On Thu, Dec 20, 2001 at 02:20:45PM +0000, Simon Cozens wrote:
> I'm sure this shouldn't be. Anyone got any clues?
Me! The following seems to be an off-by-one error:
diff -d -u -r1.1 jit.c
--- jit.c 20 Dec 2001 01:57:01 -0000 1.1
+++ jit.c 20 Dec 2001 14:24:32 -0000
@@ -70,7 +70,7 @@
k += op_assembly[*pc].nargop;
prev_address = op_address[k] = prev_address + op_assembly[*pc].size;
size += op_assembly[*pc].size;
- pc += op_assembly[*pc].nargop;
+ pc += op_assembly[*pc].nargop + 1;
}
bytecode_position = 0;
@@ -323,7 +323,7 @@
/* */
bytecode_position += op_assembly[*pc].nargop;
- pc += op_assembly[*pc].nargop;
+ pc += op_assembly[*pc].nargop + 1;
}
return arena_start;
}
However, this causes mops.pasm to take forever.
--
As in certain cults it is possible to kill a process if you know its true name.
-- Ken Thompson and Dennis M. Ritchie
