floating point exception leads to access violation

Original Reporter info from Mantis: BeniBela @benibela

• Reporter name: Benito van der Zander

Description:

After several floating point exceptions, I get an access violation

Steps to reproduce:

This works without error:

var s, sr: single;
  i, j, tempcode: Integer;
begin
  Val('8.077936E-28', sr, tempcode);

  for i := 1 to 5000 do begin
    s := 0;
    try
      for  j := 0 to Random(5) do s += Random(2) * power(2, Random(256) - 127);
    except
      on e: EMathError do s := 0; //continue;
    end;
  end;

  Val('8.077936E-28', sr, tempcode);
end.

This prints EOverflow: Floating point overflow

var s, sr: single;
  i, j, tempcode: Integer;
begin
  for i := 1 to 5000 do begin
    s := 0;
    try
      for  j := 0 to Random(5) do s += Random(2) * power(2, Random(256) - 127);
    except
      on e: EMathError do s := 0; //continue;
    end;
    Val('8.077936E-28', sr, tempcode);
  end;
end.                          

This prints EAccessViolation: Access violation

procedure test(const s: single);
var
  tempcode: integer;
  sr: single;
begin
  Val('8.077936E-28', sr, tempcode);
  if sr <> s then
end;

var s, sr: single;
  i, j, tempcode: Integer;
begin
  for i := 1 to 5000 do begin
    s := 0;
    try
      for  j := 0 to Random(5) do s += Random(2) * power(2, Random(256) - 127);
    except
      on e: EMathError do s := 0; //continue;
    end;
    test(s)
  end;
end.                             

Additional information:

Perhaps it is related to #37468 (closed) ?

Mantis conversion info:

• Mantis ID: 37926
• OS: linux
• OS Build: opensuse
• Build: r47006
• Platform: amd64
• Version: 3.3.1
• Target version: 3.2.2