Richard Smith [Wed, 19 Feb 2014 22:53:42 +0000 (22:53 +0000)]
Remove a broken attempt to cope with someone #undef'ing __has_include_next.
This was broken because __has_include_next(...) would not be valid in a
preprocessor condition if __has_include_next is not defined.
Chandler Carruth [Wed, 19 Feb 2014 22:35:01 +0000 (22:35 +0000)]
Teach Clang to provide ::max_align_t in C11 and C++11 modes.
This definition is not chosen idly. There is an unfortunate reality with
max_align_t -- the specific nature of its definition leaks into the ABI
almost immediately. Because it is part of C11 and C++11 it becomes
essential for it to match with other systems on that ABI. There is an
effort to discourage any further use of this construct as a consequence
-- using max_align_t introduces an immediate ABI problem. We can never
update it to have larger alignment even as the microarchitecture changes
to necessitate higher alignment. =/
The particular definition here exactly matches the ABI of GCC's chosen
::max_align_t definition, for better or worse. This was written with the
help of Richard Smith who was decoding the exact ABI implications of the
selected definition in GCC. Notably, in-register arguments are impacted
by the particular definition chosen. =/
No one is under the illusion that this is a "good" or "useful"
definition of max_align_t, and we are working with the standards
committee to specify a more useful interface to address this need.
Reid Kleckner [Wed, 19 Feb 2014 22:06:10 +0000 (22:06 +0000)]
MS ABI: Let non-virtual method overloads participate in vftable ordering
In the Microsoft ABI, the vftable is laid out as if all methods in every
overload set were declared in reverse order of declaration at the point
of declaration of the first overload in the set.
Previously we only considered virtual methods in an overload set, but
MSVC includes non-virtual methods for ordering purposes.
Aaron Ballman [Wed, 19 Feb 2014 20:43:58 +0000 (20:43 +0000)]
Some of these headings had the incorrect number of "underlines" and so would get warnings when generating the content from Sphinx. No functional changes intended.
Jordan Rose [Wed, 19 Feb 2014 17:44:16 +0000 (17:44 +0000)]
[analyzer] Extend IdenticalExprChecker to check logical and bitwise expressions.
IdenticalExprChecker now warns if any expressions in a logical or bitwise
chain (&&, ||, &, |, or ^) are the same. Unlike the previous patch, this
actually checks all subexpressions against each other (an O(N^2) operation,
but N is likely to be small).
Jordan Rose [Wed, 19 Feb 2014 17:44:11 +0000 (17:44 +0000)]
[analyzer] Extend IdenticalExprChecker to check the two branches of an if.
This extends the checks for identical expressions to handle identical
statements, and compares the consequent and alternative ("then" and "else")
branches of an if-statement to see if they are identical, treating a single
statement surrounded by braces as equivalent to one without braces.
This does /not/ check subsequent branches in an if/else chain, let alone
branches that are not consecutive. This may improve in a future patch, but
it would certainly take more work.
Aaron Ballman [Wed, 19 Feb 2014 14:48:31 +0000 (14:48 +0000)]
Moving the documentation for the enable_if attribute into AttrDocs. Removed the "clang introduces" phrase for style consistency, but otherwise the documentation is identical.
Tim Northover [Wed, 19 Feb 2014 11:55:06 +0000 (11:55 +0000)]
AArch64: look up EmitAArch64Scalar support before calling.
This fixes one immediate bug where an expression with side-effects
could be emitted twice during a NEON call.
It also prepares the way for folding CodeGen for many of the SISD
intrinsics into a table, reducing code size and hopefully increasing
performance eventually ("binary search + few switch cases" should be
better than "lots of switch cases").
Tim Northover [Wed, 19 Feb 2014 10:37:13 +0000 (10:37 +0000)]
ARM NEON: add vcvtX (with rounding mode) intrinsics to v8 ARM.
These instructions (well, the f32 ones) are supported on 32-bit ARMv8, not just
AArch64. Now that the arm_neon.td refactoring is complete, adding them is
surprisingly simple.
