it was not well-received.
This *could* be "fixed", by changing the diagnostics logic to something like
`if the types of the values being compared are different, but are of the same size, then do diagnose`,
and i even attempted to do so in D39462, but as @rjmccall rightfully commented,
that implementation is incomplete to say the least.
So to stop causing trouble, and avoid contaminating upcoming release, lets do this workaround:
* move these three diags (`warn_unsigned_always_true_comparison`, `warn_unsigned_enum_always_true_comparison`, `warn_tautological_constant_compare`) into it's own `-Wtautological-constant-in-range-compare`
* Disable them by default
* Make them part of `-Wextra`
* Additionally, give `warn_tautological_constant_compare` it's own flag `-Wtautological-type-limit-compare`.
I'm not happy about that name, but i can't come up with anything better.
This way all three of them can be enabled/disabled either altogether, or one-by-one.
Reviewers: aaron.ballman, rsmith, smeenai, rjmccall, rnk, mclow.lists, dim
Reviewed By: aaron.ballman, rsmith, dim
Subscribers: thakis, compnerd, mehdi_amini, dim, hans, cfe-commits, rjmccall
Richard Smith [Tue, 2 Jan 2018 23:52:42 +0000 (23:52 +0000)]
Fix and simplify handling of return type for (generic) lambda conversion function to function pointer.
Previously, we would:
* compute the type of the conversion function and static invoker as a
side-effect of template argument deduction for a conversion
* re-compute the type as part of deduced return type deduction when building
the conversion function itself
Neither of these turns out to be quite correct. There are other ways to reach a
declaration of the conversion function than in a conversion (such as an
explicit call or friend declaration), and performing auto deduction causes the
function type to be rebuilt in the context of the lambda closure type (which is
different from the context in which it originally appeared, resulting in
spurious substitution failures for constructs that are valid in one context but
not the other, such as the use of an enclosing class's "this" pointer).
This patch switches us to use a different strategy: as before, we use the
declared type of the operator() to form the type of the conversion function and
invoker, but we now populate that type as part of return type deduction for the
conversion function. And the invoker is now treated as simply being an
implementation detail of building the conversion function, and isn't given
special treatment by template argument deduction for the conversion function
any more.
[analyzer] do not crash with assertion on processing locations of bodyfarmed functions
This addresses an issue introduced in r183451: since
`removePiecesWithInvalidLocations` is called *after* `adjustCallLocations`,
it is not necessary, and in fact harmful, to have this assertion in
adjustCallLocations.
Douglas Yung [Tue, 2 Jan 2018 20:42:53 +0000 (20:42 +0000)]
[DOXYGEN] Fix doxygen and content issues in pmmintrin.h
- Fix incorrect wording in various intrinsic descriptions. Previously the descriptions used "low-order" and "high-order" when the intended meaning was "even-indexed" and "odd-indexed".
Douglas Yung [Tue, 2 Jan 2018 20:39:29 +0000 (20:39 +0000)]
[DOXYGEN] Fix doxygen and content issues in emmintrin.h
- Fixed innaccurate instruction mappings for various intrinsics.
- Fixed description of NaN handling in comparison intrinsics.
- Unify description of _mm_store_pd1 to match _mm_store1_pd.
- Fix incorrect wording in various intrinsic descriptions. Previously the descriptions used "low-order" and "high-order" when the intended meaning was "even-indexed" and "odd-indexed".
- Fix typos.
- Add missing italics command (\a) for params and fixed some parameter spellings.
Nick Lewycky [Tue, 2 Jan 2018 19:10:12 +0000 (19:10 +0000)]
Suppress undefined-template warnings when the pattern is declared in a system header.
The way to fix an undefined-template warning is to add lines to the header file that defines the template pattern. We should suppress the warnings when the template pattern is in a system header because we don't expect users to edit those.
Faisal Vali [Mon, 1 Jan 2018 18:23:28 +0000 (18:23 +0000)]
Again reverting an attempt to convert the DeclSpec enums into scoped enums.
- reverts r321622, r321625, and r321626.
