This change effects both RecordDecls and CXXRecordDecls, but does not effect EnumDecls (yet).
The motivation of this patch is as follows:
- Capture more source information, necessary for refactoring/rewriting clients.
- Pave the way to resolve ownership issues with RecordDecls with the forthcoming
addition of DeclGroups.
Current caveats:
- Until DeclGroups are in place, we will leak RecordDecls not explicitly
referenced by the AST. For example:
typedef struct { ... } x;
The RecordDecl for the struct will be leaked because the TypedefDecl doesn't
refer to it. This will be solved with DeclGroups.
- This patch also (temporarily) breaks CodeGen. More below.
High-level changes:
- As before, TagType still refers to a TagDecl, but it doesn't own it. When
a struct/union/class is first referenced, a RecordType and RecordDecl are
created for it, and the RecordType refers to that RecordDecl. Later, if
a new RecordDecl is created, the pointer to a RecordDecl in RecordType is
updated to point to the RecordDecl that defines the struct/union/class.
- TagDecl and RecordDecl now how a method 'getDefinition()' to return the
TagDecl*/RecordDecl* that refers to the TagDecl* that defines a particular
enum/struct/class/union. This is useful from going from a RecordDecl* that
defines a forward declaration to the RecordDecl* that provides the actual
definition. Note that this also works for EnumDecls, except that in this case
there is no distinction between forward declarations and definitions (yet).
- Clients should no longer assume that 'isDefinition()' returns true from a
RecordDecl if the corresponding struct/union/class has been defined.
isDefinition() only returns true if a particular RecordDecl is the defining
Decl. Use 'getDefinition()' instead to determine if a struct has been defined.
- The main changes to Sema happen in ActOnTag. To make the changes more
incremental, I split off the processing of enums and structs et al into two
code paths. Enums use the original code path (which is in ActOnTag) and
structs use the ActOnTagStruct. Eventually the two code paths will be merged,
but the idea was to preserve the original logic both for comparison and not to
change the logic for both enums and structs all at once.
- There is NO CHAINING of RecordDecls for the same RecordType. All RecordDecls
that correspond to the same type simply have a pointer to that type. If we
need to figure out what are all the RecordDecls for a given type we can build
a backmap.
- The diff in CXXRecordDecl.[cpp,h] is actually very small; it just mimics the
changes to RecordDecl. For some reason 'svn' marks the entire file as changed.
Why is CodeGen broken:
- Codegen assumes that there is an equivalence between RecordDecl* and
RecordType*. This was true before because we only created one RecordDecl* for
a given RecordType*, but it is no longer true. I believe this shouldn't be too
hard to change, but the patch was big enough as it is.
I have tested this patch on both the clang test suite, and by running the static analyzer over Postgresql and a large Apple-internal project (mix of Objective-C and C).
llvm-svn: 55839
The motivation behind this change is that chaining the RecordDecls is simply unnecessary. Once we create multiple RecordDecls for the same struct/union/class, clients that care about all the declarations of the same struct can build a back map by seeing which Decls refer to the same RecordType.
llvm-svn: 55821
Added a couple FIXME's wrt PointLikeType. If the author reads this, it would be great to get some background on this class (thanks in advance).
llvm-svn: 55778
- Change definition of store::Region and store::Binding (once again) to make
them real classes that just wrap pointers. This makes them more strictly
typed, and allows specific implementations of Regions/Bindings to just
subclass them.
- minor renamings to RegionExtent and its subclasses
- added a bunch of doxygen comments
StoreManager: (static analyzer)
- added 'iterBindings', an iteration method for iterating over the bindings of a
store. It that takes a callback object (acting like a poor man's closure).
- added 'getRVal' version for store::Binding. Will potentially phase the other
versions of GetRVal in StoreManager out.
- reimplemented 'getBindings' to be non-virtual and to use 'iterBindings'
BasicStoreManager: (static analyzer)
- implemented 'iterBindings' for BasicStoreManager
llvm-svn: 55688
- Now also searches for correct setter method.
- There are still some issues regarding validation of the setter
method and access of read-only properties.
llvm-svn: 55686
- Remove method 'isForwardDecl'; this functionality is already provided by
'isDefinition()'
- Move method definitions to be co-located with other RecordDecl methods.
llvm-svn: 55649
- Added method 'isForwardDeclaration', a predicate method that returns true
if a RecordDecl represents a forward declaration.
- Added method 'getDefinitionDecl', a query method that returns a pointer to
the RecordDecl that provides the actual definition of a struct/union.
llvm-svn: 55642
- Change constructor and create methods to accept a CXXRecordDecl* (RecordDecl*)
instead of a ScopedDecl* for PrevDecl. This causes the type checking
to be more tight and doesn't break any code.
RecordDecl:
- Don't use the NextDeclarator field in ScopedDecl to represent the previous
declaration. This is a conflated use of the NextDeclarator field, which will
be removed anyway when DeclGroups are fully implemented.
- Instead, represent (a soon to be implemented) chain of RecordDecls using a
NextDecl field. The last RecordDecl in the chain is always the 'defining'
RecordDecl that owns the FieldDecls. The other RecordDecls in the chain
are forward declarations.
llvm-svn: 55640
- Replace FIXME in Preprocessor::HandleIdentifier() with a check that avoids diagnosing extension tokens that originate from macro definitions.
llvm-svn: 55639
Note that this isn't really a complete fix; I think there are other
potential overrun situations. I don't really know what the best
systematic fix is, though.
llvm-svn: 55622
anywhere near correct in terms of missing cases and missing
diagnostics, but it's good enough to handle the uses in the
Linux system headers, which are currently a constant pain for compiling
applications on Linux.
llvm-svn: 55621
- warn about nonnull being applied to functions with no pointer arguments
- continue processing argument list in the attribute when we encounter a non-pointer parameter being marked as nonnull
- when no argument list is specified, only mark pointers as nonnull. This fixes PR 2732 and radar 6188814.
llvm-svn: 55610
- Change Obj-C runtime message API, drop the ObjCMessageExpr arg in
favor of just result type and selector. Necessary so it can be
reused in situations where we don't want to cons up an
ObjCMessageExpr.
- Update aggregate binary assignment to know about special property
ref lvalues.
- Add CodeGenFunction::EmitCallArg overload which takes an already
emitted rvalue.
Add CodeGenFunction::StoreComplexIntoAddr.
Disabled logic in Sema for parsing Objective-C dot-syntax that
accesses methods. This code does not search in the correct order and
the AST node has no way of properly representing its results.
Updated StmtDumper to print a bit more information about
ObjCPropertyRefExprs.
llvm-svn: 55561
- Added CodeGenFunction::EmitCall which just takes the callee, return
type, and a list of (Value*,QualType) pairs.
- Added CodeGenFunction::EmitCallArg which handles emitting code for
a call argument and turning it into an appropriate
(Value*,QualType) pair.
- Changed Objective-C runtime interface so that the actual emission
of arguments for message sends is (once again) done in the code to
emit a message send.
No intended functionality change, this is prep work for better ABI
support and for Objective-C property setter support.
llvm-svn: 55560
Migrated CFRefCount.cpp to use getBindings and BindingsAsString instead of
making assumptions about the Store (removed dependence on GRState::vb_iterator).
llvm-svn: 55522
Their precise semantics will be implemented by a specific StoreManager.
Use function pointer to create the StoreManager in GRStateManager. This matches how we create ConstraintsManager.
llvm-svn: 55514
- It is not clear that this is the right approach, but this is at
least consistent with how interfaces are handled.
- This means NeXT now emits the correct metadata for properties in
protocols.
- This currently introduces a spurious warning involving inherited
properties in protocols or categories; however, it also fixes some
situations where we were failing to emit a warning. I will scrub
this code tomorrow and fix this issue as well as number of other
missed warnings / error situations that appear to exist.
llvm-svn: 55407
- Only supports simple assignment and atomic semantics are ignored.
- Not quite usable yet because the methods do not actually get added
to the class metadata.
- Added ObjCPropertyDecl::getSetterKind (one of Assign, Copy, Retain).
- Rearrange CodeGenFunction so synthesis can reuse function prolog /
epilog code.
llvm-svn: 55365
- Change enum name to Kind.
- Change enum constants to English strings.
Also, fix getPropertyImplementation (which probably should be renamed)
llvm-svn: 55354
of RHSs to id type instead of void* if either has Objective-C object
type.
- This ensures the result can still be used in normal places an
object can be used, like a message send.
Add implicit conversions for ?: applied to qualified id types to
ensure that the RHSs are compatible.
