Home / Browse / OProfile / Mail / Archive

Donate

OProfile

Email Archive: oprofile-list (read-only)

Trouble running oprofile with 2.6.8 & Gentoo

From: Zarakin <zarakin@ho...> - 2004-08-23 03:14

Hi,

I've been trying to get oprofile to run in my Gentoo box and I seem to be
getting an oops everytime I run "opcontrol --start".
Some specs:
- ST61G4 Shuttle box (ATI IXP chipset) running a P4 2.8Ghz with Nvidia
MX440, 512MB RAM.
- Kernel 2.6.8

I've tried various combinations of having oprofile driver as module, having
power management off (though shouldn't matter with 2.6.x kernels?)... The
oops is always the same.
---->8----->8-----
general protection fault: 0000 [#1]
Modules linked in: ipt_LOG ipt_state ip_conntrack iptable_filter ip_tables
uhci_
hcd at76c503_rfmd at76c503 usbdfu ati_agp ohci_hcd b44 mii snd_pcm_oss
snd_mixer
_oss snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device snd_atiixp
snd_ac97_c
odec snd_pcm snd_timer snd snd_page_alloc tveeprom ivtv tuner saa7115
msp3400 i2
c_algo_bit i2c_core videodev nvidia nvidia_agp agpgart sbp2 ohci1394
ieee1394 us
b_storage ehci_hcd usbcore
CPU:    0
EIP:    0060:[<c0366ba8>]    Tainted: P
EFLAGS: 00010297   (2.6.8-gentoo)
EIP is at nmi_cpu_save_registers+0x4f/0x6c
eax: 00000000   ebx: 00000210   ecx: 000003ba   edx: 00000000
esi: d7c5b900   edi: 0000002c   ebp: d89b8000   esp: d30dfef4
ds: 007b   es: 007b   ss: 0068
Process oprofiled (pid: 9511, threadinfo=d30de000 task=d30916f0)
Stack: 0000003f 00000008 00000000 c0476fa0 dfd90d80 ffffffe9 c0366cea
c0519d08
       c0365411 ffffffea d305dc80 c036600c 00000001 d305dc80 d837c800
c0148336
       d837c800 d305dc80 00000000 00000000 d30ca000 d30de000 c014821f
d7c3476c
Call Trace:
 [<c0366cea>] nmi_setup+0x26/0x5c
 [<c0365411>] oprofile_setup+0x41/0xa3
 [<c036600c>] event_buffer_open+0x65/0x84
 [<c0148336>] dentry_open+0x115/0x1b4
 [<c014821f>] filp_open+0x62/0x64
 [<c013e872>] unmap_vma_list+0x1c/0x28
 [<c0148408>] get_unused_fd+0x33/0xb5
 [<c014852d>] sys_open+0x5b/0x8b
 [<c010571f>] syscall_call+0x7/0xb
Code: 0f 32 83 c7 01 89 44 1d 08 89 54 1d 04 83 c3 0c 3b 3c 24 72
---->8----->8-----

Any ideas what could be causing this? It seems NMI is causing some problems?

Thread View

Thread	Author	Date
Trouble running oprofile with 2.6.8 & Gentoo	Zarakin <zarakin@ho...>

Re: Trouble running oprofile with 2.6.8 & Gentoo

From: John Levon <levon@mo...> - 2004-08-23 15:08

On Sun, Aug 22, 2004 at 08:13:58PM -0700, Zarakin wrote:

> EIP:    0060:[<c0366ba8>]    Tainted: P

                               ^^^^^^^^^^

Please reproduce without proprietary modules loaded, or we can't help
you.

regards
john

Re: Trouble running oprofile with 2.6.8 & Gentoo

From: Zarakin <zarakin@ho...> - 2004-08-23 16:14

> > EIP:    0060:[<c0366ba8>]    Tainted: P
>
>                                ^^^^^^^^^^
>
> Please reproduce without proprietary modules loaded, or we can't help
> you.

