NetBSD-Bugs archive

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

kern/56445: DDB_DUMPSTACK is not documented and counterintuitive



>Number:         56445
>Category:       kern
>Synopsis:       DDB_DUMPSTACK is not documented and counterintuitive
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Oct 06 12:10:00 +0000 2021
>Originator:     Valery Ushakov
>Release:        NetBSD-9.99.90
>Organization:
>Environment:
>Description:
When current kernel panics it spews out ddb stack trace.

This is apparently (from a quick RTFS) controlled by the kernel option
DDB_DUMPSTACK that is not in any kernel config in the tree and is not
documented in ddb(4).

The behaviour is completely counter-productive when db_onpanic is true
(the user presumably knows how to use bt), especially on non-serial
consoles when large/infinite scrollback is not available.

sys/ddb/db_panic.c:

revision 1.7
date: 2018-02-17 03:41:09 +0300;  author: sevan;  state: Exp;  lines: +6 -6;  commitid: fYewV3fKl2bA89rA;
branches:  1.7.2;
Opt to print a backtrace on panic by default with the intention of improving bug reports.
Instead of relying on ddb.onpanic=2, introduce a new sysctl called dumpstack to handle this.
via <jmcneill> channeled through <mrg> on tech-kern[1]

[1] https://mail-index.netbsd.org/tech-kern/2018/02/15/msg023103.html

>How-To-Repeat:
Get a kernel panic with GENERIC on something with a wscons console.

>Fix:
Make the defaults sane, db_onpanic should be factored into the
decision to dump the stack.

Document the kernel option, add it (commented out) to configs.



Home | Main Index | Thread Index | Old Index