DragonFly BSD
DragonFly bugs List (threaded) for 2004-05
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Fwd: Re: buildworld panics with yesterday's kernel ...


From: Andrew Atrens <atrens@xxxxxxxxxxxxxxxxxx>
Date: Wed, 5 May 2004 15:27:22 -0400

I sent this a bit earlier but didn't copy the newsgroup -


----------  Forwarded Message  ----------

Subject: Re: buildworld panics with yesterday's kernel ...
Date: May 5, 2004 02:20 pm
From: Andrew Atrens <atrens@xxxxxxxxxxxxxxxxxx>

On May 5, 2004 02:07 pm, you wrote:
> :Hi All,
> :
> :With kern.mmxopt=0, my buildworld's are failing with
> :'internal compiler error' at random times, and on a
> :couple of occasions my system has panicked. Here are
> :the gdb backtraces on the vmcores -
> :
> :First one (kernel not built for debugging)
>
>     Andrew, you are getting the 'vm_page_alloc: free/cache page %p was
> dirty' panic.  This is what was happening after my initial FP code commit.
>
>     Note that this is a very dangerous panic to get, you risk filesystem
>     corruption!  I recommend that you drop back to a more reliable
>     kernel before proceeding.
>
>     What is the build date on your kernel and also make sure your source
>     tree is up-to-date!

These are the only changes that I'm missing -

Edit src/share/man/man9/Makefile,v
  Add delta 1.6 2004.05.05.16.57.11 hmp
 Edit src/share/man/man9/bus_alloc_resource.9,v
  Add delta 1.3 2004.05.05.16.57.11 hmp
 Edit src/sys/kern/kern_lockf.c,v
  Add delta 1.9 2004.05.05.09.17.04 joerg
 Edit src/sys/kern/tty.c,v
  Add delta 1.9 2004.05.04.23.22.43 dillon
 Edit src/sys/libkern/strcat.c,v
  Add delta 1.5 2004.05.05.00.17.44 hsu
 Edit src/sys/libkern/strcpy.c,v
  Add delta 1.6 2004.05.05.00.17.44 hsu
 Edit src/sys/libkern/strncpy.c,v
  Add delta 1.5 2004.05.05.00.17.44 hsu
 Edit src/sys/netinet6/ip6_input.c,v
  Add delta 1.15 2004.05.04.23.08.43 dillon
 Edit src/sys/sys/bus.h,v
  Add delta 1.12 2004.05.05.16.57.11 hmp
 Edit src/sys/sys/libkern.h,v
  Add delta 1.7 2004.05.05.00.17.45 hsu

>     Even so, the are likely still issues with the FP code in the current
>     up-to-date source tree, which I've been tracking down with the help
>     of David Rhodus.  But the current failure condition tends to be a
>     very specific 'npxdna' panic.  I hope to commit a fix soon.

Since I can so readily reproduce this feel free to throw some patches my
way :) :) ...

> 					-Matt

-------------------------------------------------------


Attachment: pgp00001.pgp
Description: signature



[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]