Tim Northover [Wed, 19 Feb 2014 10:37:09 +0000 (10:37 +0000)]
ARM NEON: use more flexible TableGen field for defs.
We used to have special handling for isCrypto and isA64 bits in the
NeonEmitter.cpp file (it knew the former was predicated on __ARM_FEATURE_CRYPTO
and the latter on __aarch64__ and went through various contortions to make sure
the correct intrinsics were emitted under the correct guard.
This is ugly and has obvious scalability problems (e.g. vcvtX intrinsics are
needed, which are ARMv8 only but available on both, yet another category). This
patch moves the #if predicate into the arm_neon.td file directly and makes
NeonEmitter.cpp agnostic about what goes in there.
It also deduplicates arm_neon.td so that each desired intrinsic is mentioned in
just one place (necessary because of the new mechanism for creating
arm_neon.h).
Tim Northover [Wed, 19 Feb 2014 10:37:05 +0000 (10:37 +0000)]
ARM & AArch64: merge the semantic checking of NEON intrinsics
There are two kinds of automatically generated tests for NEON intrinsics, both
of which can be merged without adversely affecting users.
1. We check that a valid kind of __builtin_neon_XYZ overload is requested (e.g.
we're not asking for a float32x4_t version when it only accepts integers. Since
the __builtin_neon_XYZ intrinsics should only be used in arm_neon.h, relaxing
this test and permitting AArch64 types for AArch32 should not cause a problem.
The extra arm_neon.h definitions should be #ifdefed out anyway.
2. We check that intrinsics which take immediates are actually given
compile-time constants within range. Since all NEON intrinsics should be
backwards compatible, these tests should be identical on AArch64 and AArch32
anyway.
This patch, therefore, merges the separate AArch64 and 32-bit checks.
David Majnemer [Wed, 19 Feb 2014 03:00:56 +0000 (03:00 +0000)]
Implement DR577
DR18 previously forebode typedefs to be used as parameter types if they
were of type 'void'. DR577 allows 'void' to be used as a function
parameter type regardless from where it came.
TU is not guaranteed to be initialised in all cases. In particular if CIdx or
ast_filename is NULL (or if &TU is NULL), then clang_createTranslationUnit2 will
not initialise the out parameter out_TU. This is followed by an assertion check
which may perform a branch based on unitialised memory.
Reid Kleckner [Tue, 18 Feb 2014 22:51:52 +0000 (22:51 +0000)]
MS ABI: Remove nv adjustment from direct vdtor calls and prologues
Summary:
Generally the vector deleting dtor, which we model as a vtable thunk,
takes care of non-virtual adjustment and delegates to the other
destructor variants. The other non-complete destructor variants assume
that 'this' on entry points to the virtual base subobject that first
declared the virtual destructor.
We need to change the adjustment in both the prologue and the vdtor call
setup.
Ted Kremenek [Tue, 18 Feb 2014 22:12:10 +0000 (22:12 +0000)]
Experiment with making -Wunreachable-code more immediately useful by restricting warnings to those issued in the main file.
This warning has a whole bunch of known false positives, much of them due
to code that is "sometimes unreachable". This can caused by code that
is conditionally generated by the preprocessor, branches that are defined
in terms of architecture-specific details (e.g., the size of a type), and
so on. While these are all good things to address one by one, the reality
is that this warning has received little love lately. By restricting
its purvue, we can focus on the top issues effecting main files, which
should be smaller, and then gradually widen the scope.
Sylvestre Ledru [Tue, 18 Feb 2014 17:45:06 +0000 (17:45 +0000)]
In some cases (for example, the Firefox build system), the CLANG_CXX variable is defined but empty.
Extend the test (like it is done in scan-build) to check also if the variable
is empty or not.
Aaron Ballman [Tue, 18 Feb 2014 17:36:50 +0000 (17:36 +0000)]
DeLesley Hutchins (who wrote the original thread-safety attribute functionality) and I have agreed to start migrating from lock-specific terminology to "capability"-specific terminology. This opens the door for future threading-related analysis passes so that a common nomenclature can be used.
