Re: Kernel Oops with dual core athlon 64

["Rafael J. Wysocki" <rjw@sisk.pl>]

Hi,

On Thursday, 7 of July 2005 07:58, Jon Schindler wrote:
> The dmesg is below.  After I get this Oops, I am unable to use my (PS/2) 
> keyboard, and had to ssh to my machine in order to save a copy of dmesg 
> before rebooting the machine.  I've seen a couple of other users of dual 
> core machings having this problem.  The suggestion so far has been to remove 
> the binary nvidia driver and repoducde the bug.  So, I went ahead and 
> removed the nvidia driver and used the deprecated nv driver that comes with 
> X11 and I still have this issue.  Does anyone have any ideas what might be 
> causing this?  Thanks in advance for the help.  I don't know my way around 
> the kernel, but I do have experience with C and should be able to apply any 
> SMP patches if you want me to test it

It seems to be a cpufreq issue.  You can try to apply the attached patch from
Mark Langsdorf.

Greets,
Rafael


-- 
- Would you tell me, please, which way I ought to go from here?
- That depends a good deal on where you want to get to.
		-- Lewis Carroll "Alice's Adventures in Wonderland"

--- linux-2.6.12-rc6/arch/i386/kernel/cpu/cpufreq/powernow-k8.c.old	2005-06-12 17:41:55.123651184 -0500
+++ linux-2.6.12-rc6/arch/i386/kernel/cpu/cpufreq/powernow-k8.c	2005-06-12 17:46:32.780440936 -0500
@@ -44,7 +44,7 @@
 
 #define PFX "powernow-k8: "
 #define BFX PFX "BIOS error: "
-#define VERSION "version 1.40.2"
+#define VERSION "version 1.40.4"
 #include "powernow-k8.h"
 
 /* serialize freq changes  */
@@ -978,7 +978,7 @@
 {
 	struct powernow_k8_data *data;
 	cpumask_t oldmask = CPU_MASK_ALL;
-	int rc;
+	int rc, i;
 
 	if (!check_supported_cpu(pol->cpu))
 		return -ENODEV;
@@ -1064,7 +1064,9 @@
 	printk("cpu_init done, current fid 0x%x, vid 0x%x\n",
 	       data->currfid, data->currvid);
 
-	powernow_data[pol->cpu] = data;
+	for_each_cpu_mask(i, cpu_core_map[pol->cpu]) {
+		powernow_data[i] = data;
+	}
 
 	return 0;