John McCall [Tue, 6 Jul 2010 01:34:17 +0000 (01:34 +0000)]
Validated by nightly-test runs on x86 and x86-64 darwin, including after
self-host. Hopefully these results hold up on different platforms.
I tried to keep the GNU ObjC runtime happy, but it's hard for me to test.
Reimplement how clang generates IR for exceptions. Instead of creating new
invoke destinations which sequentially chain to the previous destination,
push a more semantic representation of *why* we need the cleanup/catch/filter
behavior, then collect that information into a single landing pad upon request.
Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional
control flow) are generated, since it's actually fairly closely tied in with
the former. Remove the need to track which cleanup scope a block is associated
with.
Document a lot of previously poorly-understood (by me, at least) behavior.
The new framework implements the Horrible Hack (tm), which requires every
landing pad to have a catch-all so that inlining will work. Clang no longer
requires the Horrible Hack just to make exceptions flow correctly within
a function, however. The HH is an unfortunate requirement of LLVM's EH IR.
Chris Lattner [Mon, 5 Jul 2010 20:21:00 +0000 (20:21 +0000)]
in the "coerce" case, the ABI handling code ends up making the
alloca for an argument. Make sure the argument gets the proper
decl alignment, which may be different than the type alignment.
Although in C++ class name has external linkage, usually the definition of the
class is available in the same translation unit when it's needed. So we make
all of them invalid Entity.
Fix a regression of a previous commit of mine (rdar://8158953).
Some of the invariant checks for creating Record/Enum types don't hold true during PCH reading.
Introduce more suitable ASTContext::getRecordType() and getEnumType().
When adding a visible decl, deserialize the visible decls and add it.
Before this commit, visible decls added before deserialization were ignored.
This was not an issue since name lookup (that usually comes before the addition) forces deserialization
but it is an issue for lazily declared class implicit members.
Don't try to install the __[u]int128_t identifier if it is already installed by PCHReader.
Currently, adding it to visible decls of a PCH'ed translation unit has no effect because
adding visible decls before deserialization has no effect (the decls won't be visible).
This will be fixed in a future commit; then it will force deserialization of visible decls, so avoid pointlessly installing it.
Nick Lewycky [Sun, 4 Jul 2010 21:29:45 +0000 (21:29 +0000)]
Add TypeLoc traversal to the RecursiveASTVisitor! Because the TypeLocs don't
contain all the same information that their Types do, we will fall back to
traversing the Types instead. The default TypeLoc visitor calls the matching
Type visitor so that existing clients should continue to work with no change.
Nick Lewycky [Sun, 4 Jul 2010 01:41:30 +0000 (01:41 +0000)]
Fix ObjCInterfaceTypeLoc to inherit from ObjCObjectTypeLoc so as to match the
Type hierarchy. Without this, TypeLocNodes.def will give you wrong type
hierarchy information (claiming that ObjCObjectTypeLoc is the base of
ObjCInterfaceTypeLoc, which it wasn't).
Jordy Rose [Sun, 4 Jul 2010 00:00:41 +0000 (00:00 +0000)]
Add a new symbol type, SymbolExtent, to represent the extents of memory regions that may not be known at compile-time (such as those created by malloc). This replaces the old setExtent/getExtent API on Store, which used the GRState's GDM to store SVals.
Also adds a getKnownValue() method to SValuator, which gets the integer value of an SVal if it is known to only have one possible value. There are more places in the code that could be using this, but in general we want to be dealing entirely in SVals, so its usefulness is limited.
The only visible functionality change is that extents are now honored for any DeclRegion, such as fields and Objective-C ivars, rather than just variables. This shows up in bounds-checking and cast-size-checking.
Charles Davis [Sat, 3 Jul 2010 16:56:59 +0000 (16:56 +0000)]
Mangle Objective-C pointers and block pointers in the Microsoft C++ Mangler.
ObjC pointers were easy enough (as far as the ABI is concerned, they're
just pointers to structs), but I had to invent a new mangling for block
pointers. This is particularly worrying with the Microsoft ABI, because
it is a vendor-specific ABI; extending it could come back to bite us
later when MS extends it on their own (and you know they will).
