Fatal trap 12: page fault while in kernel mode

[Adam McDougall]

On Thu, Apr 19, 2007 at 02:33:29PM +0200, Kai wrote:

  On Wed, Apr 11, 2007 at 12:53:32PM +0200, Kai wrote:
  > 
  > Hello all,
  > 
  > We're running into regular panics on our webserver after upgrading
  > from 4.x to 6.2-stable:
  
  Hi Again,
  
  The panics keep happening, so I'm trying alternate kernel setups. This is a
  trace of a panic on a default SMP kernel with debugging symbols.
  
<snip>

  Fatal trap 12: page fault while in kernel mode
  cpuid = 0; apic id = 00
  fault virtual address   = 0x34
                          ^^^^^^^^^
  fault code              = supervisor read, page not present
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  #7  0xc06bdefa in vfs_vmio_release (bp=0xdbec2560) at atomic.h:146
                    ^^^^^^^^^^^^^^^^
  #8  0xc06be728 in getnewbuf (slpflag=0, slptimeo=0, size=6585, maxsize=8192)
      at ../../../kern/vfs_bio.c:1779
  #9  0xc06bfccc in getblk (vp=0xca2cfdd0, blkno=8438, size=6585, slpflag=0, 
      slptimeo=0, flags=0) at ../../../kern/vfs_bio.c:2497
  #10 0xc075ad41 in nfs_getcacheblk (vp=0xca2cfdd0, bn=8438, size=6585, 
      td=0xc8cd1c00) at ../../../nfsclient/nfs_bio.c:1261
  #11 0xc075a978 in nfs_write (ap=0x0) at ../../../nfsclient/nfs_bio.c:1069
  #12 0xc089fde6 in VOP_WRITE_APV (vop=0xc0984440, a=0xeb9cfbec)
      at vnode_if.c:698
  #13 0xc06dbb26 in vn_write (fp=0xc8940e10, uio=0xeb9cfcbc, 
      active_cred=0xc89ee880, flags=0, td=0xc8cd1c00) at vnode_if.h:372
  #14 0xc0698f63 in dofilewrite (td=0xc8cd1c00, fd=5, fp=0xc8940e10, 
      auio=0xeb9cfcbc, offset=Unhandled dwarf expression opcode 0x93
  ) at file.h:252
  #15 0xc0698e07 in kern_writev (td=0xc8cd1c00, fd=5, auio=0xeb9cfcbc)
      at ../../../kern/sys_generic.c:402
  #16 0xc0698d2d in write (td=0xc8cd1c00, uap=0xc8cd1c00)
      at ../../../kern/sys_generic.c:326

I believe I am seeing the same panic on my samba servers, sometimes from NFS
and sometimes from FFS.  I see it on i386 and amd64 alike.  I do not know how
to manually trigger it, but I do have two servers sitting in DDB from after the
panic, waiting for more experienced hands to continue the debugging from what
I have already done.  I filed a PR with as much details as I could think of,
and it would be wonderful if someone could look at it and either tell me what
else to do in DDB, or I could provide remote access to the existing DDB session
to a developer.  Both servers crashed in vfs_vmio_release but one was through
NFS and one through FFS.  

pr 111831
http://docs.freebsd.org/cgi/mid.cgi?200704181924.l3IJOMUL088901