- This prevents a codegen crash (creating invalid PHI nodes).
- Again, this relates to the fact that qualified id types have no
canonical types.
- Note that the implicit type casted to is incorrect, however this
doesn't currently cause problems because of the flexibility of the
id type.
Test cases for above.
llvm-svn: 55346
casting pointers to integers.
Eventually, we should check whether we can evaluate an expression
using Expr::tryEvaluate, and this codepath should be tightened to only
handle standard-compliant cases.
llvm-svn: 55331
Add CodeGenFunction::EmitUnsupportedLValue
- Gives error and returns undef value.
Swap some asserts() over to using EmitUnsupportedLValue
- Rumor has it users (and even some developers) prefer carat
diagnostics to backtraces.
- Works better in Release-Asserts to boot.
llvm-svn: 55328
EmitScalarConversion().
- Important for allowing Obj-C void * to id<X> casts and so on.
- Not sure about this fix however, perhaps Type should understand
that id is effectively a pointer type.
llvm-svn: 55311
- This ensures that references to undefined classes cause link
errors.
- NOTE: This relies on platform specific asm directives currently,
this should be factored out.
Also, don't emit a SYMBOLS metadata entry if there are no symbols.
llvm-svn: 55302
an APInt directly to an ostream now, so add some hacks. It would
be better to switch all of the bugreport (and friends) stuff over
to raw_ostream.
llvm-svn: 55264
Implement Obj-C lvalue message sends (aggregate returns).
Update several places to emit more precise ErrorUnsupported warnings
for currently unimplemented Obj-C features (main missing chunks are
property references, Obj-C exception handling, and the for ... in
syntax).
llvm-svn: 55234
- Used to autoselect runtime when neither -fnext-runtime nor
-fgnu-runtime is specified.
- Default impl is false, all darwin targets set it to true.
llvm-svn: 55231
- Returns an RValue.
- Reduced to only taking the CodeGenFunction, Expr, and Receiver.
- Becomes responsible for emitting the arguments.
Add CodeGenFunction::EmitCallExprExt
- Takes optional extra arguments to insert at the head of the call.
- This allows the Obj-C runtimes to call into this and isolates the
argument and call instruction generation code to one place. Upshot
is that we now pass structures (more) correctly.
Also, fix one aspect of generating methods which take structure
arguments (for NeXT). This probably needs to be merged with the
SetFunctionAttributes code in CodeGenModule.cpp
llvm-svn: 55223
-The Parser calls a new "ActOnCXXTypeConstructExpr" action.
-Sema, depending on the type and expressions number:
-If the type is a class, it will treat it as a class constructor. [TODO]
-If there's only one expression (i.e. "int(0.5)" ), creates a new "CXXFunctionalCastExpr" Expr node
-If there are no expressions (i.e "int()" ), creates a new "CXXZeroInitValueExpr" Expr node.
llvm-svn: 55177
For example, adding the default system include paths in clients is now as
simple as
InitHeaderSearch init(headers);
init.AddDefaultSystemIncludePaths(langopts);
init.Realize();
llvm-svn: 55174
testing compatibility. This is necessary for some constructs, like merging
redeclarations.
Also, there are some ObjC changes to make sure that
typesAreCompatible(a,b) == typesAreCompatible(b,a). I don't have any
ObjC code beyond the testsuite, so please tell me if there are any cases
where this doesn't behave as expected.
llvm-svn: 55158
- Functional for simple cases but there are some glaring omissions
(protocols, properties, and other language extensions).
- The code needs a fair bit of cleaning.
llvm-svn: 55108
difference from generic x86 is the defines. The rest is non-trivial to
implement.
I'm not planning on adding any more targets myself; if there are any
targets anyone is currently using that are missing, feel free to add
them, or ask me to add them.
This concludes the work I'm planning for the TargetInfo
implementations at the moment; all the other issues with TargetInfo require
some API changes, and I haven't really thought it through. Some of the
remaining issues: allowing targets to define size_t and wchar_t properly,
adding some sort of __builtin_type_info intrinsic so we can finish clang's
limits.h and float.h and get rid of a massive number of macro
definitions, allowing target-specific command-line options, allowing
target-specific defaults for certain command-line options like
-fms-extensions, exposing vector alignment outside of the description
string, exposing endianness outside of the description string, allowing
targets to expose special bit-field layout requirements, exposing some
sort of custom hook for call generation in CodeGen, and adding CPU
selection to control defines like __SSE__.
llvm-svn: 55098
This approach allows adding OS-specific targets/defines/etc. without
completely breaking unknown subtargets. No new subtargets yet, although
I plan to add x86-Linux soon. Others can add targets that they use as
needed; adding a new subtarget takes very little code.
Also does some fixups for description strings; a lot of them were
unspecified. I think all the ones I added are correct, but
they're unverified; corrections are welcome.
llvm-svn: 55091
cleaned it up a bit, including fixing the definition of va_list; this
shouldn't break anything, but anyone using Sparc should watch for
regressions.
llvm-svn: 55041
- UsualUnaryConversions takes an Expr *& and may modify its argument,
this broke when it was refactored into Sema::CheckCastTypes. This
meant that we were missing implicit casts in some places.
- Seems pretty sad that this got through our tests.
llvm-svn: 55039
visible effects, but this will significantly reduce the amount of
boilerplate code necessary to add subtargets.
If this looks okay, I'll do the rest of the processors (PPC, Sparc, ARM)
soon.
llvm-svn: 55036
This patch extends BasicStoreManager::getInitialStore() to include code that symbolicates input variables.
It also removes redundant handling of ImplicitParamDecl, since it is a subclass of VarDecl.
llvm-svn: 54993
Change CodeGenFunction::EmitParmDecl to take either a ParmVarDecl or an
ImplicitParamDecl.
Drop hasAggregateLLVMType from CodeGenModule.cpp (use version in
CodeGenFunction).
Change the Objective-C method generation to use EmitParmDecl for
implicit parameters.
llvm-svn: 54838
- Remove ConstNotEq from GRState/GRStateManager (!= tracking uses GDM instead).
- GRStateManager now can book-keep "contexts" (e.g., factory objects) for uses
with data elements stored into the GDM.
- Refactor pretty-printing of states to use GRState::Printer objects
exclusively. This removed a huge amount of pretty-printing logic from
GRExprEngine.
CFRefCount
- Simplified some API calls based on refinements to the GDM api.
llvm-svn: 54835
ObjCInterfaceDecl.
Change ObjCRuntime::GenerateMessageSendSuper to take the
ObjCInterfaceDecl for the super class, instead of just its name.
Change EmitObjCMessageExpr to make the right runtime calls for super
sends in class methods (i.e. a super send with the class object as
the receiver).
llvm-svn: 54833
- Add CodeGenModule::EmitTopLevelDecl which uses switch on kind
instead of ugly & slow dyn_cast chains.
- Drop some simple forwarding methods into the ObjC runtime.
- Privatize appropriate methods.
llvm-svn: 54827
- This is in prep for implementation class support for the NeXT
runtime, for which the existing methods don't provide enough
information (and additionally make too many assumptions about how
things should be emitted).
llvm-svn: 54824
- Added as private members for each because it is not clear where to
put the common definition. Perhaps the IdentifierInfos all of these
"pseudo-keywords" should be collected into one place (this would
KnownFunctionIDs and Objective-C property IDs, for example).
Remove Token::isNamedIdentifier.
- There isn't a good reason to use strcmp when we have interned
strings, and there isn't a good reason to encourage clients to do
so.
llvm-svn: 54794
of using CheckerState).
Removed CheckerState from GRState.
Added class GRStateRef which wraps GRState* and GRStateManager*. This is handy
for generating new states with a single handle.
Added member template set/get functions to GRStateRef/GRState/GRStateManager for
accessing the Generic Data Map.
llvm-svn: 54788
and put the caret on the ()'s. e.g. produces:
t.c:13:9: error: called object is not a function or function pointer
((B)a)();
~~~~~~^
instead of:
t.c:13:3: error: called object is not a function or function pointer
((B)a)();
^~~~~~~~
llvm-svn: 54775
- Returns addr of constant for argument + '\0'.
- I couldn't think of a better name.
- Move appropriate users of GetAddrOfConstantString to this.
Rename getStringForStringLiteral to GetStringForStringLiteral.
Add GetAddrOfConstantStringFromLiteral
- This combines GetAddrOfConstantString and
GetStringForStringLiteral. This method can be, but is not yet, more
efficient.
