Richard Smith [Tue, 7 May 2013 02:55:48 +0000 (02:55 +0000)]
C++ status:
- fix paper links to point to isocpp.org, where most of the papers are already up
- update "SVN" features to "Clang 3.3" to distinguish them from features which we
complete after the branch
- document use of -std=c++1y to enable c++1y support
John McCall [Mon, 6 May 2013 21:39:12 +0000 (21:39 +0000)]
Grab-bag of bit-field fixes:
- References to ObjC bit-field ivars are bit-field lvalues;
fixes rdar://13794269, which got me started down this.
- Introduce Expr::refersToBitField, switch a couple users to
it where semantically important, and comment the difference
between this and the existing API.
- Discourage Expr::getBitField by making it a bit longer and
less general-sounding.
- Lock down on const_casts of bit-field gl-values until we
hear back from the committee as to whether they're allowed.
The one user has been changed to use getLValue on the compound literal
expression and then use the normal bindLoc to assign a value. No need
to special case this in the StoreManager.
Jordan Rose [Mon, 6 May 2013 16:48:20 +0000 (16:48 +0000)]
[analyzer] Handle CXXTemporaryObjectExprs in compound literals.
This occurs because in C++11 the compound literal syntax can trigger a
constructor call via list-initialization. That is, "Point{x, y}" and
"(Point){x, y}" end up being equivalent. If this occurs, the inner
CXXConstructExpr will have already handled the object construction; the
CompoundLiteralExpr just needs to propagate that value forwards.
Note the two CXXBindTemporaryExprs. The InitListExpr is really part of the
CompoundLiteralExpr, not an object in its own right. By introducing a new
entity initialization kind in Sema specifically for compound literals, we
avoid the treatment of the inner InitListExpr as a temporary.
Ulrich Weigand [Mon, 6 May 2013 16:26:41 +0000 (16:26 +0000)]
Add SystemZ support
This patch then adds all the usual platform-specific pieces for SystemZ:
driver support, basic target info, register names and constraints,
ABI info and vararg support. It also adds new tests to verify pre-defined
macros and inline asm, and updates a test for the minimum alignment change.
This version of the patch incorporates feedback from reviews by
Eric Christopher and John McCall. Thanks to all reviewers!
Ulrich Weigand [Mon, 6 May 2013 16:23:57 +0000 (16:23 +0000)]
Allow targets to define minimum alignment for global variables
This patch adds a new common code feature that allows platform code to
request minimum alignment of global symbols. The background for this is
that on SystemZ, the most efficient way to load addresses of global symbol
is the LOAD ADDRESS RELATIVE LONG (LARL) instruction. This instruction
provides PC-relative addressing, but only to *even* addresses. For this
reason, existing compilers will guarantee that global symbols are always
aligned to at least 2. [ Since symbols would otherwise already use a
default alignment based on their type, this will usually only affect global
objects of character type or character arrays. ] GCC also allows creating
symbols without that extra alignment by using explicit "aligned" attributes
(which then need to be used on both definition and each use of the symbol).
To enable support for this with Clang, this patch adds a
TargetInfo::MinGlobalAlign variable that provides a global minimum for the
alignment of every global object (unless overridden via explicit alignment
attribute), and adds code to respect this setting. Within this patch, no
platform actually sets the value to anything but the default 1, resulting
in no change in behaviour on any existing target.
This version of the patch incorporates feedback from reviews by
Eric Christopher and John McCall. Thanks to all reviewers!
Daniel Jasper [Mon, 6 May 2013 10:24:51 +0000 (10:24 +0000)]
Don't break comments after includes.
LLVM/Clang basically don't use such comments and for Google-style,
include-lines are explicitly exempt from the column limit. Also, for
most cases, where the column limit is violated, the "better" solution
would be to move the comment to before the include, which clang-format
cannot do (yet).
Daniel Jasper [Mon, 6 May 2013 08:27:33 +0000 (08:27 +0000)]
Change indentation when breaking after a type.
clang-format did not indent any declarations/definitions when breaking
after the type. With this change, it indents for all declarations but
does not indent for function definitions, i.e.:
While it might seem inconsistent to indent function declarations, but
not definitions, there are two reasons for that:
- Function declarations are very similar to declarations of function
type variables, so there is another side to consistency to consider.
- There can be many function declarations on subsequent lines and not
indenting can make them harder to identify. Function definitions
are already separated by their body and not indenting
makes the function name slighly easier to find.
