www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 3700] New: Ref foreach over large static arrays fails

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3700

           Summary: Ref foreach over large static arrays fails
           Product: D
           Version: 2.038
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: major
          Priority: P2
         Component: DMD
        AssignedTo: nobody puremagic.com
        ReportedBy: bugzilla kyllingen.net



00:21:30 PST ---
This code correctly compiles, but causes a segmentation fault when run:

    void main()
    {
        double[1_000_000] a;
        foreach (ref e; a) e = 1.0;
    }


In case it helps, here's a disassembly of the main function. A run through the
debugger shows that the offending instruction is located at 0x080490cd.

080490c4 <_Dmain>:
 80490c4:       55                      push   %ebp
 80490c5:       8b ec                   mov    %esp,%ebp
 80490c7:       81 ec 1c 24 f4 00       sub    $0xf4241c,%esp
 80490cd:       53                      push   %ebx
 80490ce:       57                      push   %edi
 80490cf:       b9 80 84 1e 00          mov    $0x1e8480,%ecx
 80490d4:       31 c0                   xor    %eax,%eax
 80490d6:       8d bd e4 db 0b ff       lea    -0xf4241c(%ebp),%edi
 80490dc:       f3 ab                   rep stos %eax,%es:(%edi)
 80490de:       8d 8d e4 db 0b ff       lea    -0xf4241c(%ebp),%ecx
 80490e4:       bb 40 42 0f 00          mov    $0xf4240,%ebx
 80490e9:       89 9d e4 ed 85 ff       mov    %ebx,-0x7a121c(%ebp)
 80490ef:       89 8d e8 ed 85 ff       mov    %ecx,-0x7a1218(%ebp)
 80490f5:       89 85 ec ed 85 ff       mov    %eax,-0x7a1214(%ebp)
 80490fb:       8b 95 ec ed 85 ff       mov    -0x7a1214(%ebp),%edx
 8049101:       3b 95 e4 ed 85 ff       cmp    -0x7a121c(%ebp),%edx
 8049107:       73 42                   jae    804914b <_Dmain+0x87>
 8049109:       8b 85 ec ed 85 ff       mov    -0x7a1214(%ebp),%eax
 804910f:       3b 85 e4 ed 85 ff       cmp    -0x7a121c(%ebp),%eax
 8049115:       72 0a                   jb     8049121 <_Dmain+0x5d>
 8049117:       b8 04 00 00 00          mov    $0x4,%eax
 804911c:       e8 33 00 00 00          call   8049154 <_D1d7__arrayZ>
 8049121:       8b 8d ec ed 85 ff       mov    -0x7a1214(%ebp),%ecx
 8049127:       8b 95 e8 ed 85 ff       mov    -0x7a1218(%ebp),%edx
 804912d:       8b 85 e4 ed 85 ff       mov    -0x7a121c(%ebp),%eax
 8049133:       8d 1c ca                lea    (%edx,%ecx,8),%ebx
 8049136:       c7 03 00 00 00 00       movl   $0x0,(%ebx)
 804913c:       c7 43 04 00 00 f0 3f    movl   $0x3ff00000,0x4(%ebx)
 8049143:       ff 85 ec ed 85 ff       incl   -0x7a1214(%ebp)
 8049149:       eb b0                   jmp    80490fb <_Dmain+0x37>
 804914b:       31 c0                   xor    %eax,%eax
 804914d:       5f                      pop    %edi
 804914e:       5b                      pop    %ebx
 804914f:       c9                      leave  
 8049150:       c3                      ret    
 8049151:       90                      nop
 8049152:       90                      nop
 8049153:       90                      nop

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 13 2010
next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3700


Don <clugdbug yahoo.com.au> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |clugdbug yahoo.com.au
         Resolution|                            |INVALID



Nothing to do with foreach. It's just a stack overflow. Reduced test case:

void main()
{
   double[1000000] a;
}

Error: Stack Overflow

You can't expect 8Mb of data to be storable on the stack!
This will happen in C, as well.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 14 2010
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3700




06:56:27 PST ---
You may be right. But consider the following:

* The "Arrays" page on the D homepage says that "The total size of a static
array cannot exceed 16Mb." I am only allocating half of that.

* When I've run my example, it has *always* caused a segmentation fault.

* The following, on the other hand, has so far *never* caused a segmentation
fault on my machine:

void main()
{
  double[1_000_000] a;
  for (i; 0 .. a.length)  a[i] = 1.0;
}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 14 2010
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3700


Leandro Lucarella <llucax gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |llucax gmail.com



PST ---
In Linux the default stack size is 8MiB, I don't think D runtime changes that.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 14 2010
prev sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=3700





 You may be right. But consider the following:
 
 * The "Arrays" page on the D homepage says that "The total size of a static
 array cannot exceed 16Mb." I am only allocating half of that.
That limit is for static arrays (ie, global arrays, or arrays marked 'static'), not fixed-length arrays. (I wish the spec would stop talking about 'static arrays' when it means 'fixed-length'). The size of the stack is always limited by the OS, sometimes severely so. -- Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email ------- You are receiving this mail because: -------
Jan 14 2010