Here you go...
---->8---->8----
general protection fault: 0000 [#1]
PREEMPT
Modules linked in: ipt_LOG ipt_state ip_conntrack iptable_filter ip_tables
uhci_
hcd at76c503_rfmd at76c503 usbdfu ati_agp agpgart ohci_hcd b44 mii
snd_pcm_oss s
nd_mixer_oss snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device
snd_atiixp sn
d_ac97_codec snd_pcm snd_timer snd snd_page_alloc tuner saa7115 msp3400
i2c_algo
_bit i2c_core videodev sbp2 ohci1394 ieee1394 usb_storage ehci_hcd usbcore
CPU:    0
EIP:    0060:[<c0359114>]    Not tainted
EFLAGS: 00010297   (2.6.8-gentoo)
EIP is at nmi_cpu_save_registers+0x4f/0x6c
eax: 00000000   ebx: 00000210   ecx: 000003ba   edx: 00000000
esi: dbc46080   edi: 0000002c   ebp: de320000   esp: da15def0
ds: 007b   es: 007b   ss: 0068
Process oprofiled (pid: 9218, threadinfo=da15c000 task=da16f360)
Stack: 0000003f 00000008 00000000 c0462420 dfdede80 00000001 c035927b
c04fb748
       c0357911 ffffffea da156e80 c035854e 00000001 da156e80 de673380
c014943c
       de673380 da156e80 40015000 00000000 00000000 df4ca000 da15c000
c01492f7
Call Trace:
 [<c035927b>] nmi_setup+0x26/0x5c
 [<c0357911>] oprofile_setup+0x41/0xa3
 [<c035854e>] event_buffer_open+0x65/0x84
 [<c014943c>] dentry_open+0x143/0x222
 [<c01492f7>] filp_open+0x62/0x64
 [<c0149552>] get_unused_fd+0x37/0xd9
 [<c01496c5>] sys_open+0x5b/0x8b
 [<c0105983>] syscall_call+0x7/0xb
Code: 0f 32 83 c7 01 89 44 1d 08 89 54 1d 04 83 c3 0c 3b 3c 24 72
---->8---->8----

Re: Trouble running oprofile with 2.6.8 & Gentoo

From: John Levon <levon@mo...> - 2004-08-23 17:17

On Sun, Aug 22, 2004 at 08:13:58PM -0700, Zarakin wrote:

> I've been trying to get oprofile to run in my Gentoo box and I seem to be
> getting an oops everytime I run "opcontrol --start".

Can you do:

cat /root/.oprofile/daemonrc

> EIP is at nmi_cpu_save_registers+0x4f/0x6c

gdb /path/to/your/vmlinux   # or /path/to/oprofile/module if built as a module

disas nmi_cpu_save_registers

Also, can you safely boot with nmi_watchdog=2 on this machine?

regards
john

Re: Trouble running oprofile with 2.6.8 & Gentoo

From: Zarakin <zarakin@ho...> - 2004-08-24 01:37

> > I've been trying to get oprofile to run in my Gentoo box and I seem to
be
> > getting an oops everytime I run "opcontrol --start".
>
> Can you do:
>
> cat /root/.oprofile/daemonrc

CHOSEN_EVENTS[0]=GLOBAL_POWER_EVENTS:100000:1:1:1
NR_CHOSEN=1
SEPARATE_LIB=0
SEPARATE_KERNEL=0
SEPARATE_THREAD=0
SEPARATE_CPU=0
VMLINUX=none
IMAGE_FILTER=
CPU_BUF_SIZE=0
CALLGRAPH=0