Change GetAddrOfConstantString to not add terminating '\0'
- <rdar://problem/6140956>
llvm-svn: 54768
The following case resulted in an invalid start location:
// start location not being set for ObjCProtocolDecl AST (when a forward reference is in scope).
@protocol Buggy;
@protocol Buggy
@optional
- whatever;
@end
llvm-svn: 54740
ObjCProtocolDecl directly.
Implement CodeGen support for forward protocol decls (no-ops are so
nice to implement).
Also moved CGObjCRuntime.h out of CodeGenModule.h
llvm-svn: 54709
Changed CGObjCRuntime::GenerateConstantString interface to take
std::string instead of char* and size.
Change ObjC functions which call on GenerateConstantString to bitcast
result to appropriate type.
llvm-svn: 54659
- Maps to LangOptions.Exceptions
- Currently always off, should autoselect based on language.
Update CodeGen to set unwind attribute on functions definitions based
on LangOptions.Exceptions.
- Still need to set attributes appropriately on calls.
llvm-svn: 54643
- Kill unnecessary #includes in .cpp files. This is an automatic
sweep so some things removed are actually used, but happen to be
included by a previous header. I tried to get rid of the obvious
examples and this was the easiest way to trim the #includes in one
fell swoop.
- We now return to regularly scheduled development.
llvm-svn: 54632
- Drop {Decl.h,DeclObjC.h,IdentifierTable.h} from Expr.h
- Moved Sema::getCurMethodDecl() out of line (dependent on
ObjCMethodDecl via dyn_cast).
llvm-svn: 54629
temporarily, I assumed GetAddrForConstantString literal was being
used consistently but it doesn't look like it is.
Factored out a CodeGenModule::getStringForStringLiteral which handles
extracting a std::string for the bytes of a StringLiteral, padded to
match the type.
Update EmitLValue to use getStringForStringLiteral, this was
previously not padding strings correctly. Good thing we only emit
strings in 4 different places!
llvm-svn: 54621
* Move FormatError() from TextDiagnostic up to DiagClient, remove now
empty class TextDiagnostic
* Make DiagClient optional for Diagnostic
This fixes the following problems:
* -html-diags (and probably others) does now output the same set of
warnings as console clang does
* nothing crashes if one forgets to call setHeaderSearch() on
TextDiagnostic
* some code duplication is removed
llvm-svn: 54620
1) New public methods added:
-EnableBacktrackAtThisPos
-DisableBacktrack
-Backtrack
-isBacktrackEnabled
2) LookAhead() implementation is replaced with a more efficient one.
3) LookNext() is removed.
llvm-svn: 54611
member count. The count returned by numStructUnionElements is the
number of initializers that will be consumed, not the number of members
to iterate through. Fixes PR2534.
llvm-svn: 54601
Even though the test case this fixes is in "tentative-decls.c", this bug didn't have anything to do with our handling of tentative definitions (which is what I first expected). In any event, this is a tricky area of the spec.
llvm-svn: 54583
- Move checking from MergeVarDecl->FinializeDeclaratorGroup. Since MergeVarDecl is called before the initializer is attacted, it can't be done there (this removes a long standing FIXME).
- Add Sema::isTentativeDefinition() and Sema::CheckForFileScopedRedefinitions().
- Remove FIXME's and touch-up test case.
Still some more work to do (forthcoming)...
llvm-svn: 54533
The motivation is that clients may either:
(a) query the ASTConsumer object after AST parsing to collect data/etc.
(b) reuse the ASTConsumer.
llvm-svn: 54502
type.
- This generates somewhat less optimal code than before but this is
not hard to rectify once stable (at the cost of slightly more
complex code).
- This currently always uses little-endian ordering of the bitfield.
- This breaks the CodeGen/bitfield.c test because it was grepping for
hard-coded assembly instructions. Will fix once a better test case
is constructed (hard to do without execution).
- This fixes SingleSource/UnitTests/2006-01-23-InitializedBitField.c
and Regression/C/PR1386.c from the test suite.
- <rdar://problem/6085090>, <rdar://problem/6094169>
llvm-svn: 54395
- Changed EmitSub to work on all cases (arithmetic, ptr - int, ptr -
ptr) so that implementation matches that of other operators.
- Modify EmitCompoundAssign to compute and perform the appropriate
casts of left, right, and result types for the assorted pointer
arithmetic cases.
- Fix EmitSub (ptr - int) case to negate the rhs *after*
extension. This is critical when the rhs is unsigned (and needs
extension).
- This fixes cfrac.
- <rdr://6115726>
llvm-svn: 54392
scimark2 on Darwin.
- Added Sema support for asm-label on variables, which I forgot before.
- Update CodeGen to use GlobalDeclMap to determine if static Decls
require emission (instead of LLVM module name lookup). Important
since the Decl name and the LLVM module name can differ.
- <rdar://problem/6116729>
llvm-svn: 54388
having multiple bindings from all the possible decls which
conceptually map to the same global).
- This eliminates CodeGen depending on the LLVM module for name
lookup.
- This also eliminates the need for ReplaceMapValuesWith (hurrah).
- This handles lookups for FunctionDecls correctly in the presence of
aliases, this was previously broken.
- WIP: Can still clean up & unify variable and function emission.
llvm-svn: 54382
Updated a few clients of DeclStmt::getNextDeclarator() to use decl_iterator instead. Will update other clients after additional testing.
llvm-svn: 54368
- ActOnDeclarator now takes an additional parameter which is the
AsmLabel if used. Its unfortunate that this bubbles up this high,
but we cannot just lump it in as an attribute without mistakenly
*accepting* it as an attribute.
- The actual asm-label itself is, however, encoded as an AsmLabelAttr
on the FunctionDecl.
- Slightly improved parser error recovery on malformed asm-labels.
- CodeGen support still missing...
llvm-svn: 54339
- Like EXTENSION but always generates a warning (even without
-pedantic).
- Updated ptr -> int, int -> ptr, and incompatible cast warnings to
be EXTWARN.
- Other EXTENSION level diagnostics should be audited for upgrade.
- Updated several test cases to fix code which produced unanticipated
warnings.
llvm-svn: 54335
- Follows emission scheme used by llvm-gcc, i.e. invent an id for
each label whose address is taken and replace each indirect goto by
a switch to each possible target.
- Currently we emit a switch for each indirect goto instead of
merging them as llvm-gcc does.
llvm-svn: 54318
move getAsArrayType into ASTContext instead of being a method on type.
This is required because getAsArrayType(const AT), where AT is a typedef
for "int[10]" needs to return ArrayType(const int, 10).
Fixing this greatly simplifies getArrayDecayedType, which is a good sign.
llvm-svn: 54317
- There is an miscompilation issue remaining due to a poor
interaction between the delayed emission of static functions and
the emission of constructors, but that already existed prior to
this change.
llvm-svn: 54258
instead of mapping the decl to a bitcast of the global to the correct
type.
- GetAddrOf{Function,GlobalVar} introduce the bitcast on every use now.
- This solves a problem where a dangling pointer could be introduced
by the RAUW done when replacing a forward or tentative
definition. See testcase for more details.
- Fixes <rdar://problem/6108358>
llvm-svn: 54211
- No (intended) functionality change.
- Primary purpose is to clearly separate (lazy) construction of
globals that are a forward declaration or tentative definition from
those that are the final definition.
- Lazy construction is now encapsulated in
GetAddrOf{Function,GlobalVar} while final definitions are
constructed in EmitGlobal{Function,Var}Definition.
- External interface for dealing with globals is now limited to
EmitGlobal and GetAddrOf{Function,GlobalVar}.
- Also updated helper functions dealing with statics, annotations,
and ctors to be private.
llvm-svn: 54179
1) reject stuff like "id<foo> short" and "<foo> short". 2) set
the declspec range correctly 3) only parse protocol qualifiers
when in objc mode.
llvm-svn: 54086
First, fix canonical type handling of these, since protocol qualified id's are always
canonical. Next, enhance SemaType to actually make these when used (instead of int)
allowing them to actually be used when appropriate. Finally remove a bunch of logic
relating to the mishandling of canonical types with protocol-qual id's. This fixes
rdar://5986251
llvm-svn: 54083
like "id<foo>". This 1) fixes an infinite loop in the parser on things
like "short<foo>" 2) emits a warning about this bogus construct and 3)
changes the testcase to be substantially reduced.
llvm-svn: 54082
problem with &&/||. 2) use canonical types for comparison instead
of raw types. 3) emit an ext-warn for a gnu extension.
