kernel crashes and portupgrade

[Kent Stewart]

On Wednesday 30 April 2003 04:28 pm, Terry Lambert wrote:
> Kent Stewart wrote:
> > I just had one running /usr/libexec/locate.updatedb from kde as
> > root. I cvsuped, updated the system, and tried it from the command
> > line and had no problem. I am just letting it sit with kde running
> > and setiathome.
>
> If you are panic'ing, and it's repeatable, then you should
> minimally post:
>
> 1)	The panic message, with the registers, etc. (e.g. "page
> 	fault in kernel mode; Trap 12", etc.).
>

This is going to take longer than I thought. I get the message

dev=ad0s3e block=567360, fs=/var
panic: ffs_blkfree: freeing free block

syncing disks, buffers remaing... panic: bremfree: removing a buffer not 
on a queue

To force it, all I had to do is update the locate.db twice and then ran  
startx. I received the above message immediately.

The system was fsck'ed from single user mode after all panics and then, 
the last 2 were booted to 

FreeBSD opal 5.0-CURRENT FreeBSD 5.0-CURRENT #13: Wed Apr 30 19:40:40 
PDT 2003     root at opal:/usr/obj/usr/src/sys/OPAL  i386

This is a really clean system and has never been setup to produce debug 
results. I will have to do that first. I thought I had created a debug 
kernel but that didn't work. I have some reading to do first.

Kent

> 2)	A stack traceback from DDB.
>
> 3)	If you are running something other than a tagged branch
> 	(e.g. if you are running -CURRENT, or using something
> 	other than a -RELEASE ISO for the install), then you need
> 	to take the traceback information, and either post a
> 	"gdb-k" dump of the lines of the file in which it occurred,
> 	or you need to go into an editor like "vi", turn on line
> 	numbering, and cut-and paste at least 5 lines before and
> 	after the line on which it crashed.
>
> After that, someone can probably help you, if not give you a patch.
>
> My personal "pat answer" for any Trap 12 panic is going to be to
> reguild the kernel with options DISABLE_PG_G and DISABLE_PSE, on
> the theory that Matt Dillon and Jeff Robertson's changes which
> mask the problem happen to *not* be masking the problem on your
> particular machine.
>
> After that, we can talk about what else might be causing the problem.
>
> -- Terry

-- 
Kent Stewart
Richland, WA

http://users.owt.com/kstewart/index.html