The following attributes have been (silently) deprecated, with their replacements listed:
Richard Smith [Tue, 18 Feb 2014 03:51:47 +0000 (03:51 +0000)]
PR18876: The special-case rule that ignores the destruction of a top-level
temporary in a decltype expression only applies if that temporary was created
by a function call, not by a function-style cast or other flavour of
expression.
Bob Wilson [Mon, 17 Feb 2014 19:21:09 +0000 (19:21 +0000)]
Change PGO instrumentation to compute counts in a separate AST traversal.
Previously, we made one traversal of the AST prior to codegen to assign
counters to the ASTs and then propagated the count values during codegen. This
patch now adds a separate AST traversal prior to codegen for the
-fprofile-instr-use option to propagate the count values. The counts are then
saved in a map from which they can be retrieved during codegen.
This new approach has several advantages:
1. It gets rid of a lot of extra PGO-related code that had previously been
added to codegen.
2. It fixes a serious bug. My original implementation (which was mailed to the
list but never committed) used 3 counters for every loop. Justin improved it to
move 2 of those counters into the less-frequently executed breaks and continues,
but that turned out to produce wrong count values in some cases. The solution
requires visiting a loop body before the condition so that the count for the
condition properly includes the break and continue counts. Changing codegen to
visit a loop body first would be a fairly invasive change, but with a separate
AST traversal, it is easy to control the order of traversal. I've added a
testcase (provided by Justin) to make sure this works correctly.
3. It improves the instrumentation overhead, reducing the number of counters for
a loop from 3 to 1. We no longer need dedicated counters for breaks and
continues, since we can just use the propagated count values when visiting
breaks and continues.
To make this work, I needed to make a change to the way we count case
statements, going back to my original approach of not including the fall-through
in the counter values. This was necessary because there isn't always an AST node
that can be used to record the fall-through count. Now case statements are
handled the same as default statements, with the fall-through paths branching
over the counter increments. While I was at it, I also went back to using this
approach for do-loops -- omitting the fall-through count into the loop body
simplifies some of the calculations and make them behave the same as other
loops. Whenever we start using this instrumentation for coverage, we'll need
to add the fall-through counts into the counter values.
Anton Yartsev [Mon, 17 Feb 2014 18:25:34 +0000 (18:25 +0000)]
[analyzer] Improved checker naming in CFG dump.
This implements FIXME from Checker.cpp (FIXME: We want to return the package + name of the checker here.) and replaces hardcoded checker names with the new ones obtained via getCheckName().getName().
Aaron Ballman [Mon, 17 Feb 2014 18:23:02 +0000 (18:23 +0000)]
Forcing it to be an error when there is no Documentation list specified for an attribute. This is a bit of a (harmless) hack, but the FIXME explains why and when this hack can be removed. It's a justified hack because this prevents attribute authors from forgetting to add documentation when they add a new attribute.
Aaron Ballman [Mon, 17 Feb 2014 15:27:10 +0000 (15:27 +0000)]
Implements a declarative approach to documenting individual attributes in Clang via a Documentation tablegen class. Also updates the internals manual with information about how to use this new, required, documentation feature.
This patch adds some very, very sparse initial documentation for some attributes. Additional effort from attribute authors is greatly appreciated.
Daniel Jasper [Mon, 17 Feb 2014 07:57:46 +0000 (07:57 +0000)]
clang-format: Don't wrap "const" etc. of function declarations.
Generally people seem to prefer wrapping the first function parameter
over wrapping the trailing tokens "const", "override" and "final". This
does not extend to function-like annotations and probably not to other
non-standard annotations.
Manman Ren [Sat, 15 Feb 2014 01:29:02 +0000 (01:29 +0000)]
PGO: fix a bug in parsing pgo data.
When a function has a single counter, we will offset the pointer by 1 when
parsing the next function. If a function has multiple counters, we are
okay after skipping rest of the counters.