Also simplify the code to make it less nested.
llvm-svn: 54068
(a) removes a bogus warning.
(b) removes an undesirable usage of the ObjCMessageExpr constructor that takes an IdentifierInfo * (which I will abolish).
llvm-svn: 54042
t2.c:1:17: warning: hexadecimal floating constants are a C99 feature
long double d = 0x0.0000003ffffffff00000p-16357L;
^
instead of emitting a weird error message that doesn't make sense:
t2.c:1:41: error: hexadecimal floating constants require an exponent
long double d = 0x0.0000003ffffffff00000p-16357L;
^
rdar://6096838
llvm-svn: 54035
- Fix multiple issues with the way case ranges were emitted, see test
cases details about the specific issues.
- The root issue was not being careful about how basic blocks were
emitted which led to them being chained together incorrectly,
resulting in improper control flow.
- Fixes <rdar://problem/6098585>
llvm-svn: 54006
This fixes a critical rewriter bug (<rdar://problem/6096760> clang ObjC rewriter: 'self' not expected value in class method called with 'super').
Also added a couple FIXME's since I'm not happy with my fix to Sema. It would be nicer if the super handling for class/instance messages was the same (based on PreDefinedExpr).
llvm-svn: 53994
- Also cleaned up emission slightly
- Inspection of the code revealed several other bugs, however. Case
ranges are not properly wired and can result in switch cases being
dropped or even infinite loops. See: <rdar://problem/6098585>
Completes: <rdar://problem/6094119>
llvm-svn: 53975
This change also fixes a subtle bug where the access control of an ivar would be initialized to garbage if we didn't have an explicit visibility specifier (e.g., @private).
llvm-svn: 53955
of a specific smallvector size.
Fix protocol lists to pass down proper location info, so we get diagnostics
like this:
t.m:3:35: error: cannot find protocol definition for 'NSCopying', referenced by 'NSWhatever'
@interface NSWhatever : NSObject <NSCopying>
^
instead of this:
t.m:3:44: error: cannot find protocol definition for 'NSCopying', referenced by 'NSWhatever'
@interface NSWhatever : NSObject <NSCopying>
^
Add a new IdentifierLocPair typedef which is just a pair<IdentifierInfo*, SourceLocation>
llvm-svn: 53883
the standard "set these as the list of protocols" interface instead of a
strange "set this as the size and then set each one to the value" interface.
The problem with the later is that it a) is completely different from
everything else, b) is awkward, and c) doesn't handle the case when a
referenced protocol is invalid: it set it to null.
This meant that all clients downstream would have to handle null protocols
in the protocol list, and empirically they didn't. Fix this by not setting
invalid protocols in the referenced protocol list, fixing the crash on
test/Sema/objc-interface-1.m
While I'm at it, clean up some locations so that we produce:
t.m:1:25: error: cannot find interface declaration for 'NSObject', superclass of 'NSWhatever'
@interface NSWhatever : NSObject <NSCopying>
~~~~~~~~~~~~~~~~~~~~~ ^
instead of:
t.m:1:1: error: cannot find interface declaration for 'NSObject', superclass of 'NSWhatever'
@interface NSWhatever : NSObject <NSCopying>
^
llvm-svn: 53846
have better source ranges. Before:
t.m:11:53: error: member reference is not to a structure or union
CGFloat maxOffsetY = [_outlineLayer contentSize].height - [_outlineLayer frame].size.height;
^~~~~~~
after:
t.m:11:54: error: member reference base type ('id') is not a structure or union
CGFloat maxOffsetY = [_outlineLayer contentSize].height - [_outlineLayer frame].size.height;
~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^
llvm-svn: 53834
from:
t.m:8:7: error: member reference is not to a structure or union
pool->farm = 0;
^ ~~~~
to:
t.m:8:7: error: 'NSAutoreleasePool' has member named 'farm'
pool->farm = 0;
~~~~^ ~~~~
llvm-svn: 53824
interfaces. Just because they x->isPointerType() doesn't mean it is
valid to just cast to a pointertype. We have to handle typedefs etc
as well.
llvm-svn: 53819
Modified the new EvalBinOpNN to generate states instead of nodes. This is a much simpler interface and is what clients will want to do.
llvm-svn: 53750
- Make sure ObjCIvarDecl propagates the bitfield width.
- RewriteObjC::SynthesizeIvarOffsetComputation(): Avoid using the __OFFSETOF__ mumbo jumbo for bitfields (since it isn't legal C). This fixes <rdar://problem/5986079> clang ObjC rewriter: bitfields and ivar access don't mix.
llvm-svn: 53694
The regression was the casts from integers to pointers where not being handled: they would just return UnknownVal. This would greatly decrease path-sensitivity.
llvm-svn: 53659
Add some code to handle vector comparisons, which is the language side
of the llvm vicmp/vfcmp instructions. Also make the vector-vector and
vector-scalar asign checks a bit more sane under the presence of lax vector
conversions.
llvm-svn: 53565
Refactored error reporting in CheckObjCDealloc and CheckObjCInstMethSignature to use this new bug reporting interface (major code simplification).
llvm-svn: 53560
ObjCImplementationDecls and sees if a ancestor class defines a method with the
same selector but with a different type signature. Right now it just compares
return types, and mainly looks at differences in primitive values. The checking
will be expanded in the future.
llvm-svn: 53482
current store implementation is now encapsulated by BasicStore.
These changes prompted some long due constification of ValueState. Much of the
diffs in this patch include adding "const" qualifiers.
llvm-svn: 53423
-Remove ParseExpressionWithLeadingIdentifier and ParseAssignmentExprWithLeadingIdentifier.
-Separate ParseLabeledStatement from ParseIdentifierStatement.
llvm-svn: 53376
Constraints. These concepts are already present in the current ValueState, but
the implementation is monolothic. Making ValueState more modular opens up new
design choices for customizing the analysis engine.
In the context of the analysis engine, the "Environment" is the binding between
Expr* (expressions) and intermediate symbolic values (RValues).
llvm-svn: 53252
also fix the correspondent test (it was expecting more errors than it should. please confirm my fix is correct (at least gcc agrees with me)
llvm-svn: 53174
passed-by-reference to a function. This allows us to build up constraints for
their new values and restore some lost path-sensitivity. This addresses a few
false positives since in Adium.
llvm-svn: 53125
http://lists.cs.uiuc.edu/pipermail/cfe-dev/2008-July/002157.html
Essentially the observer mechanism in LiveVariables was observing block-level
expressions multiple times, leading to a case where the dead store checker could
see a value as dead when it was really live.
llvm-svn: 53115
1) add a new ASTContext::getFloatTypeSemantics method.
2) Use it from SemaExpr.cpp, CodeGenTypes.cpp and other places.
3) Change the TargetInfo.h get*Format methods to return their
fltSemantics byref instead of by pointer.
4) Change CodeGenFunction::EmitBuiltinExpr to allow builtins which
sometimes expand specially and othertimes fall back to libm.
5) Add support for __builtin_nan("") to codegen, cases that don't pass
in an empty string are currently lowered to libm calls.
6) Fix codegen of __builtin_infl.
llvm-svn: 52914
decl attributes out of the various places they can hide. This makes
us correctly reject things like this:
t.c:2:22: error: mode attribute only supported for integer and floating-point types
int **__attribute((mode(HI)))* i32;
^
because you can't make a pointer be HImode.
llvm-svn: 52876
to work list the rest of the attr handlers. Also, rename
it to HandleVectorSizeAttribute to match its attr name.
No functionality change.
llvm-svn: 52872
the single attribute they look at by reference instead of by pointer.
This is a subtle indicator that they take the specified attribute, not
a whole list of them.
This also make HandleExtVectorTypeAttribute work the same way as the rest
of the attributes, adds some comments etc. No functionality change.
llvm-svn: 52871
integers which have the same width and different signedness work
correctly. (The testcase in PR2501 uses a comparison between long and
unsigned int).
llvm-svn: 52853
used to mutate the attribute list for declspecs when the type was
converted, breaking the case where one declspec was shared by multiple
declarators.
This fixes rdar://6032532.
llvm-svn: 52769
Tracked objects now have their type information tracked with them.
Enhanced summaries for ObjC methods to include the type information of the receiver.
Used the enhanced summaries to support the idiom that NSWindow owns itself (it sends a release message to itself upon close).
Added some comments.
Did some cleanups with the checker logic using operator overloading (reduced redundant code which I was concerned about being the source of bugs).
llvm-svn: 52741
Note that Parser::ParseCXXMemberSpecification is temporarily disabled until the Sema support is in place.