When setting the anonymous namespace at PCH reading, it may still be initializing so avoid
the invariant checks at NamespaceDecl::setAnonymousNamespace().
Charles Davis [Sat, 3 Jul 2010 02:41:45 +0000 (02:41 +0000)]
Fix mangling of array parameters for functions in the Microsoft C++ Mangler.
Only actual functions get mangled correctly; I don't know how to fix it for
function pointers yet. Thanks to John McCall for the hint.
Also, mangle anonymous tag types. I don't have a suitable testcase yet; I have
a feeling that that's going to need support for static locals, and I haven't
figured out exactly how MSVC's scheme for mangling those works.
Douglas Gregor [Sat, 3 Jul 2010 00:47:00 +0000 (00:47 +0000)]
Lazily declare default constructors. We now delay the construction of
declarations for implicit default constructors, copy constructors,
copy assignment operators, and destructors. On a "simple" translation
unit that includes a bunch of C++ standard library headers, we
generate relatively few of these implicit declarations now:
4/159 implicit default constructors created
18/236 implicit copy constructors created
70/241 implicit copy assignment operators created
0/173 implicit destructors created
And, on this translation unit, this optimization doesn't really
provide any benefit. I'll do some more performance measurements soon,
but this completes the implementation work for <rdar://problem/8151045>.
Douglas Gregor [Fri, 2 Jul 2010 17:43:08 +0000 (17:43 +0000)]
Move the "current scope" state from the Parser into Action. This
allows Sema some limited access to the current scope, which we only
use in one way: when Sema is performing some kind of declaration that
is not directly driven by the parser (e.g., due to template
instantiatio or lazy declaration of a member), we can find the Scope
associated with a DeclContext, if that DeclContext is still in the
process of being parsed.
Use this to make the implicit declaration of special member functions
in a C++ class more "scope-less", rather than using the NULL Scope hack.
Fix PR 7475 by enhancing the static analyzer to also invalidate bindings for non-static global variables
when calling a function/method whose impact on global variables we cannot accurately estimate.
This change introduces two new MemSpaceRegions that divide up the memory space of globals, and causes
RegionStore and BasicStore to consult a binding to the NonStaticGlobalsMemSpaceRegion when lazily
determining the value of a global.
Jordy Rose [Thu, 1 Jul 2010 20:09:55 +0000 (20:09 +0000)]
Add an ivar to SymbolReaper for the current statement, and then stop passing the current statement around everywhere. Preparation for symbolic extents.
Douglas Gregor [Thu, 1 Jul 2010 17:57:27 +0000 (17:57 +0000)]
Move declaration of a class's implicit copy constructor into a
separate function. Aside from making the loops infinitely faster, no
functionality change.
Douglas Gregor [Thu, 1 Jul 2010 05:10:53 +0000 (05:10 +0000)]
Reinstate fix for PR7526, which was failing because, now that we
aren't dropping all exception specifications on destructors, the
exception specifications on implicitly-declared destructors were
detected as being wrong (which they were).
Introduce logic to provide a proper exception-specification for
implicitly-declared destructors. This also fixes PR6972.
Note that the other implicitly-declared special member functions also
need to get exception-specifications. I'll deal with that in a
subsequent commit.
Daniel Dunbar [Thu, 1 Jul 2010 00:16:31 +0000 (00:16 +0000)]
Revert "IRgen: Make sure any prolog instructions get debug info.", the lexical
scope hasn't been set up yet so this isn't valid. It was just a cleanup to the
IR, so I'm going to ignore it for now.
Douglas Gregor [Thu, 1 Jul 2010 00:00:45 +0000 (00:00 +0000)]
Implement C++ DR481, which clarifies that the scope of template
parameters starts at the end of the template-parameter rather than at
the point where the template parameter name is encounted. For example,
given:
Daniel Dunbar [Wed, 30 Jun 2010 23:56:13 +0000 (23:56 +0000)]
Driver/Darwin: Add one -L for the DarwinClang toolchain to point into the gcc
library directory, as we still rely on a few libraries that are hiding there (in
particular libstdc++.dylib).
projects / clang / log