out of the llvm namespace.  This makes the clang namespace be a sibling of
llvm instead of being a child.

The good thing about this is that it makes many things unambiguous.  The
bad things is that many things in the llvm namespace (notably data structures
like smallvector) now require an llvm:: qualifier.  IMO, libsystem and libsupport
should be split out of llvm into their own namespace in the future, which will fix
this issue.

llvm-svn: 39659

Submitted by:
Reviewed by:
Fix CheckRelationalOperands/CheckEqualityOperands to deal with null pointer
constants. The new logic also deals (more) correctly for non-pointer/integer
operands.

llvm-svn: 39654

int test(int X, short Y, float Z) {
  return (int)(X*Y+Z);
}

to:

define i32 @test(i32 %X, i16 %Y, float %Z) {
entry:
        %promote = sext i16 %Y to i32           ; <i32> [#uses=1]
        %mul = mul i32 %promote, %X             ; <i32> [#uses=1]
        %promote3 = sitofp i32 %mul to float            ; <float> [#uses=1]
        %add = add float %promote3, %Z          ; <float> [#uses=1]
        %conv = fptosi float %add to i32                ; <i32> [#uses=1]
        ret i32 %conv
}

with:

$ clang -emit-llvm t.c | llvm-as | opt -std-compile-opts | llvm-dis

llvm-svn: 39652

Submitted by:
Reviewed by:
Fixed typechecking bugs wrt UsualUnaryConversions. Includes two distinct fixes:

#1: Call UsualUnaryConversions in CheckRelationalOperands/CheckEqualityOperands.
#2: UsualArithmeticConversions arguments are now output parameters. This insures
the implicit conversion is seen by clients (and fixes bugs in CheckAdditionOperands
and CheckSubtractionOperands when doing pointer arithmetic).
~

llvm-svn: 39649

llvm-svn: 39644

Submitted by:
Reviewed by:
Implement semantic analysis for vector_size attribute!
We now produce the following diagnostics...

[administrators-powerbook59:~/llvm/tools/clang] admin% ../../Debug/bin/clang vector.c
vector.c:2:29: error: attribute requires precisely 1 argument
typedef int __attribute__(( vector_size )) tVecp;
                            ^
vector.c:7:32: error: attribute requires precisely 1 argument
extern int foo __attribute__(( vector_size ));
                               ^
vector.c:8:34: error: attribute requires precisely 1 argument
extern float bar __attribute__(( vector_size(16,18) ));
                                 ^
vector.c:11:34: error: vector_size requires integer constant (attribute ignored)
extern char foo2 __attribute__(( vector_size(16.2) ));
                                 ^           ~~~~
vector.c:21:47: error: invalid vector type 'struct s'
struct s { int a; } structVar __attribute__(( vector_size(16) ));

llvm-svn: 39643

static int seminal(from, to)
{
}

llvm-svn: 39642

llvm-svn: 39637

llvm-svn: 39634

Submitted by:
Reviewed by:
Lot's of attribute scaffolding.
Modernized ParseArraySubscriptExpr...call DefaultFunctionArrayConversion (which
simplified the logic considerably) and upgrade Diags to use the range support.

llvm-svn: 39628

like: int X, Y, Z;

This is required for the code gen to get to all of the declarations in a
DeclStmt, and should simplify some other code.

llvm-svn: 39623

llvm-svn: 39620

llvm-svn: 39619

Submitted by:
Reviewed by:

The following code illustrates a bug in the semantic analysis for assignments:

int func() {
  int *P;
  char *x;
  P = x; // type of this assignment expression should be "int *", NOT "char *".
}

While the type checking/diagnostics are correct, the type of the assignment
expression is incorrect (which shows up during code gen).  With the fix,
the llvm code looks correct...

[dylan:~/llvm/tools/clang] admin% ../../Debug/bin/clang cast.c -emit-llvm
cast.c:4:5: warning: incompatible pointer types assigning 'char *' to 'int *'
  P = x; // type of assignment expression is "int *", NOT "char *".
  ~ ^ ~
; ModuleID = 'foo'

define i32 @func() {
entry:
        %P = alloca i32*                ; <i32**> [#uses=1]
        %x = alloca i8*         ; <i8**> [#uses=1]
        %allocapt = bitcast i32 undef to i32            ; <i32> [#uses=0]
        %tmp = load i8** %x             ; <i8*> [#uses=1]
        %conv = bitcast i8* %tmp to i32*                ; <i32*> [#uses=1]
        store i32* %conv, i32** %P
        ret i32 undef
}

Even though the fix was simple, I decided to rename/refactor the surrounding code
to make a clearer distinction between constraint checking and conversion.

- Renamed AssignmentConversionResult -> AssignmentCheckResult.
- Renamed UsualAssignmentConversions -> CheckAssignmentConstraints.
- Changed the return type of CheckAssignmentConstraints and CheckPointerTypesForAssignment
from QualType -> AssignmentCheckResult. These routines no longer take a reference to the result (obviously).
- Changed CheckAssignmentOperands to return the correct type (with spec annotations).

llvm-svn: 39601

  int *P2;
  P2(1, 2, 3);

  register short X;
  X();

emit:

ds.c:10:3: error: called object is not a function or function pointer
  P2(1, 2, 3);
  ^~~~~~~~~~~
ds.c:13:3: error: called object is not a function or function pointer
  X();
  ^~~

instead of aborting.

llvm-svn: 39599

int func() {
 int (*FP)();
 FP();
 (*****FP)();
}

llvm-svn: 39598

llvm-svn: 39597

llvm-svn: 39591

Submitted by:
Reviewed by:
Touch up a couple comments (one was incorrect/out-of-date).

llvm-svn: 39589

check canonical types in a few places.  Tighten up VerifyConstantArrayType
to diagnose more errors, now that we can evaluate i-c-e's.  Add some fixmes
about poor diagnostics.