Once ParseCXXMemberSpecification is enabled, the Parser/cxx-class.cpp test will pass.
llvm-svn: 52694
1) Check if a dead store appears as a subexpression. For such cases, we emit
a verbose diagnostic so that users aren't confused. This addresses:
<rdar://problem/5968508> checker gives misleading report for dead store in loop
2) Don't emit a dead store warning when assigning a null value to a pointer.
This is a common form of defensive programming. We may wish to make
this an option to the the checker one day.
This addresses the feature request in the following email:
http://lists.cs.uiuc.edu/pipermail/cfe-dev/2008-June/001978.html
llvm-svn: 52555
clang as a Release build.
The big change is that all AST nodes (subclasses of Stmt) whose children are
Expr* store their children as Stmt* or arrays of Stmt*. This is to remove
strict-aliasing warnings when using StmtIterator. None of the interfaces of any
of the classes have changed (except those with arg_iterators, see below), as the
accessor methods introduce the needed casts (via cast<>). While this extra
casting may seem cumbersome, it actually adds some important sanity checks
throughout the codebase, as clients using StmtIterator can potentially overwrite
children that are expected to be Expr* with Stmt* (that aren't Expr*). The casts
provide extra sanity checks that are operational in debug builds to catch
invariant violations such as these.
For classes that have arg_iterators (e.g., CallExpr), the definition of
arg_iterator has been replaced. Instead of it being Expr**, it is an actual
class (called ExprIterator) that wraps a Stmt**, and provides the necessary
operators for iteration. The nice thing about this class is that it also uses
cast<> to type-checking, which introduces extra sanity checks throughout the
codebase that are useful for debugging.
A few of the CodeGen functions that use arg_iterator (especially from
OverloadExpr) have been modified to take begin and end iterators instead of a
base Expr** and the number of arguments. This matches more with the abstraction
of iteration. This still needs to be cleaned up a little bit, as clients expect
that ExprIterator is a RandomAccessIterator (which we may or may not wish to
allow for efficiency of representation).
This is a fairly large patch. It passes the tests (except CodeGen/bitfield.c,
which was already broken) on both a Debug and Release build, but it should
obviously be reviewed.
llvm-svn: 52378
qualifier in the lvalue, and changes lvalue loads/stores to honor
the volatile flag. Places which need some further attention are marked
with FIXMEs.
Patch by Cédric Venet.
llvm-svn: 52264
isn't guaranteed to exist. This fixes a crash with conflicting typedefs
coming from stdin.
This also fixes the crash in PR2406, but doesn't completely fix the
issue; it appears there's something strange about the physical location
for the definition of int64_t in stdlib.h.
llvm-svn: 52209
pointer cast hack currently in isIntegerConstantExpr
(in lib/AST/Expr.cpp). Also removes an odd test that even gcc doesn't accept.
The reason the pointer cast hack is relevant here is that it makes Sema
end up misinterpreting the relevant expression as a null pointer constant.
The reason for this patch is that I plan to remove the pointer cast hack
sometime soon because it causes strange issues, especially in its
current form; see my recent email to cfe-dev
"[PATCH] add constant expression evaluation to the AST and fix PR2413".
llvm-svn: 52120
type; this isn't explicitly stated in the standard, but it doesn't
really make sense for them to have an effect here. Fixes the included
testcase, sent to me by Steve Naroff.
llvm-svn: 52113
reported on cfe-dev by Cédric Venet.
Note that I seriously doubt that this perticular construct is useful,
though: it's a pointer in an alternate address space pointing into
unqualified address space.
llvm-svn: 52076
have Destroy methods of ObjcMethodDecl and ObjCInterfaceDecl which recursively
destroy their owned Decls and Stmts. There are a few cases where it is not
clear what to do (FIXMEs included in the patch).
llvm-svn: 52050
- Changed Sema::ObjCActOnStartOfMethodDef() to more accurately type "self" in factory methods.
- Changed Sema::ActOnInstanceMessage() to use the new type to restrict the lookup.
llvm-svn: 52005
In addition to fixing the crasher, this commit fixes further improves property lookup (by searching protocols of qualified interfaces..."NSObject <prot>").
llvm-svn: 52001
/* Make "<SomeProtocol>" equivalent to "id <SomeProtocol>" - nisse@lysator.liu.se. */
This commit adds the parser magic. The type associated with <p> is still incorrect. Will discuss with Chris.
llvm-svn: 51972
hardcoded data layout in getTargetDescription. Hopefully fixes a test
failure.
Of course, this should be fixed properly, but that's a bigger fix.
llvm-svn: 51948
moves the check for the invalid construct to a point where it doesn't
affect other uses of isIntegerConstantExpr, and we can warn properly
when the extension is used. This makes it a bit more complicated, but
it's a lot cleaner.
Steve, please tell me if this check is sufficient to handle the
relevant system header. I know it's enough to handle the testcase, but
I don't know what exactly the original looks like.
llvm-svn: 51918
much closer to passing the gcc struct layout tests.
It might be possible to refactor this a bit, but I'm not sure there's
actually enough common code for that to be useful.
To get the calling convention completely correct, a bit of
platform-specific code is necessary even for x86-Linux. On x86-Linux, the
alignment of function parameters is extremely strange; as far as I can tell,
it's always 4 except for SSE vectors or structs containing SSE vectors. I'm
continuing to investigate this.
llvm-svn: 51839
This eliminates a bogus warning identified in the test below.
This fixes <rdar://problem/5968256> clang on xcode: error: incompatible type initializing 'NSObject<XCSelectionSource> *', expected 'id<NSObject,XCSelectionSource>'
llvm-svn: 51832
TranslationUnit object instead of an ASTContext. By default it calls
Initialize(ASTConstext& Context) (to match with the current interface used by
most ASTConsumers).
Modified the ObjC-Rewriter to use InitializeTU, and to tell the TranslationUnit
to not free its Decls. This is a workaround for: <rdar://problem/5966749>
llvm-svn: 51825
this does is reconstruct the type for structs and arrays if the type
wouldn't be compatible otherwise.
The assertion about packing in the struct type reconstruction code
sucks, but I don't see any obvious way to fix it. Maybe we need a general
utility method to take a list of types and alignments and try to construct an
unpacked type if possible?
llvm-svn: 51785
associated declaration. This is a prerequisite to handling
general union initializations; for example, an array of unions involving
pointers has to be turned into a struct because the elements can have
incompatible types.
I refactored the code a bit to make it more readable; now, the logic for
definitions is all in EmitGlobalVarInit.
The second parameter for GetAddrOfGlobalVar is now dead; I'll remove it
separately.
By itself, this patch should not cause any visible changes.
llvm-svn: 51783
required by the standard (the standard doesn't know anything about
implicit casts).
Disallow pointers cast to non-integral arithmetic types as constant
expressions. This was previously allowed by accident.
llvm-svn: 51779
bit-field initialization; ugly code, X86-only, but it works, at least
for basic stuff. Separates/adds union initialization; currently disabled,
though, because the struct/array code needs modifications to support
elements of the wrong type.
Fixes PR2381 and PR2309 with the bit-field initialization. And NetHack
compiles and appears to work with a few tweaks (to work around the lack
of transparent_union support, and clang being a bit strict about
conflicting declarations).
llvm-svn: 51763
and union codepaths and fixes some minor bugs.
I'm reasonably confident this is accurate, at least for X86. I'll
correct any bugs as I find them; I haven't found any for a while,
though.
llvm-svn: 51762
While it is far from complete, it does fix the following <rdar://problem/5967199> clang on xcode: error: member reference is not to a structure or union
llvm-svn: 51719
emit incomplete types, because they crash llc, and always use the
logical location as the current location so we don't crash doing invalid
queries on CurLoc.
llvm-svn: 51675
nothing fundamentally wrong with it. Emitting unpacked structs where
possible is more work for almost no practical benefit. We'll probably
want to fix it at some point anyway, but it's low priority.
The issue with long double in particular is that LLVM thinks an X86 long
double is 10 bytes, while clang considers it for all purposes to be
either 12 or 16 bytes, depending on the platform, even in a packed
struct.
llvm-svn: 51673
encountered. Mixing up the decls is unintuitive, and confuses the AST
destruction code. Fixes PR2360.