Richard Smith [Sun, 5 May 2013 21:17:10 +0000 (21:17 +0000)]
Factor out duplication between lvalue-to-rvalue conversions and variable
assignments in constant expressions. No significant functionality changes
(slight improvement to potential constant expression checking).
Aaron Ballman [Sat, 4 May 2013 16:58:37 +0000 (16:58 +0000)]
Properly parsing __declspec(safebuffers), though there is no semantic hookup. For more information about safebuffers, see MSDN: http://msdn.microsoft.com/en-us/library/dd778695(v=vs.110).aspx
Richard Smith [Sat, 4 May 2013 07:12:37 +0000 (07:12 +0000)]
Tweaks to C++ status: add a link to Faisal's generic lambdas implementation,
and mark "clarifying memory allocation" as done, since it turns out that our
optimizations here (such as they are) already conform to the new rules.
Richard Smith [Sat, 4 May 2013 07:00:32 +0000 (07:00 +0000)]
Implement most of N3638 (return type deduction for normal functions).
Missing (somewhat ironically) is support for the new deduction rules
in lambda functions, plus PCH support for return type patching.
Richard Smith [Sat, 4 May 2013 06:44:46 +0000 (06:44 +0000)]
Don't build a call expression referring to a function which we're not allowed
to use. This makes very little difference right now (other than suppressing
follow-on errors in some cases), but will matter more once we support deduced
return types (we don't want expressions with undeduced return types in the
AST).
Sean Callanan [Sat, 4 May 2013 02:04:27 +0000 (02:04 +0000)]
Added a function to check whether a Decl is in
the list of Decls for a given DeclContext. This
is useful for LLDB's implementation of
FindExternalLexicalDecls.
Richard Smith [Sat, 4 May 2013 01:26:46 +0000 (01:26 +0000)]
Separate out and special-case the diagnostic for 'auto' in a
conversion-type-id, in preparation for this becoming valid in c++1y mode.
No functionality change; small diagnostic improvement.
Ted Kremenek [Sat, 4 May 2013 01:13:01 +0000 (01:13 +0000)]
[analyzer; alternate edges] prune edges whose end/begin locations have the same statement parents.
This change required some minor changes to LocationContextMap to have it map
from PathPieces to LocationContexts instead of PathDiagnosticCallPieces to
LocationContexts. These changes are in the other diagnostic
generation logic as well, but are functionally equivalent.
Interestingly, this optimize requires delaying "cleanUpLocation()" until
later; possibly after all edges have been optimized. This is because
we need PathDiagnosticLocations to refer to the semantic entity (e.g. a statement)
as long as possible. Raw source locations tell us nothing about
the semantic relationship between two locations in a path.
Douglas Gregor [Fri, 3 May 2013 22:58:43 +0000 (22:58 +0000)]
When building a module, forward diagnostics to the outer diagnostic consumer.
Previously, we would clone the current diagnostic consumer to produce
a new diagnostic consumer to use when building a module. The problem
here is that we end up losing diagnostics for important diagnostic
consumers, such as serialized diagnostics (where we'd end up with two
diagnostic consumers writing the same output file). With forwarding,
the diagnostics from all of the different modules being built get
forwarded to the one serialized-diagnostic consumer and are emitted in
a sane way.
[Preprocessor] For the MacroExpands preprocessor callback, also pass the MacroArgs object that provides information about
the argument tokens for a function macro.
Adrian Prantl [Fri, 3 May 2013 20:11:48 +0000 (20:11 +0000)]
Reapply r180982 with repaired logic and an additional testcase.
Un-break the gdb buildbot.
- Use the debug location of the return expression for the cleanup code
if the return expression is trivially evaluatable, regardless of the
number of stop points in the function.
- Ensure that any EH code in the cleanup still gets the line number of
the closing } of the lexical scope.
- Added a testcase with EH in the cleanup.
Ben Langmuir [Fri, 3 May 2013 19:00:33 +0000 (19:00 +0000)]
Move CapturedStmt parameters to CapturedDecl
Move the creation of CapturedStmt parameters out of CodeGen and into
Sema, making it easier to customize the outlined function. The
ImplicitParamDecls are stored in the CapturedDecl using an
ASTContext-allocated array.
Hans Wennborg [Fri, 3 May 2013 09:10:16 +0000 (09:10 +0000)]
Support __wchar_t in -fms-extensions and -fms-compatibility modes.