We now correctly typecheck this example:

void s(void) {
  typedef int a[(int) +1.0];
  static a b;    // invalid, static VLA

  static int c[(int) +1.0];  // invalid, static VLA
}

void t(void) {
  typedef int a[(int)1.0];
  static a b;    // valid, constant size
}

void u() {
  static int X[-1];
  static int Y[0];
}

producing:

static-vla.c:3:12: error: variable length array declared outside of any function
  static a b;    // invalid, static VLA
           ^
static-vla.c:5:14: error: variable length array declared outside of any function
  static int c[(int) +1.0];  // invalid, static VLA
             ^ ~~~~~~~~~~
static-vla.c:15:14: error: array size is negative
  static int X[-1];
             ^ ~~
static-vla.c:16:14: warning: zero size arrays are an extension
  static int Y[0];
             ^ ~

llvm-svn: 39587

llvm-svn: 39583

Submitted by:
Reviewed by:
Fixed an "impurity" in the typechecking code (to more closely model the
C99 spec)...

- Added function Sema::DefaultFunctionArrayConversion(). This is called
by UsualUnaryConversions(). It is also called from several contexts that
don't call for integer promotions (logical negation for expressions, and
while/if/do/for conditions in statements). It is also used in
UsualAssignmentConversions (which is cleaner than the check I was using
before).

- Changed the name from UsualUnaryConversion->UsualUnaryConversions.

- Added comments to the header.

Since several contexts don't call for integer
promotions,

llvm-svn: 39581

Submitted by:
Reviewed by:
Apply UsualUnaryConversion() to statement conditions that expect scalars.
UsualUnaryConversion() converts function/arrays to pointers.

This fixes the following...

int func() {
  int A[10];
  while (A) {
  }
  if (A) ;
  for (; A; ) ;
}

llvm-svn: 39580

llvm-svn: 39575

ds.c:11:16: error: case label does not reduce to an integer constant
  case 0 ? 0 : foo():
       ~~~~~~~~^~~~~

llvm-svn: 39574

Reviewed by: Chris Lattner

- Disallow references to references, pointers to references, or arrays of
  references.

llvm-svn: 39569

Reviewed by: Chris Lattner

- If the LHS and/or RHS is a reference, then see if they're compatible.
  If so, the type is that of the LHS.

llvm-svn: 39568

typedef information associated with the input.

llvm-svn: 39562

llvm-svn: 39558

- Added C99 reference to why "auto" and "register" cannot be used as a
  storage class specifier for file scoped variable declarations.

llvm-svn: 39557

of one by reference, making it optional.

llvm-svn: 39552

Submitted by:
Reviewed by:
After speaking with Chris, decided not to have GCC "attributes" inherit
from Decl. This will enable us to grow the attribute hierarchy over time
without effecting Decls.

llvm-svn: 39543

Submitted by:
Reviewed by:
Implement support for GCC __attribute__.

- Implement "TODO" in Parser::ParseAttributes. Changed the return type from
void to Parser::DeclTy. Changed all call sites to accept the return value.
- Added Action::ParseAttribute and Sema::ParseAttribute to return an
appropriate AST node. Added new node AttributeDecl to Decl.h.

Still to do...hook up to the Decl...

llvm-svn: 39539

llvm-svn: 39536

Submitted by:
Reviewed by:
- ParseForStatement(): Put back a test/assignment. My removal of
ParseExprStmt() was a bit over zealous:-(thanks to Chris for pointing it out)
- Add assert to VisitDeclStmt().
- Removed an out-of-date FIXME
- Added some curlies for a couple multi-line calls to Diag().

llvm-svn: 39528

Comment format changes.

llvm-svn: 39527

Submitted by:
Reviewed by:
- Many tweaks to various diagnostics.
- Fixed some location/range issues.
- Bug fix to Sema::ParseDeclStmt() - error return code is "true", not 0.

llvm-svn: 39526

Submitted by:
Reviewed by:
Unified the diagnostics for function calls. Since we have range support,
there is no need for including the argument number. Instead, I've made
the diags more expressive by including more type info.

Also improved the indentation of many calls to Diag (which can be 2-3
lines now).

llvm-svn: 39523

Submitted by:
Reviewed by:
Implement some FIXME's that stand in the way of fully typechecking "for"
statements. This involved:

- Adding a DeclStmt AST node (with statement visitor). The DeclStmt
printer is preliminary.
- Added a ParseDeclStmt action, called from Parser::ParseForStatement()
and Parser::ParseStatementOrDeclaration(). DID NOT add to
Parser::ParseIdentifierStatement()...probably could have, however didn't
really understand the context of this rule (will speak with Chris).
- Removed ParseExprStmt (and it's clients)...it was vestigial.

llvm-svn: 39518

Submitted by:
Reviewed by:
- Finished Sema::ParseReturnStmt(). Still need to tweak ranges.
- Tweaked location for function arguments (they now point at the expression directly, no parens or commas).
- Added InvalidOperands helper...was sick of looking at the same 3 lines in ~9 Check functions.
- Added a few diags and moved a group of statement diags to the proper comment/category.

llvm-svn: 39517