- the use of bit-fields is still resulting in warnings - even though we can use static-asserts to harden the code and ensure the bit-fields are wide enough. The bots still complain of warnings being seen.
- to silence the warnings requires specifying the bit-fields with the underlying enum type (as opposed to the enum type itself), which then requires lots of unnecessary static casts of each enumerator within DeclSpec to the underlying-type, which even though could be seen as implementation details, it does hamper readability - and given the additional litterings, makes me question the value of the change.
Benjamin Kramer [Mon, 1 Jan 2018 17:07:23 +0000 (17:07 +0000)]
[Sema] Fix build with GCC
tools/clang/lib/Sema/DeclSpec.cpp: In member function 'void clang::DeclSpec::Finish(clang::Sema&, const clang::PrintingPolicy&)':
tools/clang/lib/Sema/DeclSpec.cpp:1116:8: error: could not convert 'clang::DeclSpec::TSW_unspecified' from 'const TSW {aka const clang::TypeSpecifierWidth}' to 'int'
tools/clang/lib/Sema/DeclSpec.cpp:1117:8: error: could not convert 'clang::DeclSpec::TSW_short' from 'const TSW {aka const clang::TypeSpecifierWidth}' to 'int'
tools/clang/lib/Sema/DeclSpec.cpp:1118:8: error: could not convert 'clang::DeclSpec::TSW_longlong' from 'const TSW {aka const clang::TypeSpecifierWidth}' to 'int'
tools/clang/lib/Sema/DeclSpec.cpp:1128:8: error: could not convert 'clang::DeclSpec::TSW_long' from 'const TSW {aka const clang::TypeSpecifierWidth}' to 'int'
Faisal Vali [Mon, 1 Jan 2018 16:36:47 +0000 (16:36 +0000)]
Use 'unsigned int' instead of enum bit-fields to silence some warnings from r321622
- bots were complaining that the bit-field width was less than the width of the underlying type (note, underlying types of enums can not be bit-fields)
- add static_asserts for TSS and TSW to ensure that the bit-fields can hold all the enumerators - and add comments next to the last enumerator warning not to reorder.
See https://reviews.llvm.org/rC321622 for the patch that introduced the warnings.
Faisal Vali [Mon, 1 Jan 2018 15:42:13 +0000 (15:42 +0000)]
[NFC] Modernize enums TypeSpecifierWidth, TypeSpecifierSign & TypeSpecifierType into scoped enums with underlying types.
- Since these enums are used as bit-fields - for the bit-fields to be interpreted as unsigned, the underlying type must be specified as unsigned.
Previous failed attempt - wherein I did not specify an underlying type - was the sum of:
https://reviews.llvm.org/rC321614
https://reviews.llvm.org/rC321615
Serge Pavlov [Mon, 1 Jan 2018 13:27:01 +0000 (13:27 +0000)]
Enable configuration files in clang
Clang is inherently a cross compiler and can generate code for any target
enabled during build. It however requires to specify many parameters in the
invocation, which could be hardcoded during configuration process in the
case of single-target compiler. The purpose of configuration files is to
make specifying clang arguments easier.
A configuration file is a collection of driver options, which are inserted
into command line before other options specified in the clang invocation.
It groups related options together and allows specifying them in simpler,
more flexible and less error prone way than just listing the options
somewhere in build scripts. Configuration file may be thought as a "macro"
that names an option set and is expanded when the driver is called.
Use of configuration files is described in `UserManual.rst`.
[Sema] Improve diagnostics for const- and ref-qualified member functions
(Re-submission of D39937 with fixed tests.)
Adjust wording for const-qualification mismatch to be a little more clear.
Also add another diagnostic for a ref qualifier mismatch, which previously produced a useless error (this error path is simply very old; see rL119336):
Before:
error: cannot initialize object parameter of type 'X0' with an expression of type 'X0'
After:
error: 'this' argument to member function 'rvalue' is an lvalue, but function has rvalue ref-qualifier
[Sema] Improve diagnostics for const- and ref-qualified member functions
Summary:
Adjust wording for const-qualification mismatch to be a little more clear.
