Fatal trap 12: page fault while in kernel mode

Hi
Since the last base system upgrade (14-STABLE from tongiht) my server panics after 10s of usage
The crashdump is attached.
I **know** that my memory is in bad state :

numenor.groumpf.org hardware dmi status:
--- /var/log/dmidecode.today 2025-07-23 04:21:15.265767000 +0200
+++ /tmp/security.du3KoWI8LJ 2025-08-01 04:48:24.887650000 +0200
@@ -5 +5 @@
-Table at 0xDDF3D018.
+Table at 0xDDF43018.
@@ -324,2 +324,2 @@
- Total Width: 64 bits
- Data Width: 64 bits
+ Total Width: 16 bits
+ Data Width: 8 bits
@@ -327 +327 @@
- Form Factor: DIMM
+ Form Factor: SODIMM
@@ -373,2 +373,2 @@
- Total Width: 64 bits
- Data Width: 64 bits
+ Total Width: 16 bits
+ Data Width: 8 bits
@@ -376 +376 @@
- Form Factor: DIMM
+ Form Factor: SODIMM
@@ -400 +400 @@
- Change Token: 0x000001B6
+ Change Token: 0x000001B7

Could it be the root cause ?
I have already ordered memory on eBay
Regards,
Xavier
 

Attachments

Maybe you have included some faulty commit after update, maybe update the sources and then rebuild world and kernel then test again, if you have time?
 
Just a wild guess but possibly something loaded via /boot/loader.conf (including kernel) would be causing the issue, unless physical memory itself is faulty.

As you already mentioned that the memory is in bad state (only missingly recognized?), what you need to do would be going into BIOS/firmware menu and check (and fix, if possible) the configuration. Possibly you need to replace your memory module(s), but if the problem is in SPD and your motherboard allows overriding detailed memory configurations, you can override broken configuration with correct (if you know / remember it) ones.

If the memory is OK, you can boot with previous kernel (if correctly saved and known to work) or unload suspicious modules on loader prompt.
 
The previous kernel panics also ;-(
Just before the panic, I have a kernel message about ipmi0 KCS start write failed. Don't know if it is related
 
Are you loading ipmi.ko and/or ipmi_linux.ko in your /boot/loader.conf?
If yes and you don't have strong reason to do so, does anything change if you unload it via loader prompt? Basic loader commands can be seen in loader_simp(8) manpage.

And if you're loading any huge but not mandatory on boot modules like GPU drivers in your /boot/loader.conf, stop it and load them via kld_list variable in /etc/rc.conf[.local].
 
Back
Top