> 
> On Jul 20, 2005, at 7:32 PM, Jason Thorpe wrote:
> 
> > So, that would neatly avoid the entire situation.  If I cook up a  
> > patch, would someone be willing to test it?  (I don't have an MP  
> > Alpha system that is in a runnable state, at the moment.)
> 
> Attached is a patch against the pmap in -current.  Can someone try it  
> out?
> 
I tried it on a 2-CPU 8400 system, and the spurious hangings that
occurred before disappeared, so suddenly the machine got usable :-)

After some hours of stressing with "build.sh -j 8" over NFS it
paniced though, in something that seems like an MP problem.
Here's some DDB dumps, tell me if I should try out some more.

-- Ragge

db{1}> bt
cpu_Debugger() at netbsd:cpu_Debugger+0x4
panic() at netbsd:panic+0x1e8
fpusave_proc() at netbsd:fpusave_proc+0x22c
cpu_lwp_fork() at netbsd:cpu_lwp_fork+0x198
uvm_lwp_fork() at netbsd:uvm_lwp_fork+0xa8
newlwp() at netbsd:newlwp+0x114
fork1() at netbsd:fork1+0x4b4
sys___vfork14() at netbsd:sys___vfork14+0x38
syscall_plain() at netbsd:syscall_plain+0xe0
XentSys() at netbsd:XentSys+0x60
--- syscall (282) ---
--- user mode ---
db{1}> sho reg
v0          0x6
t0          0xfffffc00005c2b38  db_onpanic
t1          0x1
t2          0xfffffc003ff56000
t3          0x1d56600011101
t4          0x1
t5          0xfffffc003ff56000
t6          0xad4
t7          0xfffffc00005efd3e  proclist_lock+0xe
s0          0x100
s1          0xfffffc00005c3714  msgbufenabled
s2          0x2710
s3          0x1
s4          0
s5          0xfffffc00004e6f00  child_return
s6          0xfffffc000059e021  msgperiod.3+0x6b1
a0          0
a1          0
a2          0x20000000
a3          0x1
a4          0
a5          0x8
t8          0
t9          0x187c0
t10         0x8
t11         0xfffffc003d949650
ra          0xfffffc0000433918  panic+0x1e8
t12         0xfffffc00004f01e0  cpu_Debugger
at          0x20001ff8
gp          0xfffffc00005b8b30  osf1_sysent+0xa000
sp          0xfffffe00138ddc18
pc          0xfffffc00004f01e4  cpu_Debugger+0x4
ps          0
ai          0xfffffc003d949650
pv          0xfffffc00004f01e0  cpu_Debugger

db{1}> machine cpu
addr            dev     id      flags   ipis    curproc         fpcurproc
0xfffffc00005f4f88      cpu0    0       1d      0       0xfffffc003c181690      0
xfffffc003c1813c0
0xfffffe000009bd00      cpu1    1       4       0       0xfffffc003c1813c0      0

db{1}> machine cpu 0
CPU 0 has no register state