[LLVMbugs] [Bug 10810] New: Function prototype with K&R definition and resulting call.

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Mon Aug 29 21:10:31 PDT 2011


           Summary: Function prototype with K&R definition and resulting
           Product: clang
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P
         Component: -New Bugs
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: jpbonn-keyword-llvmbug.a51747 at corniceresearch.com
                CC: llvmbugs at cs.uiuc.edu

I believe this to be a bug.  There is a function prototype in scope but the
function definition is K&R style.  My understanding is the actual call to the
function should be non-varargs since the prototype is present.  What I'm seeing
is the call is being made as a varargs call.  I believe either an warning/error
should be generated or the generated code should be non-varargs.

This code was derived from a test in the gcc test suite. This happens when
using clang with the -O0 option. 

If the prototype and definition of the function is:
static void attr_eq(char *);
static void
attr_eq (name)
    char * name;

The call generated is:
call void (...)* bitcast (void (i8*)* @attr_eq to void (...)*)(i8*
getelementptr inbounds ([5 x i8]* @arg0, i32 0, i32 0))

Here's the complete code:
extern void abort (void);

static char arg0[] = "arg0";

static void attr_eq        (char *);

void attr_rtx        (char *);
char *attr_string        (char *);

static void
attr_eq (name)
    char * name;
attr_eq (char * name)
  attr_rtx (attr_string (name));


int main()
  attr_eq (arg0);
  exit (0);

Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

More information about the llvm-bugs mailing list