Also add another diagnostic for a ref qualifier mismatch, which previously produced a useless error (this error path is simply very old; see rL119336):
Before:
error: cannot initialize object parameter of type 'X0' with an expression of type 'X0'
After:
error: 'this' argument to member function 'rvalue' is an lvalue, but function has rvalue ref-qualifier
Serge Pavlov [Sat, 30 Dec 2017 17:59:26 +0000 (17:59 +0000)]
Enable configuration files in clang
Clang is inherently a cross compiler and can generate code for any target
enabled during build. It however requires to specify many parameters in the
invocation, which could be hardcoded during configuration process in the
case of single-target compiler. The purpose of configuration files is to
make specifying clang arguments easier.
A configuration file is a collection of driver options, which are inserted
into command line before other options specified in the clang invocation.
It groups related options together and allows specifying them in simpler,
more flexible and less error prone way than just listing the options
somewhere in build scripts. Configuration file may be thought as a "macro"
that names an option set and is expanded when the driver is called.
Use of configuration files is described in `UserManual.rst`.
Alex Lorenz [Fri, 29 Dec 2017 17:42:40 +0000 (17:42 +0000)]
[driver][darwin] Take the OS version from -m<os>-version-min argument when
-target has no OS version
This ensures that Clang won't warn about redundant -m<os>-version-min
argument for an invocation like
`-target x86_64-apple-macos -mmacos-version-min=10.11`
Craig Topper [Fri, 29 Dec 2017 06:39:16 +0000 (06:39 +0000)]
Revert r321504 "[X86] Don't accidentally enable PKU on cannon lake and icelake or CLWB on cannonlake."
I based that commit on what was in Intel's public documentation here https://software.intel.com/sites/default/files/managed/c5/15/architecture-instruction-set-extensions-programming-reference.pdf
Which specifically said CLWB wasn't until Icelake.
But I've since cross checked with SDE and it thinks these features exist on CNL and ICL. So now I don't know what to believe.
I've added test coverage of the current behavior as part of the revert so at least now have proof of what we're doing.
Coby Tayree [Wed, 27 Dec 2017 11:25:07 +0000 (11:25 +0000)]
[x86][icelake][vbmi2]
added vbmi2 feature recognition
added intrinsics support for vbmi2 instructions
_mm[128,256,512]_mask[z]_compress_epi[16,32]
_mm[128,256,512]_mask_compressstoreu_epi[16,32]
_mm[128,256,512]_mask[z]_expand_epi[16,32]
_mm[128,256,512]_mask[z]_expandloadu_epi[16,32]
_mm[128,256,512]_mask[z]_sh[l,r]di_epi[16,32,64]
_mm[128,256,512]_mask_sh[l,r]dv_epi[16,32,64]
matching a similar work on the backend (D40206)
Differential Revision: https://reviews.llvm.org/D41557
Coby Tayree [Wed, 27 Dec 2017 09:22:34 +0000 (09:22 +0000)]
[hotfix]
fixinig test failures as seen here:
http://lab.llvm.org:8011/builders/llvm-clang-lld-x86_64-scei-ps4-ubuntu-fast/builds/22791/steps/test/logs/stdio
which resulted by rL321480
Coby Tayree [Wed, 27 Dec 2017 09:00:31 +0000 (09:00 +0000)]
[x86][icelake][vpclmulqdq]
added vpclmulqdq feature recognition
added intrinsics support for vpclmulqdq instructions
_mm256_clmulepi64_epi128
_mm512_clmulepi64_epi128
matching a similar work on the backend (D40101)
Differential Revision: https://reviews.llvm.org/D41573
Coby Tayree [Wed, 27 Dec 2017 08:16:54 +0000 (08:16 +0000)]
[x86][icelake][vaes]
added vaes feature recognition
added intrinsics support for vaes instructions, matching a similar work on the backend (D40078)
_mm256_aesenc_epi128
_mm512_aesenc_epi128
_mm256_aesenclast_epi128
_mm512_aesenclast_epi128
_mm256_aesdec_epi128
_mm512_aesdec_epi128
_mm256_aesdeclast_epi128
_mm512_aesdeclast_epi128
Faisal Vali [Mon, 25 Dec 2017 22:23:20 +0000 (22:23 +0000)]
Add a fixit for attributes incorrectly placed prior to 'struct/class/enum' keyword.
