How can I identify the cause of "Bad page map in process ..." kernel errors?

  • 1
  • Question
  • Updated 4 years ago
  • Doesn't Need an Answer
Running 6.1r2 build 1365 on HiveAP 330s, I am observing the following occurring intermittently on some access points:

2013-12-11 17:15:05 alert
kernel: BUG: Bad page map in process ah_dcd pte:cd470028 pmd:cd50c000

2013-12-11 17:15:05 alert
kernel: addr:4cbf8000 vm_flags:00100073 anon_vma:cd261e14 mapping:(null) index:4cbf8

(Repeatedly written 'ad nauseam'...)

Any thoughts on how I might identify the cause/trigger of this bad page table entry, in this case for the Device Control Daemon (ah_dcd) that manages the wireless interfaces (AR9390s in the 330/350s)? It is so intermittent that I am presently at a loss of how to approach this methodically.

Is there an implemented/supported mechanism in HiveOS that would allow me to get a memory dump to pass on via reseller when this occurs so that the issue can be hopefully isolated? (I don't have the inner binaries, and obviously no debugging symbols or source so couldn't do anything particularly meaningful myself with GDB et al.)

The code in the Linux Kernel's mm/memory.c here that writes this is, of course:

"BUG: Bad page map in process %s pte:%08llx pmd:%08llx\n",
(long long)pte_val(pte), (long long)pmd_val(*pmd));

It is called under print_bad_pte(...), documented as:

* This function is called to print an error when a bad pte
* is found. For example, we might have a PFN-mapped pte in
* a region that doesn't allow it.

And... a more general, abstract musing... Why does a watchdog mechanism not force a reboot of the access point if-and-when something like this occurs rather than leaving the device in a pathological state? Does the boot loader (U-Boot) normally check/exercise the RAM before booting to probabilistically rule out hardware faults there?

Should I just wait out 6.1r3 and see if the issue reoccurs? Is this, perhaps, due to a known bug that has already been resolved?


Photo of Nick Lowe

Nick Lowe, Official Rep

  • 2491 Posts
  • 451 Reply Likes

Posted 5 years ago

  • 1
Photo of Nick Lowe

Nick Lowe, Official Rep

  • 2491 Posts
  • 451 Reply Likes
Just a quick update on this...

I have not yet seen the issue reoccur on 6.1r3 so consider the case closed at the moment.
Photo of Amanda


  • 396 Posts
  • 25 Reply Likes
Thank you for the update Nick.