system crash when probe all syscalls on Power5

["Guanglei Li" <guanglei at cn dot ibm dot com>]

Hi,

  I tried to probe all the syscalls entry&return for both 
32-bit/64-bit application on Power5, the following is the function 
list I probed:
  sys_*,compat_sys_*, 
sys32_getdents,sys32_adjtimex,sys32_pause,sys32_gettimeofday,sys32_settimeofday,sys32_ipc,sys32_execve,sys32_sysctl,sys32_olduname,sys32_utimes,ppc_rtas,sys64_time,ppc64_newuname

  But when I started the probe: stap -I./mytapsets -b syscall.stp
  And doing something at the same time(such ls or cp). It showed my 
system is very easy to crash.

  I tried to probe all syscalls on my PC, but it seems much more 
stable than Power5.

  When the system crashed, the xmon will be executed. The following is 
the info of my system and the info given by xmon
  I compiled my kernel using kernel-2.6.9-22.17.EL.src.rpm and 
rpmbuild command.

root:/root>uname -r
2.6.9-22.17.EL.root

gcc version 3.4.4 20050721 (Red Hat 3.4.4-2)

2:mon> e (print exception information)
cpu 0x2: Vector: 700 (Program Check) at [c00000019d53b630]
    pc: c000000000047968: .trampoline_probe_handler+0xec/0x150
    lr: c0000000000478c4: .trampoline_probe_handler+0x48/0x150
    sp: c00000019d53b8b0
   msr: 8000000000021032
  current = 0xc0000003a3aa42c0
  paca    = 0xc0000000003ef400
    pid   = 4319, comm = ls


2:mon> t (print backtrace)
[c00000019d53b8b0] c0000000000478b4 
.trampoline_probe_handler+0x38/0x150 (unreliable)
[c00000019d53b950] c000000000047d70 
.kprobe_exceptions_notify+0x340/0x5f4
[c00000019d53ba10] c000000000071030 .notifier_call_chain+0x68/0x98
[c00000019d53baa0] c0000000000127b0 .ProgramCheckException+0x80/0x17c
[c00000019d53bb40] c00000000000b048 ProgramCheck_common+0xc8/0x100
--- Exception: 700 (Program Check) at c000000000047874 
.kretprobe_trampoline_holder+0x0/0x8
--- Exception: c01 (System Call) at 000000000ffe0518
SP (ffffe6f0) is in userspace

2:mon> r  (print registers)
R00 = 0000000000000001   R16 = 00000000ffffffff
R01 = c00000019d53b8b0   R17 = 00000000ffffffff
R02 = c0000000004e86d0   R18 = 0000000000000000
R03 = 0000000000000000   R19 = 0000000000000000
R04 = c00000019d53bbb0   R20 = 0000000000000000
R05 = c00000019d53bb10   R21 = 0000000000000000
R06 = c0000003a3aa4a10   R22 = 0000000000000000
R07 = 0000000000000000   R23 = 0000000000000000
R08 = c0000000003941b0   R24 = c0000000004e7208
R09 = 0000000000000001   R25 = c0000001b2ea77b8
R10 = 0000000000000000   R26 = c00000019d53bbb0
R11 = 0000000000000000   R27 = c000000000047874
R12 = 8000000000029032   R28 = 8000000000009032
R13 = c0000000003ef400   R29 = c000000000414cb0
R14 = 00000000100ff870   R30 = 0000000000000000
R15 = 00000000100eb858   R31 = c000000000047874
pc  = c000000000047968 .trampoline_probe_handler+0xec/0x150
lr  = c0000000000478c4 .trampoline_probe_handler+0x48/0x150
msr = 8000000000021032   cr  = 28222442
ctr = c00000000004787c   xer = 0000000000000010   trap =      700

2:mon> S (print special registers)
msr  = 8000000000001032  sprg0= 0000000000000000
pvr  = 00000000003b0200  sprg1= c0000000003ef400
dec  = 0000000087811f89  sprg2= c0000000003ef400
sp   = c00000019d53aef0  sprg3= c0000000003ef400
toc  = c0000000004e86d0  dar  = c00000038d587fa0
srr0 = c000000000036c40  srr1 = 8000000000001032

  Could anybody give me some help about this?
  thanks.