Suggest moving the following erroneous attrib list (based on location)
[[]] struct X;
to
struct [[]] X;
Additionally, added a fixme for the current implementation that diagnoses misplaced attributes to consider using the newly introduced diagnostic (that I think is more user-friendly).
Sam McCall [Sat, 23 Dec 2017 19:31:24 +0000 (19:31 +0000)]
[Index] Reduce size of SymbolInfo struct.
Summary:
This is currently 16 bytes, the patch reduces it to 4.
(Building with clang on linux x84, I guess others are similar)
The only subfield that might need a bigger type is SymbolPropertySet,
I've moved it to the end of the struct so if it grows, SymbolInfo will
only be 8 bytes.
With a full index of namespace-scope symbols from the LLVM project (200k)
loaded into clangd, this saves ~2MB of RAM.
Faisal Vali [Sat, 23 Dec 2017 18:56:34 +0000 (18:56 +0000)]
[NFC] Update the template-parameter parsers and analyzers to return NamedDecl (vs Decl)
This patch addresses a FIXME and has the template-parameter processing functions return a more derived common type NamedDecl (as opposed to a type needlessly higher up in the inheritance hierarchy : Decl).
Serge Pavlov [Fri, 22 Dec 2017 15:22:45 +0000 (15:22 +0000)]
Unit tests for TBAA metadata generation.
Now tests for metadata created by clang involve compiling code snippets
placed into c/c++ source files and matching interesting patterns in the
obtained textual representation of IR. Writting such tests is a painful
process as metadata often form complex tree-like structures but textual
representation of IR contains only a pile of metadata at the module end.
This change implements IR matchers that may be used to match required
patterns in the binary IR representation. In this case the metadata
structure is not broken and creation of match patterns is easier.
The change adds unit tests for TBAA metadata generation.
Ivan A. Kosarev [Fri, 22 Dec 2017 09:57:24 +0000 (09:57 +0000)]
[CodeGen] Represent array members in new-format TBAA type descriptors
Now that in the new TBAA format we allow access types to be of
any object types, including aggregate ones, it becomes critical
to specify types of all sub-objects such aggregates comprise as
their members. In order to meet this requirement, this patch
enables generation of field descriptors for members of array
types.
Ivan A. Kosarev [Fri, 22 Dec 2017 09:54:23 +0000 (09:54 +0000)]
[CodeGen] Support generation of TBAA info in the new format
Now that the MDBuilder helpers generating TBAA type and access
descriptors in the new format are in place, we can teach clang to
use them when requested.
[Modules] Map missing private submodules from Foo.Private to Foo_Private
In case `@import Foo.Private` fails because the submodule doesn't exist,
look for `Foo_Private` (if available) and build/load that module
instead. In that process emit a warning and tell the user about the
assumption.
The intention here is to assist all existing private modules owners
(in ObjC and Swift) to migrate to the new `Foo_Private` syntax.
Faisal Vali [Fri, 22 Dec 2017 03:50:55 +0000 (03:50 +0000)]
Diagnose the various invalid decl-specifiers on nontype template parameters.
The standard correctly forbids various decl-specifiers that dont make sense on non-type template parameters - such as the extern in:
template<extern int> struct X;
This patch implements those restrictions (in a fashion similar to the corresponding checks on function parameters within ActOnParamDeclarator).
Credit goes to miyuki (Mikhail Maltsev) for drawing attention to this issue, authoring the initial versions of this patch, and supporting the effort to re-engineer it slightly. Thank you!
For details of how this patch evolved please see: https://reviews.llvm.org/D40705
[Modules] Change private modules rules and warnings
We used to advertise private modules to be declared as submodules
(Foo.Private). This has proven to not scale well since private headers
might carry several dependencies, introducing unwanted content into the
main module and often causing dep cycles.
Change the canonical way to name it to Foo_Private, forcing private
modules as top level ones, and provide warnings under -Wprivate-module
to suggest fixes for other private naming. Update documentation to
reflect that.
