> Date: Sun, 1 Apr 2012 13:54:10 +0300
> From: Paul Irofti <[email protected]>
>
> [--snip-some-of-the-gyping-from-my-initial-diff--]
Not really gyped. Already had some of it in my tree from when I
looked at the arithnmetic diff. It was just that your casting
mistakes made me realize what the right way to go was ;).
> > case AML_OBJTYPE_STRING:
> > - if (ival == -1)
> > + if (ival == (uint64_t)-1)
>
> This is what I tried first as well. But this cast fails on my x61s where
> ival is set to a 32-bit -1.
Well, the real bug is that ival gets set to "32-bit -1" (by which you
mean 0xffffffff I presume). That should never happen. May very well
be a bug that you introduced in your diff because:
> Index: dev/acpi/dsdt.c
> ===================================================================
> RCS file: /cvs/src/sys/dev/acpi/dsdt.c,v
> retrieving revision 1.193
> diff -u -p -r1.193 dsdt.c
> --- dev/acpi/dsdt.c 15 Mar 2012 18:36:53 -0000 1.193
> +++ dev/acpi/dsdt.c 30 Mar 2012 22:45:15 -0000
> @@ -1451,11 +1448,11 @@ static char osstring[] = "Macrosift Wind
> struct aml_defval {
> const char *name;
> int type;
> - int64_t ival;
> + u_int64_t ival;
> const void *bval;
> struct aml_value **gval;
> } aml_defobj[] = {
> - { "_OS_", AML_OBJTYPE_STRING, -1, osstring },
> + { "_OS_", AML_OBJTYPE_STRING, (unsigned)-1, osstring },
> { "_REV", AML_OBJTYPE_INTEGER, 2, NULL },
> { "_GL", AML_OBJTYPE_MUTEX, 1, NULL, &aml_global_lock },
> { "_OSI", AML_OBJTYPE_METHOD, 1, aml_callosi },
...the (unsigned) cast you added there will give you one!
