Re: Recent i386-GENERIC panic

[christos%astron.com@localhost (Christos Zoulas)]

In article <Pine.NEB.4.64.1109160613001.17072%david.technoskunk.fur@localhost>,
John D. Baker <jdbaker%mylinuxisp.com@localhost> wrote:
>On Thu, 15 Sep 2011, John D. Baker wrote:
>
>> Following an update, a second machine now suffers the exact same panic.
>> It's an Intel D850MV board.  This one has a serial port, so I should
>> be able to capture the console output.
>
>After finding a working USB/RS232 dongle and wrestling with a
>temperamental KVM switch, I've captured the dmesg with panic and
>backtrace from this second machine.  It's essentially the same as
>on the machine about which I first posted.
>
>   http://bobdbob.com/~jdbaker/nbsd-debug/fred-panic_bt-201109160300Z
>
>This reflects a release built from the latest HEAD (as of last night).
>
>
>If I revert "sys/dev/pci/pci.c" to r1.141 I can produce a kernel that
>doesn't panic, but produces the warnings about builtin dependencies.
>
>I suppose I should see if a DIAGNOSTIC kernel will provide more
>information.  DEBUG?

Try adding a print to give more info (not compile-tested):

christos

Index: kern_module.c
===================================================================
RCS file: /cvsroot/src/sys/kern/kern_module.c,v
retrieving revision 1.81
diff -u -u -r1.81 kern_module.c
--- kern_module.c       14 Sep 2011 12:29:22 -0000      1.81
+++ kern_module.c       16 Sep 2011 12:57:44 -0000
@@ -689,6 +689,9 @@
 
                /* Add references to the requisite modules. */
                for (i = 0; i < mod->mod_nrequired; i++) {
+if (mod->mod_required[i])
+printf("module %s requires module %s!\n",
+       mod->mod_name, mod->mod_required[i].mod_name)l
                        KASSERT(mod->mod_required[i] != NULL);
                        mod->mod_required[i]->mod_refcnt++;
                }