Note that there is a need to look up the characteristics and
declarations of a function associated with a particular name or decl,
but the original swapping code doesn't solve it properly.
http://lists.cs.uiuc.edu/pipermail/cfe-dev/2008-May/001644.html is one
suggestion for how to fix that.
llvm-svn: 51584
it fixes PR2204. Not too much to say about the implementation; it works
in a similar way to the vector size attribute.
At some point, we need to modify the targets to provide information
about the appropriate types.
llvm-svn: 51577
a few bugs, but I don't know of any in particular. This has good effects
besides cleanup, though: it also should make it easier to implement the
aligned and packed attributes, and also target-specific struct layouts,
because the code won't have to be duplicated in codegen.
llvm-svn: 51576
required for correctness in cases of copying a struct to itself or to
an overlapping struct (itself for cases like *a = *a, and overlapping
is possible with unions).
Hopefully, this won't end up being a perf issue; LLVM *should* be able
to optimize memmove to memcpy in a lot of cases, and for small copies
the generated code *should* be mostly comparable. (In reality, LLVM
is currently horrible at optimizing memmove, but that's a bug, not a
fundamental issue.)
gcc currently generates wrong code; that's
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32667.
llvm-svn: 51566
Turns out that there are multiple places where a redefinition diagnostic can be emitted. A cleaner solution (without touching Sema) is to have
the Driver turn off these diagnostics. (will submit this patch soon)
llvm-svn: 51502
llvm::Type::isSingleValueType. Currently these two functions have
the same behavior, but soon isFirstClassType will return true for
struct and array types.
Clang may some day want to use of isFirstClassType for some of
these some day as an optimization, but it'll require some
consideration.
llvm-svn: 51446
tracked object to "escape": it's reference count might be incremented by the
called function, thus causing an object's lifetime to extend beyond when the
local reference count is decremented to 0.
This addresses: <rdar://problem/5933215>
llvm-svn: 51433
actually work (instead of crashing llc), and there's enough info emitted
to get line number information in gdb. This should hopefully be helpful
for debugging non-working programs.
I got rid of the begin/endregion calls because the implementation wasn't
working; someone who knows the debugging info a bit better might try to
add it. I really have no clue how a compiler is supposed to emit them.
This commit shouldn't have any effect without -g.
llvm-svn: 51404
lib/CodeGen/CGExpr.cpp and to change include/clang/AST/Attr.h to
use its own enum for visibility types instead of using
llvm::GlobalValue::VisibilityTypes. These changes eliminate
dependencies in the AST library on LLVM's VMCore library.
llvm-svn: 51398
in Sema, per discussion on mailing list. This doesn't cause any changes
in the test results. I'll probably add some more tests at some point,
but it's an open question what we need to accept to be compatible with
real code.
This doesn't touch the existing isConstantExpr method on Expr; that
should be addressed somehow eventually (either removed or refined to
whatever is appropriate).
llvm-svn: 51318
consensus was that it was fine; iso646 is trivial.
We actually don't need that many headers overall... on Ubuntu, at least,
all we need are stdint.h, stdarg.h, limits.h, and float.h to stop
depending on having gcc's headers around for normal code.
On a side note, the path searching needs some improvement; I had to
hardcode the path to my header directory into clang.cpp to get the
driver to use it consistently.
llvm-svn: 51317
as far as I can tell, and it fixes code like test/CodeGen/functions.c.
(Whatever performance effect it might have, crashing on a
construct like this isn't really acceptable; I've run into this
multiple times.)
llvm-svn: 51312
extend beyond the end of the function.
I'm not completely sure this is the right way to fix this bug, so
someone familiar with the parser should double-check.
llvm-svn: 51311
Fixed a bug in ParmVarDecl::param_end(): Handle the case where there are no
ParmVarDecls for a FunctionDecl, but its function prototype has formal arguments
(can happen with typedefs).
llvm-svn: 51297
1) Sema::ParseAST now constructs a TranslationUnit object to own the top-level Decls, which releases the top-level Decls upon exiting ParseAST.
2) Bug fix: TranslationUnit::~TranslationUnit handles the case where a Decl is added more than once as a top-level Decl.
3) Decl::Destroy is now a virtual method, obviating the need for a special dispatch based on DeclKind.
3) FunctionDecl::Destroy now releases its Body using its Destroy method.
4) Added Stmt::Destroy and Stmt::DestroyChildren, which recursively delete the child ASTs of a Stmt and call their dstors. We may need to special case dstor/Destroy methods for particular Stmt subclasses that own other dynamically allocated objects besides AST nodes.
5) REGRESSION: We temporarily are not deallocating attributes; a FIXME is provided.
llvm-svn: 51286
implement bitfield codegen (although I don't envy the person who
implements it). This also prevents a crash on code like that from PR2309
(it's still broken, but it fails more gracefully).
llvm-svn: 51285
they were causing bad code to be emitted. There are two fixes here: one
makes sure we emit a string that is long enough, and one makes sure we
properly handle string initialization in init lists.
llvm-svn: 51259
used for _Bool is not the same as the primitive width (which for _Bool
is 1 bit). The load and store changes add some casts to make the
types consistent. The EmitLValue changes make sure that the pointer is
of an appropriate type for loading the bitfield.
This isn't perfect, but it's an improvement, and getting everything
right depends on actually laying out structs in an ABI-compliant way.
llvm-svn: 51224
the condidtional have compatible types, they are not necessarily the
same type. Therefore, we cast to the composite type. As a hack, for
the moment we assume that the composite type is the type of the
left-hand expression; this isn't correct, but it's good enough for most
purposes.
llvm-svn: 51202
in Sema::CheckFunctionCall:
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080512/005706.html
The bug was that the logic from the helper methods used by CheckFunctionCall
were being inverted (a subtle bug). This would cause the parser to discard
any valid AST nodes involving several builtins (see patch).
This removes the last regression failure I'm seeing in the test suite: Analysis-Apple/NoReturn.
llvm-svn: 51168
expressing the full flexibility of the LLVM shufflevector instruction.
The expected immediate usage is in *mmintrin.h, so that they don't
depend on the mess of gcc-inherited (and not completely implemented)
shuffle builtins.
llvm-svn: 51113
ScopedDecl with the backpatcher, and *not* a local variable on the stack. The
value of DeclCtx gets filled in *later* by the backpatcher.
This fixes: http://llvm.org/bugs/show_bug.cgi?id=2308
llvm-svn: 50976
-identifierResolver exposes an iterator interface to get all decls through the scope chain.
-The semantic staff (checking IdentifierNamespace and Doug's checking for shadowed tags were moved out of IdentifierResolver and back into Sema. IdentifierResolver just gives an iterator for all reachable decls of an identifier.
llvm-svn: 50923
vector of the same element type and half the width, with the high, low, even,
and odd elements respectively.
Allow member references to member references, so that .hi.hi gives you the high
quarter of a vector. This is fairly convenient syntax for some insert/extract
operations.
Remove some unnecessary methods/types in the ExtVectorElementExpr class.
llvm-svn: 50892
function that returns a CFxxxRef, not just functions whose name begins with
CF. This implements <rdar://problem/5917879>.
Added test case for this feature.
Updated calls to CStrInCStrNoCase to swap their arguments, per compatibility
with strcasestr.
llvm-svn: 50829
logic for function calls and message expressions.
Use the following heuristic to infer "allocating" instance methods:
[ClassName classWithXXX] allocates an object
Update testcase to reflect this heuristic.
llvm-svn: 50708
the object was last used. This can be confusing to users.
For example:
// 'y' is leaked
x = foo(y);
instead:
x = foo(y);
// 'y' is leaked
llvm-svn: 50661
highlight the most nested subexpression that appears most responsible (giving
the user better diagnostic feedback).
Updated test cases to illustrate this feature.
Implements: <rdar://problem/5880443>
llvm-svn: 50647
related to pp-expressions. Doing so is pretty simple and this
patch implements it, yielding nice diagnostics like:
t.c:2:7: error: division by zero in preprocessor expression
#if 1 / (0 + 0)
~ ^ ~~~~~~~
t.c:5:14: error: expected ')' in preprocessor expression
#if (412 + 42
~~~~~~~~^
t.c:5:5: error: to match this '('
#if (412 + 42
^
t.c:10:10: warning: left side of operator converted from negative value to unsigned: -42 to 18446744073709551574
#if (-42 + 0U) / -2
~~~ ^ ~~
t.c:10:16: warning: right side of operator converted from negative value to unsigned: -2 to 18446744073709551614
#if (-42 + 0U) / -2
~~~~~~~~~~ ^ ~~
5 diagnostics generated.
llvm-svn: 50638
a) correct rejection of ',' in pp expressions.
b) the precedence of ',' was wrong w.r.t. ?:.
