Re: core dump in cvsup caused by _once()?

On Wed, 2 Dec 2009 09:37:23 -0500
John Baldwin <jhb_at_freebsd.org> wrote:

> 
> Can you get a ktrace?  I wonder if it is trying to handle a signal 11 (name of 
> '0xb' is rather bogus) and triggering SIGILL due to a problem with that.
> 

Yup, you hit the nail on the head.  The last few lines from kdump

  2416 cvsup    CALL  gettimeofday(0x774a08,0x774a18)
  2416 cvsup    RET   gettimeofday 0
  2416 cvsup    CALL  gettimeofday(0x774a98,0x774aa8)
  2416 cvsup    RET   gettimeofday 0
  2416 cvsup    CALL  open(0x794270,O_RDONLY,<unused>0)
  2416 cvsup    RET   open 12/0xc
  2416 cvsup    CALL  fstat(0xc,0x774cf8)
  2416 cvsup    RET   fstat 0
  2416 cvsup    CALL  mmap(0,0x4db47,PROT_READ,MAP_SHARED,0xc,0)
  2416 cvsup    RET   mmap 6221824/0x8005ef000
  2416 cvsup    CALL  close(0xc)
  2416 cvsup    RET   close 0
  2416 cvsup    CALL  issetugid
  2416 cvsup    RET   issetugid 0
  2416 cvsup    CALL  open(0x7745c8,O_RDONLY,<unused>0x2e)
  2416 cvsup    RET   open 12/0xc
  2416 cvsup    CALL  fstat(0xc,0x7749d8)
  2416 cvsup    RET   fstat 0
  2416 cvsup    CALL  read(0xc,0x765c68,0xa1e8)
  2416 cvsup    RET   read 118/0x76
  2416 cvsup    CALL  close(0xc)
  2416 cvsup    RET   close 0
  2416 cvsup    PSIG  SIGSEGV caught handler=0x499ea0 mask=0x0 code=0x2

---
Gary Jennejohn