Nate,
Can you try out the patch at http://www.freebsd.org/~jhb/patches/mtrr.patch?
It's similar to the patch in the PR with a slight cleanup. The patch is
against current, but hopefully it applies ok to stable. (It should I think).
The patch is also included below for other people to look at in case they have
gripes. The basic issue is that the MTRR's on Nate's box don't work in a way
that XFree86 wants them to, so XFree86 causes his kernel to panic. :( The
patch adds a new loader tunable 'machdep.disable_mtrrs' that can be set to a
non-zero value to disable probing and use of MTRR's.
I realize that being able to automatically disable MTRR's for known bad AMD
chips would be nice, but until we know what chips are bad, this is at least a
functional workaround.
Index: i686_mem.c
===================================================================
RCS file: /usr/cvs/src/sys/i386/i386/i686_mem.c,v
retrieving revision 1.13
diff -u -r1.13 i686_mem.c
--- i686_mem.c 2001/04/27 19:28:19 1.13
+++ i686_mem.c 2001/08/02 21:38:08
@@ -60,6 +60,9 @@
#define mrcopyflags(curr, new) (((curr) & ~MDF_ATTRMASK) | ((new) &
MDF_ATTRMASK))
+static int mtrrs_disabled;
+TUNABLE_INT("machdep.disable_mtrrs", &mtrrs_disabled);
+
static void i686_mrinit(struct mem_range_softc *sc);
static int i686_mrset(struct mem_range_softc *sc,
struct mem_range_desc *mrd,
@@ -569,7 +572,7 @@
i686_mem_drvinit(void *unused)
{
/* Try for i686 MTRRs */
- if ((cpu_feature & CPUID_MTRR) &&
+ if (!mtrrs_disabled && (cpu_feature & CPUID_MTRR) &&
((cpu_id & 0xf00) == 0x600) &&
((strcmp(cpu_vendor, "GenuineIntel") == 0) ||
(strcmp(cpu_vendor, "AuthenticAMD") == 0))) {
--
John Baldwin <[EMAIL PROTECTED]> -- http://www.FreeBSD.org/~jhb/
PGP Key: http://www.baldwin.cx/~john/pgpkey.asc
"Power Users Use the Power to Serve!" - http://www.FreeBSD.org/
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
