[Ilugc] "Recursive entry to debugger" panic seen with KGDB (PowerPC)

  • From: itjegan@xxxxxxxxx (Jegathesan Durairaj)
  • Date: Mon, 17 Mar 2014 11:21:51 +0530

Prasanna,

According to the documentation given by official KGDB, I have been asked to
try connect KGDB only after triggering.

https://www.kernel.org/doc/htmldocs/kgdb/EnableKGDB.html

Once the trigger is done, it is supposed to hang and I need to connect. But
in my case it reboots.

I also tried connecting before triggering KGDB, but connection failed
because there was no KGDB hang in other side

*GDB Log*

(gdb) target remote 10.31.185.33:3028
Remote debugging using 10.31.185.33:3028
Sending packet: $Hc-1#09...putpkt: Junk:{{{}
     Welcome to Console Server Management Server HQ1
Nak
Sending packet: $Hc-1#09...putpkt: Junk: 2E17
Nak
Sending packet: $Hc-1#09...putpkt: Junk: TS2 port S028

HQ1
Nak
Sending packet: $Hc-1#09...putpkt: Junk: 2E17
Nak
Packet received: Hc-1
Sending packet: $qC#b4...Packet instead of Ack, ignoring it
Packet instead of Ack, ignoring it
Packet instead of Ack, ignoring it
Ack
Packet received: qC
Sending packet: $qOffsets#4b...Ack
Packet received: qOffsets
Malformed response to offset query, qOffsets  (I hope here connection
failed)
(gdb)

Then tried triggering KGDB, it failed with same reboot.

Correct me if I am wrong.

Thanks
Jegan


On Mon, Mar 17, 2014 at 10:37 AM, PrasannaKumar Muralidharan <
prasannatsmkumar at gmail.com> wrote:

This is what is expected I guess. With "echo g > /proc/sysrq-trigger"
you are triggering a kernel panic. You should have connected the
target to gdb before a kernel panic. Try connecting the target to gdb
and then trigger kernel panic.

Hope this helps,
PrasannaKumar
_______________________________________________
ILUGC Mailing List:
http://www.ae.iitm.ac.in/mailman/listinfo/ilugc
ILUGC Mailing List Guidelines:
http://ilugc.in/mailinglist-guidelines


Other related posts: