Fix TestFormatters.py stepping too far

[rovka]

Summary

TestFormatters.py has a sequence of three 'next' commands to get past
all the initializations in the test function. On AArch64 (and
potentially other platforms), this was one 'next' too many and we ended
up outside our frame.

This patch replaces the sequence with a 'thread until ' the line of the
return from the function, so we should stop after all the
initializations but before actually returning.

[labath]

Looks reasonable, but I'd suggest using something more unique for the breakpoint placement. Feel free to add some comment to the .cpp file and then search for that in python (that's what most of our other tests tend to do).

[rovka]

Addressed @labath's comment (test run in progress). Thanks for having a look!

lldb/packages/Python/lldbsuite/test/commands/expression/formatters/TestFormatters.py

def test(self):
 
         error = lldb.SBError()
         self.assertTrue(
             a_data.GetUnsignedInt32(
                 error,
                 0) == 122,
             '*a_ptr = 122')
 
-        self.runCmd("n")
-        self.runCmd("n")
-        self.runCmd("n")
+        ret = line_number("main.cpp", "Done initializing")
+        self.runCmd("thread until " + str(ret))
 
         self.expect("frame variable numbers",
                     substrs=['1', '2', '3', '4', '5'])
 
         self.expect("expression numbers",
                     substrs=['1', '2', '3', '4', '5'])
 
         frozen = frame.EvaluateExpression("&numbers")

lldb/packages/Python/lldbsuite/test/commands/expression/formatters/main.cpp

int main(int argc, char** argv)
   foo2.a = 7777; // Stop here
   *(foo2.b.i_ptr) = 8888;
     foo2.b.b.h = 9999;
   
   *(foo1.a_ptr) = 9999;
   foo1.b.i = 9999;
   
   int numbers[5] = {1,2,3,4,5};
-  
-  return 0;
-  
+  return 0; // Done initializing
 }