Thanks again to Neil for finding these and providing testcases.
llvm-svn: 50625
NOTE: This work is incomplete and still fails many tests (as a result, it isn't enabled yet). Nevertheless, I wanted to check it in so I can work on it from home.
llvm-svn: 50544
EvalStore/EvalLoad to handle all loads/stores from symbolic memory, allowing us
to do checks for null dereferences, etc., at any arbitrary load/store (these
were missed checks before). This also resulted in some major cleanups, some
conceptual, and others just in the structure of the code.
This temporarily introduces a regression in the test suite (null-deref-ps.c)
before I add a new LVal type for structure fields.
llvm-svn: 50443
std::pair<CFGBlock*, CFGBlock*> that have an 8-byte alignment for use with
ProgramPoint. This fixes a bug reported by Argiris where using std::set<> on
Windows would result in a 4-byte alignment, not an 8-byte alignment.
Fixes: <rdar://problem/5892265>
llvm-svn: 50364
-NamespaceDecl for the AST
-Checks for name clashes between namespaces and tag/normal declarations.
This commit doesn't implement proper name lookup for namespaces.
llvm-svn: 50321
empty CFGBlocks that only contained a terminator.
Added improved diagnostics for break and continue statements and default branches in switch statements.
This fixes <rdar://problem/5889244>.
llvm-svn: 50286
Implemented support for better localized leaks in the CF reference count checker.
Now leaks should be flagged close to where they occur.
This should implement the desired functionality in <rdar://problem/5879592>, although the diagnostics still need to be improved.
llvm-svn: 50241
This fixes <rdar://problem/5886141>.
Part of this change resulted in creating sparser summaries. I also added
some more comments and did some minor code cleanups.
llvm-svn: 50214
Problem:
In the recently refactored VisitDeref (which processes dereferences), we
were incorrectly skipping the node just generated for the subexpression
of the dereference. This was a horrible regression.
llvm-svn: 50176
when it is constructed via deserialization. This is done by recording a flag
indicating that this is the case, and it deletes these objects by getting
the references stored in the ASTContext object. This fixes some memory
leaks that occurs when we deserialize translation units from bitcode files.
The rationale between having TranslationUnit sometimes own these objects and
sometimes not is that a TranslationUnit object can be constructed from
state generated by the parser (Preprocessor; semantic analyzer, etc.), and thus
in these cases won't own the IdentifierTable or Selectors, etc. During
deserialization, there is no Preprocessor, so somebody needs to own these
objects in order for them to be properly reclaimed.
llvm-svn: 50149
DeclContext *CtxDecl -> DeclContext *DeclCtx
DeclContext *CD -> DeclContext *DC
It makes the code more consistent."
Patch by Zhongxing Xu!
llvm-svn: 50105
them to not be stack-allocated.
HTMLDiagnostics now batches PathDiagnostics before emitting HTML in its dtor.
This is a workaround for a problem when we trampled the Preprocessor state
when highlighting macros (sometimes resulting in an assertion failure).
llvm-svn: 50102
Added iterators, methods to find property and categories.
Use them in doing semantic analysis on property implementation
declarations. Fixed typos.
llvm-svn: 50050
1. If we hit a semantic error, try harder to recover to emit
diagnostics for later initializer errors (PR2241).
2. Don't leak parsed initializers on an error.
llvm-svn: 49998
literal, indicate what character the error is about or where
it would be. For example:
pr2241.c:1:17: error: exponent has no digits
float f[] = { 1e , 1ee0 };
^
llvm-svn: 49996
deleting dead code, making other code more obvious, and fixing a
bug. This fixes these (bogus) warnings in release mode:
SemaExprObjC.cpp:383: warning: 'RHSProtoE' may be used uninitialized in this function
SemaExprObjC.cpp:383: warning: 'RHSProtoI' may be used uninitialized in this function
llvm-svn: 49984
can decide the policy on how to cache related bugs. This allows us to
properly to handle warning about multiple leaks in the same location in the
ref count checker (not yet done).
llvm-svn: 49918
from registered BugTypes. This helps with debugging.
Add detection of NULL values in ref count checker; this suppresses false positives.
llvm-svn: 49912
Shuffled around code in CFRefCount to better pair classes with implementation,
and started adding subclasses of RangedBugReport to handle better diagnostics
for reference count bugs.
llvm-svn: 49889
EmitPathWarning into one method. We now properly handle emitting warnings
without a PathDiagnosticClient when the warning does not involve a particular
statement.
llvm-svn: 49884
clang.cpp: InitializePreprocessor now makes a copy of the contents of PredefinesBuffer and
passes it to the preprocessor object.
clang.cpp: DriverPreprocessorFactory now calls "InitializePreprocessor" instead of this being done in main().
html::HighlightMacros() now takes a PreprocessorFactory, allowing it to conjure up a new
Preprocessor to highlight macros.
class HTMLDiagnostics now takes a PreprocessorFactory* that it can use for html::HighlightMacros().
Updated clients of HTMLDiagnostics to use this new interface.
llvm-svn: 49875
Added assertion if unexpected property decls are found where they don't belong.
Consolidated property decl. printing by using a helper function.
llvm-svn: 49862
-Added TranslationUnitDecl class to serve as top declaration context
-ASTContext gets a TUDecl member and a getTranslationUnitDecl() function
-All ScopedDecls get the TUDecl as DeclContext when declared at global scope
llvm-svn: 49855
Refactored LiveVariables to use getTerminatorCondition() in VisitTerminator().
Bug fix: CFG now computes Block-level expression numbers using information
from block terminators. This fixes <rdar://problem/5868189>.
llvm-svn: 49818
with the Lexer. This is cheaper and gives us some advantages. For now we start
highlighting preprocessor directives (which need improvement), and disable
comments. Comments to be restored later.
llvm-svn: 49815
two worklists: for nodes whose locations are block edges with loop terminators
and another for nodes with all other locations. We only dequeue from the loop
worklist when the other is empty. Exploration of the graph is still in
reverse-BFS.
llvm-svn: 49791
allowing us to use a cheaper means to highlight keywords and making
it so that comments won't foul up macro expansions.
Start highlighting macro expansions.
llvm-svn: 49779
problems, including the fact that it doesn't work well with multi-line
comments due to Ted's crazy table. However, that could be fixed, and it
does work with single-line ones :).
llvm-svn: 49778
a SourceLocation to get a RewriteBuffer, poke the RewriteBuffer
with an offset directly. THis is no faster, but results in
cleaner code.
llvm-svn: 49774
This is a fairly mechanical/large change. As a result, I avoided making any changes/simplifications that weren't directly related. I did break two Analysis tests. I also have a couple FIXME's in UninitializedValues.cpp. Ted, can you take a look? If the bug isn't obvious, I am happy to dig in and fix it (since I broke it).
llvm-svn: 49748
LiveVariables analysis now does a flow-insensitive analysis to determine what variables have their address taken; these variables are now always treated as being live.
The DataflowSolver now uses "SetTopValue()" when getting the initial value for the entry/exit block.
llvm-svn: 49734
in a block's terminator. This expression is visited within a block, but it is accessed by the
terminator. This is important to observe because for live-variables analysis the block-level
expression is live between the terminator and where the expression occurs in the block. So far
this hasn't been an issue to not observe this because the block-level expression used in the
terminator is always the last one in the block, and we have never queried the liveness information
about this point (but before the terminator).
llvm-svn: 49709
the files of different SourceLocations. These methods correctly handle the
case where a file may have multiple FileIDs due to it being large enough
to be spread across several chunks.
llvm-svn: 49682
a nice shiny B+ Tree variant. This fixes the last of the known algorithmic
issues with the rewriter, allowing a significant speedup. For example,
-emit-html on Ted's 500K .i file speeds up from 26.8s -> 0.64s in a
debug build (41x!) and 5.475s -> 0.132s (41x!) in an optimized build.
This code is functional but needs to be cleaned up, ifdefs removed, better
commented, and moved to a .cpp file. I plan to do this tomorrow.
llvm-svn: 49635
inheritance in C++. It'll parse the base-specifier list, e.g.,
class D : public B1, virtual public B2 { };
and do some of the simpler semantic checks (B1 and B2 are classes;
they aren't unions or incomplete types, etc).
llvm-svn: 49623
allows the parsing of "class" in addition to "struct" and "union" to
declare a record. So this patch allows:
class C { };
class C c1;
But it does not contain the lookup bits, so this won't work yet:
C c2;
Patch by Doug Gregor!
llvm-svn: 49613
This results in an (IMO) simpler algorithm, results in fewer splits, and
is more amenable to delta handling (there is no reason to mutate the tree
at all when adding a delta to a position that already exists in the tree).
llvm-svn: 49609
(but simple!) datastructures in the rewriter with a more complex but
more efficient one.
