Subject: port-sparc/30327: using X crashes NetBSD/sparc on SS20 + cg14 (UP and SMP)
To: None <port-sparc-maintainer@netbsd.org, gnats-admin@netbsd.org,>
From: None <francois.cami@free.fr>
List: netbsd-bugs
Date: 05/24/2005 22:26:01
>Number: 30327
>Category: port-sparc
>Synopsis: using X crashes NetBSD/sparc on SS20 + cg14 (UP and SMP)
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: port-sparc-maintainer
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Tue May 24 22:26:00 +0000 2005
>Originator: François Cami
>Release: 2.0.2
>Organization:
>Environment:
NetBSD forestiere 2.0.2 NetBSD 2.0.2 (GENERIC) #0: Wed Mar 23 01:25:37 UTC 2005 jmc@faith.netbsd.org:/home/builds/ab/netbsd-2-0-2-RELEASE/sparc/200503220140Z-obj/home/builds/ab/netbsd-2-0-2-RELEASE/src/sys/arch/sparc/compile/GENERIC sparc
>Description:
hardware configuration :
SS20 + 2x SM71 or 2x HM125 + cg14 + hme + 4GB sun/seagate hdd
************************
software configuration :
2.0.2 binaries from ftp.netbsd.org
UP kernel used (SMP kernel crashes too, but stack trace is from UP) on the 2x SM71 config
************************
Using X (cg3 emulation on cg14) easily crashes X, causing either a
"Level 15 interrupt"
or a
"watchdog reset"
or a
kernel panic (2.0.2) with stack trace
************************
sample stack trace (I can generate others if needed, it's almost too easy) :
panic: illegal instruction
syncing disks... 20 20 20 20 20 19 19 18 18 18 17 17 16 16 15 15 14 13 13 12 12 12 10 10 9 9 8 8 7 6 6 6 6 giving up
Frame pointer is at 0xf3706b50
Call traceback:
pc = 0xf027a208 args = (0xf02e1800, 0x5, 0x0, 0x0, 0xf3706c70, 0x1, 0xf3706bb8) fp = 0xf3706bb8
pc = 0xf01a3d24 args = (0x100, 0x0, 0x0, 0x0, 0x0, 0x42, 0xf3706c28) fp = 0xf3706c28
pc = 0xf0287b68 args = (0xf02ed690, 0x2, 0xf3706e34, 0xf3706e38, 0xf035ec00, 0x100, 0xf3706c90) fp = 0xf3706c90
pc = 0xf0008694 args = (0x2, 0x400010c1, 0xf3706e34, 0xf3706dc8, 0x0, 0x0, 0xf3706d68) fp = 0xf3706d68
pc = 0xf3706e2c args = (0x0, 0xf3706e58, 0xf3706e7c, 0xf308ba70, 0xf02b2174, 0xf36f6740, 0xf3706e18) fp = 0xf3706e18
pc = 0xeffff510 args = (0x1f0001, 0x14701cc, 0x40008, 0x0, 0x2, 0xeffff520, 0xeffff518) fp = 0xeffff518
dump to dev 7,1 not possible
rebooting
Resetting ...
****************************
image of stack trace :
http://tinypic.com/5cnn06
****************************
gdb /kernel says :
0xf027a208 <cpu_reboot+452>: 0x4000004d
0xf01a3d24 <panic+232>: 0x400358c8
0xf0287b68 <trap+140>: 0x7ffc7035
0xf0008694 <Lslowtrap_reenter+56>: 0x4009fd12
0xf3706e2c: Cannot access memory at address 0xf3706e2c
0xeffff510: Cannot access memory at address 0xeffff510
****************************
NB : I have been unable to crash either OpenBSD 3.7 or Debian/Sarge (XFree 4.3 + 2.4.27).
>How-To-Repeat:
1/ full install of netbsd 2.0.2
2/ install the evilwm package (compiled for 2.0)
3/ put a .xinitrc with the only line "exec evilwm" inside
4/ start X
5/ launch an xterm on virtual desktops #1 and #2
now :
* switching from desktop #1 and #2 (sc 1 and 2) back and forth does NOT crash (mouse is over only one xterm) which is obviously the expected behaviour
* switching from desktop #1 and #2 (sc 3 and 4) back and forth does crash after a few times (between 4 and 20) (mouse is over both xterms)
(1) http://tinypic.com/5catlu
(2) http://tinypic.com/5catmr
(3) http://tinypic.com/5catn7
(4) http://tinypic.com/5catqd
addendum :
CTRL+ALT+1 (not numeric keypad) is used to activate desktop 1
CTRL+ALT+2 (not numeric keypad) is used to activate desktop 2
CTRL+ALT+ENTER is used to launch an xterm at approximately the mouse's position
>Fix:
none known ( :/ )