Craig Topper [Thu, 21 Dec 2017 23:50:22 +0000 (23:50 +0000)]
[X86] Allow _mm_prefetch (both the header implementation and the builtin) to accept bit 2 which is supposed to indicate the prefetched addresses will be written to
Add the appropriate _MM_HINT_ET0/ET1 defines to match gcc.
Erich Keane [Thu, 21 Dec 2017 23:27:36 +0000 (23:27 +0000)]
Correct hasFeature/isValidFeatureName's handling of shstk/adx/mwaitx
https://bugs.llvm.org/show_bug.cgi?id=35721 reports that x86intrin.h
is issuing a few warnings. This is because attribute target is using
isValidFeatureName for its source. It was also discovered that two of
these were missing from hasFeature.
Additionally, shstk is and ibu are reordered alphabetically, as came
up during code review.
Richard Smith [Thu, 21 Dec 2017 22:26:47 +0000 (22:26 +0000)]
Suppress "redundant parens" warning for "A (::B())".
This is a slightly odd construct (it's more common to see "A (::B)()") but can
happen in friend declarations, and the parens are not redundant as they prevent
the :: binding to the left.
Aaron Ballman [Thu, 21 Dec 2017 21:42:42 +0000 (21:42 +0000)]
Re-commit r321223, which adds a printing policy to the ASTDumper.
This allows you to dump C++ code that spells bool instead of _Bool, leaves off the elaborated type specifiers when printing struct or class names, and other C-isms.
Fixes the -Wreorder issue and fixes the ast-dump-color.cpp test.
Volodymyr Sapsai [Thu, 21 Dec 2017 19:42:37 +0000 (19:42 +0000)]
[CodeGen] Fix crash when a function taking transparent union is redeclared.
When a function taking transparent union is declared as taking one of
union members earlier in the translation unit, clang would hit an
"Invalid cast" assertion during EmitFunctionProlog. This case
corresponds to function f1 in test/CodeGen/transparent-union-redecl.c.
We decided to cast i32 to union because after merging function
declarations function parameter type becomes int,
CGFunctionInfo::ArgInfo type matches with ABIArgInfo type, so we decide
it is a trivial case. But these types should also be castable to
parameter declaration type which is not the case here.
The fix is in checking for the trivial case if ABIArgInfo type matches with
parameter declaration type. It exposed inconsistency that we check
hasScalarEvaluationKind for different types in EmitParmDecl and
EmitFunctionProlog, and comment says they should match.
Additional tests in Sema/transparent-union.c capture current behavior and make
sure there are no regressions.
Artem Dergachev [Thu, 21 Dec 2017 18:43:02 +0000 (18:43 +0000)]
[analyzer] Fix zero-initialization of stack VLAs under ObjC ARC.
Using ARC, strong, weak, and autoreleasing stack variables are implicitly
initialized with nil. This includes variable-length arrays of Objective-C object
pointers. However, in the analyzer we don't zero-initialize them. We used to,
but it accidentally regressed after r289618.
Under ARC, the array variable's initializer within DeclStmt is an
ImplicitValueInitExpr. Environment doesn't maintain any bindings for this
expression kind - instead it always knows that it's a known constant
(0 in our case), so it just returns the known value by calling
SValBuilder::makeZeroVal() (see EnvironmentManager::getSVal().
Commit r289618 had introduced reasonable behavior of SValBuilder::makeZeroVal()
for the arrays, which produces a zero-length compoundVal{}. When such value
is bound to arrays, in RegionStoreManager::bindArray() "remaining" items in the
array are default-initialized with zero, as in
RegionStoreManager::setImplicitDefaultValue(). The similar mechanism works when
an array is initialized by an initializer list that is too short, eg.
int a[3] = { 1, 2 };
would result in a[2] initialized with 0. However, in case of variable-length
arrays it didn't know if any more items need to be added,
because, well, the length is variable.
Add the default binding anyway, regardless of how many actually need
to be added. We don't really care how many, because the default binding covers
the whole array anyway.
projects / clang / log