MSVC provides __wchar_t, either as an alias for the built-in wchar_t
type, or as a separate type depending on language (C vs C++) and flags
(-fno-wchar).
In -fms-extensions, Clang will simply accept __wchar_t as an alias for
whatever type is used for wide character literals. In -fms-compatibility, we
try to mimic MSVC's behavior by always making __wchar_t a builtin type.
John McCall [Fri, 3 May 2013 07:33:41 +0000 (07:33 +0000)]
Correctly emit certain implicit references to 'self' even within
a lambda.
Bug #1 is that CGF's CurFuncDecl was "stuck" at lambda invocation
functions. Fix that by generally improving getNonClosureContext
to look through lambdas and captured statements but only report
code contexts, which is generally what's wanted. Audit uses of
CurFuncDecl and getNonClosureAncestor for correctness.
Bug #2 is that lambdas weren't specially mapping 'self' when inside
an ObjC method. Fix that by removing the requirement for that
and using the normal EmitDeclRefLValue path in LoadObjCSelf.
Jordan Rose [Fri, 3 May 2013 05:47:31 +0000 (05:47 +0000)]
[analyzer] Check the stack frame when looking for a var's initialization.
FindLastStoreBRVisitor is responsible for finding where a particular region
gets its value; if the region is a VarRegion, it's possible that value was
assigned at initialization, i.e. at its DeclStmt. However, if a function is
called recursively, the same DeclStmt may be evaluated multiple times in
multiple stack frames. FindLastStoreBRVisitor was not taking this into
account and just picking the first one it saw.
Jordan Rose [Fri, 3 May 2013 05:47:24 +0000 (05:47 +0000)]
[analyzer] Fix trackNullOrUndef when tracking args that have nil receivers.
There were actually two bugs here:
- if we decided to look for an interesting lvalue or call expression, we
wouldn't go find its node if we also knew we were at a (different) call.
- if we looked through one message send with a nil receiver, we thought we
were still looking at an argument to the original call.
Put together, this kept us from being able to track the right values, which
means sub-par diagnostics and worse false-positive suppression.
Adrian Prantl [Fri, 3 May 2013 00:44:13 +0000 (00:44 +0000)]
Attempt to un-break the gdb buildbot.
- Use the debug location of the return expression for the cleanup code
if the return expression is trivially evaluatable, regardless of the
number of stop points in the function.
- Ensure that any EH code in the cleanup still gets the line number of
the closing } of the lexical scope.
- Added a testcase with EH in the cleanup.
John McCall [Fri, 3 May 2013 00:10:13 +0000 (00:10 +0000)]
Move parsing of identifiers in MS-style inline assembly into
the actual parser and support arbitrary id-expressions.
We're actually basically set up to do arbitrary expressions here
if we wanted to.
Assembly operands permit things like A::x to be written regardless
of language mode, which forces us to embellish the evaluation
context logic somewhat. The logic here under template instantiation
is incorrect; we need to preserve the fact that an expression was
unevaluated. Of course, template instantiation in general is fishy
here because we have no way of delaying semantic analysis in the
MC parser. It's all just fishy.
I've also fixed the serialization of MS asm statements.
Ted Kremenek [Thu, 2 May 2013 23:56:33 +0000 (23:56 +0000)]
Change LocationContextMap to be a temporary instead of shared variable in BugReporter.
BugReporter is used to process ALL bug reports. By using a shared map,
we are having mappings from different PathDiagnosticPieces to LocationContexts
well beyond the point where we are processing a given report. This
state is inherently error prone, and is analogous to using a global
variable. Instead, just create a temporary map, one per report,
and when we are done with it we throw it away. No extra state.
Douglas Gregor [Thu, 2 May 2013 23:25:32 +0000 (23:25 +0000)]
Use attribute argument information to determine when to parse attribute arguments as expressions.
This change partly addresses a heinous problem we have with the
parsing of attribute arguments that are a lone identifier. Previously,
we would end up parsing the 'align' attribute of this as an expression
"(Align)":
The code that handles the alignment attribute would completely ignore
the parameter name, so the while the first of these would do what's
expected, the second would silently be equivalent to
i.e., use the maximal alignment rather than the specified alignment.
Address this by sniffing the "Args" provided in the TableGen
description of attributes. If the first argument is "obviously"
something that should be treated as an expression (rather than an
identifier to be matched later), parse it as an expression.
projects / clang / log