www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 3139] New: compiler dies "Error: out of memory" with case range

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3139

           Summary: compiler dies "Error: out of memory" with case range
           Product: D
           Version: 2.031
          Platform: x86
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: nobody puremagic.com
        ReportedBy: echochamber gmail.com


% cat -n foo.d
     1  import std.conv;
     2  import std.stdio;
     3
     4  void main(string[] args)
     5  {
     6      int i = to!int(args[0]);
     7
     8      switch (i) {
     9      case -9: .. case -1:        // line 9
    10          writefln("negative");
    11          break;
    12      case 0:
    13          writefln("zero");
    14          break;
    15      default:
    16          writefln("positive");
    17          break;
    18      }
    19  }
% dmd foo.d
Error: out of memory
%

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jul 06 2009
next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3139


Don <clugdbug yahoo.com.au> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch, wrong-code


--- Comment #1 from Don <clugdbug yahoo.com.au> 2010-05-12 12:38:56 PDT ---
It goes into an infinite loop because the loop termination condition in
CaseRangeStatement::semantic() uses <=. 
This fails whenever the last case is -1 or ulong.max.
x <= -1u is true for all uint x. Worse, if the range goes from negative to
positive, wrong code is generated.

In my patch, I've also added an error message to detect wrongly-ordered ranges.
eg, case 3:..case 2:
currently produces "more than 256 cases in case range" which is
a bit silly. This also fixes a related accepts-invalid test case.
--
TEST CASES FOR TEST SUITE
--
void hang3139(int x)
{
   switch(x) {
        case -9: .. case -1:
        default:
   }
}

int wrongcode3139(int x)
{
   switch(x) {
        case -9: .. case 2: return 3;
        default:
        return 4;
   }   
}
static assert(wrongcode3139(-5)==3);

// bug 3139, accepts-invalid in DMD2.045.
static assert(!is(typeof(
        (long x) { switch(x) { case long.max: .. case -long.max:
        default:} return 4; }(3)
   )));
---

PATCH
Index: statement.c
===================================================================
--- statement.c    (revision 484)
+++ statement.c    (working copy)
   -3033,6 +3033,14   
     last = last->optimize(WANTvalue | WANTinterpret);
     dinteger_t lval = last->toInteger();

+    if ( (first->type->isunsigned()  &&  fval > lval) || 
+        (!first->type->isunsigned()  &&  (sinteger_t)fval > (sinteger_t)lval))
+    {
+        error ("first case %s must be less than last case %s",
+            first->toChars(), last->toChars());
+        lval = fval;
+    }
+    
     if (lval - fval > 256)
     {   error("more than 256 cases in case range");
         lval = fval + 256;
   -3049,7 +3057,7   
      */

     Statements *statements = new Statements();
-    for (dinteger_t i = fval; i <= lval; i++)
+    for (dinteger_t i = fval; i != lval + 1; i++)
     {
         Statement *s = statement;
         if (i != lval)

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
May 12 2010
prev sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3139


Walter Bright <bugzilla digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |bugzilla digitalmars.com
         Resolution|                            |FIXED


--- Comment #2 from Walter Bright <bugzilla digitalmars.com> 2010-05-16
11:11:02 PDT ---
changelog 491

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
May 16 2010