This replaces the Deltas vector with a specialized BTree that makes
delta lookups much more efficient. This speeds up -emit-html on a 500K
.i file from 157.154 to 27.127 seconds on my machine (5.8x).
While this code is functional, it isn't very pretty, I have much
refactoring planned for it, and will remove the USE_VECTOR ifdef.
Stay tuned.
llvm-svn: 49586
GRSimpleVals::EvalCal(), and don't change reference counts.
Remove "getDoNothingSummary()", as a NULL summary does the same thing.
Added temporary hack for the "Get" rule for objects that return a pointer type:
treat them as non-owned CF objects.
Added test case to detect the release of a non-owned object.
llvm-svn: 49555
1) objc ivar processing is split out of ActOnField into its own ActOnIvar method.
2) the new objc ivar action takes visibility info directly, eliminating
AllVisibilities in ParseObjCClassInstanceVariables.
llvm-svn: 49506
declarators. This allows the clients (C structs, objc classes, objc
properties, [future] C++ classes) etc, to do custom processing before invoking
an action.
This has two benefits in the short term:
1) objc ivar processing should be split out of ActOnField into its own ActOn method.
2) the new objc ivar action can take visibility info directly, eliminating
AllVisibilities in ParseObjCClassInstanceVariables.
3) objc properties can pass their own special sauce down to sema as well.
llvm-svn: 49468
Bugs are now reported using a combination of "BugType" (previously
BugDescription) and Bug "BugReport" objects, which are fed to BugReporter (which
generates PathDiagnostics). This provides a far more modular way of registering
bug types and plugging in diagnostics.
GRExprEngine now owns its copy of GRCoreEngine, and is not owned by the
ExplodedGraph.
ExplodedGraph is no longer templated on the "checker", but instead on the state
contained in the nodes.
llvm-svn: 49453
in CGObjCGNU.cpp and once in CGObjCEtoile.cpp. Moved its definition to
CGObjC.cpp. This resolves a build error in Xcode, and also cleans things up.
llvm-svn: 49429
While it is similar to the other compatibility predicates in ASTContext,
it is not used by them and is different.
In addition, greatly simplify ObjCQualifiedIdTypesAreCompatible and
fix some canonical type bugs. Also, simplify my Type::getAsObjC* methods.
llvm-svn: 49313
sema. This allows clients of the parser to have the unmolested
list if desired, and guarantees that noone can create an
ObjCQualifiedInterfaceType with an unsorted list.
llvm-svn: 49310
#1: To be consistent with FieldDecl::getContextDecl(), which serves the same purpose.
#2: From my perspective, getContext() is too general (and used by several other classes for different purposes).
llvm-svn: 49224
-Added ContextDecl (no TranslationUnitDecl)
-ScopedDecl class has a ContextDecl member
-FieldDecl class has a ContextDecl member, so that a Field or a ObjCIvar can be traced back to their RecordDecl/ObjCInterfaceDecl easily
-FunctionDecl, ObjCMethodDecl, TagDecl, ObjCInterfaceDecl inherit from ContextDecl. With TagDecl as ContextDecl, enum constants have a EnumDecl as their context.
-Moved Decl class to a "DeclBase.h" along with ContextDecl class
-CurContext is handled by Sema
llvm-svn: 49208
"BugReporter" and "BugDescription". BugDescription is used to describe
a bug and provide pieces of the PathDiagnostic, and BugReporter creates
the actual PathDiagnostic by crawling through the ExplodedGraph.
Migrated checks done by GRSimpleVals to be reported using the new BugReporter
mechanism.
llvm-svn: 49127
I also finished unifying access to scope decl change by converting Sema::getObjCInterfaceDecl() to use Sema::LookupDecl(). This is much cleaner now:-)
llvm-svn: 49107
(1) Remove IdLoc (it's never used).
(2) Add a bool to enable/disable lazy builtin creaation (defaults to true).
This enables us to use LookupDecl() in Sema::isTypeName(), which is also part of this commit.
To make this work, I changed isTypeName() to be a non-const member function. I'm not happy with this, however I fiddled with making LookupDecl() and friends const and it got ugly pretty quickly. We can certainly add it back if/when someone has time to fiddle with it. For now, I thought this simplification was more important than retaining the const-ness.
llvm-svn: 49087
object type". Add a predicate that checks exactly this, as it is equivalent
to checking ot see if the type is *not* a function type, which is faster
to check.
llvm-svn: 49082
object or incomplete types shall not be restrict-qualified."
2) Warn about qualifiers on function types: C99 6.7.3p8: "If the
specification of a function type includes any type qualifiers, the
behavior is undefined."
3) Implement restrict on C++ references.
4) fix some locations for various C++ reference diagnostics.
llvm-svn: 49081
First, we got several CVR propagation cases wrong, which Eli pointed
out in PR2039.
Second, we didn't propagate address space qualifiers correctly, leading
to incorrect lowering of code in CodeGen/address-space.c.
Third, we didn't uniformly propagate the specifier in the array to the
pointer ("int[restrict 4]" -> "int *restrict").
This adds an ASTContext::getArrayDecayedType member that handles the
non-trivial logic for this seemingly simple operation.
llvm-svn: 49078
- Added a DenseMap to associate an IdentifierInfo with the ObjCCompatibleAliasDecl.
- Renamed LookupScopedDecl->LookupDecl and changed it's return type to Decl. Also added lookup for ObjCCompatibleAliasDecl's.
- Removed Sema::LookupInterfaceDecl(). Converted clients to used LookupDecl().
- Some minor indentation changes.
Will deal with ObjCInterfaceDecl and getObjCInterfaceDecl() in a separate commit...
llvm-svn: 49058
This is a temporary solution to avoid running out of file descriptors (which defaults to 256).
Need to benchmark to understand the speed benefit. If the benefit is small, the simple solution is to avoid memory mapping files. If the benefit is significant, more thought is necessary.
llvm-svn: 48991
GRSimpleVals.h
Added a PathDiagnosticClient option to the driver functions for the
CFRefCountChecker and the GRSimpleVals analysis. Both analyses now accept a "-o"
argument from the driver that specifies where HTML reports should be dumped.
llvm-svn: 48989
Fix objc ivar lookup. Ivar lookup should occur between lookup
of method-local values and lookup of globals. Emulate this with
some logic in the handling of Sema::ActOnIdentifierExpr.
Two todo's left:
1) sema shouldn't turn a bare reference to an ivar into "self->ivar"
in the AST. This is a hack.
2) The new ScopedDecl::isDefinedOutsideFunctionOrMethod method does
not correctly handle typedefs and enum constants yet.
llvm-svn: 48972
remaining open issues I've communicated to him:
1) self can be assigned to, and his patch didn't handle it correctly.
2) CollectObjCIvarTypes is N^2 (because each subclass reprocesses
all parent class ivars) and flattens classes. If A derives from B,
and both have an int, I'd expect to get { {i32}, i32}, not { i32, i32}.
David, please review.
llvm-svn: 48970
Added "AnnotatedPath" class to record an annotated path that will be useful for inspecting paths.
Added some boilerplate code for simple checks of Apple's Foundation API.
llvm-svn: 48867
The effect is that if a variable is uninitialized along a branch (but initialized along another), at merge points it is considered uninitialized. Previously we had the opposite behavior. The new behavior is more conservative, and more in line with gcc's behavior.
llvm-svn: 48689
and clients can achieve a cleaner design just by inserting tags directly. Reserve
the "html" namespace for meta-level operations (e.g., escaping text, etc.)
llvm-svn: 48524
indicate whether or not the new tag should be the outermost tag at the specified
location (in the case that other tags have been inserted at the same spot).
llvm-svn: 48506
specified. Previously, the ctor would allocate memory for the list and then
it would get filled in later. Move the allocation+filling in to be more
consistent with other stuff, e.g. the addMethods method.
llvm-svn: 48427
lib dir and move all the libraries into it. This follows the main
llvm tree, and allows the libraries to be built in parallel. The
top level now enforces that all the libs are built before Driver,
but we don't care what order the libs are built in. This speeds
up parallel builds, particularly incremental ones.
llvm-svn: 48402
Ted Kremenek