Subject: Re: KDASSERT(sa->sa_nstacks > 0); in sa_upcall_userret(struct lwp *l)
To: Nathan J. Williams <>
From: Lars Heidieker <>
List: current-users
Date: 02/11/2003 00:18:44
Here the information I got from the system halt:

sa_switch(236.10): no cached LWP for upcall
sa_switch(236.11): no cached LWP for upcall
uvm_fault(0xe31746c0, 0x0, 0, 2) -> e
kernel: page fault trap, code=0
Stopped in pid 236.4 (ltdaemon) at   remrunqueue+0x40:       movl

db> t
remrunqueue(e3b12004,e3177c00,e394d8ec,c01715d9,e3b12004) at
mi_switch(e3177c00,e3b12004,0,c0b72000,e3177c00) at mi_switch+0x198
sa_switch(e3177c00,2,1,c01c2b99,0) at sa_switch+0x236
ltsleep(c029ffb0,204,c023f2bd,0,c02a0154) at ltsleep+0x259
uvm_wait(c023f2bd,6,0,c01b4dc2,e394da84) at uvm_wait+0x8a
uvmfault_anonget(e394daf0,e372d464,e35872c0,5,c01004c8) at
uvm_fault(e31746c0,489ff000,0,2,0) at uvm_fault+0x7da
trap() at trap+0x56c
--- trap (number 6) ---
i486_copyout(e3177c00,e3177c00,e394df80,e394df78,e3177c00) at
upcallret(e3177c00,0,0,e3177c00,0) at upcallret+0x61
sa_switchcall(0,2b,2b,2b,2b) at sa_switchcall+0x75

call debug_print_proc(0t236)
Process 236 (ltdaemon), state 2 address 0xe31a2e14, flags 4402
LWPs: 11 (5 running, 0 zombies)
LWP 11 address 0xe3b12004 state 3 flags b00004 wait 0xc029ffb0
LWP 10 address 0xe3b12404 state 3 flags d00004 wait 0xc029ffb0
LWP 9 address 0xe3b12104 state 2 flags d00004
LWP 8 address 0xe3b12704 state 2 flags d00004
LWP 7 address 0xe3b12384 state 2 flags d00004
LWP 6 address 0xe3b12084 state 2 flags d00004
LWP 5 address 0xe3177c80 state 2 flags d00004
LWP 4 address 0xe3177c00 state 3 flags f00004 wait 0xc029ffb0 flt_noram1

LWP 2 address 0xe3177a00 state 3 flags f00004 wait 0xc029ffb0 flt_noram1

LWP 3 address 0xe3177a80 state 3 flags d00004 wait 0xc029e574 select
LWP 1 address 0xe3177980 state 3 flags d00004 wait 0xc05cc048 anonget2
SA VP: 11
SAs: 0 cached LWPs
0 upcall stacks


"Nathan J. Williams" wrote:

> Jaromir Dolecek <> writes:
> > This should be fixed in rev. 1.8 src/sys/kern/kern_sa.c . Can you
> > update your kernel and try again?
> Wrong crash, I think. 1.8 fixed Lars's race condition with sa_stacks,
> but this looks like something connected with the "no cached LWP for
> upcall" issue. What's the actual panic message?
>         - Nathan

Mystische Erklärungen.
Die mystischen Erklärungen gelten für tief;
die Wahrheit ist, daß sie noch nicht einmal oberflächlich sind.
                --Friedrich Nietzsche