> > EIP is at nmi_cpu_save_registers+0x4f/0x6c
> disas nmi_cpu_save_registers

Dump of assembler code for function nmi_cpu_save_registers:
0xc03590c5 <nmi_cpu_save_registers+0>:  push   %ebp
0xc03590c6 <nmi_cpu_save_registers+1>:  push   %edi
0xc03590c7 <nmi_cpu_save_registers+2>:  xor    %edi,%edi
0xc03590c9 <nmi_cpu_save_registers+4>:  push   %esi
0xc03590ca <nmi_cpu_save_registers+5>:  push   %ebx
0xc03590cb <nmi_cpu_save_registers+6>:  sub    $0x8,%esp
0xc03590ce <nmi_cpu_save_registers+9>:  mov    0xc04fb744,%eax
0xc03590d3 <nmi_cpu_save_registers+14>: mov    0x1c(%esp,1),%edx
0xc03590d7 <nmi_cpu_save_registers+18>: mov    (%eax),%ecx
0xc03590d9 <nmi_cpu_save_registers+20>: mov    %ecx,0x4(%esp,1)
0xc03590dd <nmi_cpu_save_registers+24>: cmp    %ecx,%edi
0xc03590df <nmi_cpu_save_registers+26>: mov    0x4(%eax),%eax
0xc03590e2 <nmi_cpu_save_registers+29>: mov    %eax,(%esp,1)
0xc03590e5 <nmi_cpu_save_registers+32>: mov    (%edx),%esi
0xc03590e7 <nmi_cpu_save_registers+34>: mov    0x4(%edx),%ebp
0xc03590ea <nmi_cpu_save_registers+37>: jae    0xc0359107
<nmi_cpu_save_register
s+66>
0xc03590ec <nmi_cpu_save_registers+39>: xor    %ebx,%ebx
0xc03590ee <nmi_cpu_save_registers+41>: mov    (%esi,%ebx,1),%ecx
0xc03590f1 <nmi_cpu_save_registers+44>: rdmsr
0xc03590f3 <nmi_cpu_save_registers+46>: add    $0x1,%edi
0xc03590f6 <nmi_cpu_save_registers+49>: mov    %eax,0x8(%esi,%ebx,1)
0xc03590fa <nmi_cpu_save_registers+53>: mov    %edx,0x4(%esi,%ebx,1)
0xc03590fe <nmi_cpu_save_registers+57>: add    $0xc,%ebx
0xc0359101 <nmi_cpu_save_registers+60>: cmp    0x4(%esp,1),%edi
0xc0359105 <nmi_cpu_save_registers+64>: jb     0xc03590ee
<nmi_cpu_save_registers+41>
0xc0359107 <nmi_cpu_save_registers+66>: xor    %edi,%edi
0xc0359109 <nmi_cpu_save_registers+68>: cmp    (%esp,1),%edi
0xc035910c <nmi_cpu_save_registers+71>: jae    0xc0359129
<nmi_cpu_save_registers+100>
0xc035910e <nmi_cpu_save_registers+73>: xor    %ebx,%ebx
0xc0359110 <nmi_cpu_save_registers+75>: mov    0x0(%ebp,%ebx,1),%ecx
0xc0359114 <nmi_cpu_save_registers+79>: rdmsr
0xc0359116 <nmi_cpu_save_registers+81>: add    $0x1,%edi
0xc0359119 <nmi_cpu_save_registers+84>: mov    %eax,0x8(%ebp,%ebx,1)
0xc035911d <nmi_cpu_save_registers+88>: mov    %edx,0x4(%ebp,%ebx,1)
0xc0359121 <nmi_cpu_save_registers+92>: add    $0xc,%ebx
0xc0359124 <nmi_cpu_save_registers+95>: cmp    (%esp,1),%edi
0xc0359127 <nmi_cpu_save_registers+98>: jb     0xc0359110
<nmi_cpu_save_registers+75>
0xc0359129 <nmi_cpu_save_registers+100>:        add    $0x8,%esp
0xc035912c <nmi_cpu_save_registers+103>:        pop    %ebx
0xc035912d <nmi_cpu_save_registers+104>:        pop    %esi
0xc035912e <nmi_cpu_save_registers+105>:        pop    %edi
0xc035912f <nmi_cpu_save_registers+106>:        pop    %ebp
0xc0359130 <nmi_cpu_save_registers+107>:        ret
End of assembler dump.

> Also, can you safely boot with nmi_watchdog=2 on this machine?

No - I got an another Oops in "clear_msr_range" - EIP at <c0110d4b>

Dump of assembler code for function clear_msr_range:
0xc0110d33 <clear_msr_range+0>: push   %edi
0xc0110d34 <clear_msr_range+1>: xor    %edi,%edi
0xc0110d36 <clear_msr_range+3>: push   %esi
0xc0110d37 <clear_msr_range+4>: push   %ebx
0xc0110d38 <clear_msr_range+5>: mov    0x14(%esp,1),%ebx
0xc0110d3c <clear_msr_range+9>: mov    0x10(%esp,1),%esi
0xc0110d40 <clear_msr_range+13>:        cmp    %ebx,%edi
0xc0110d42 <clear_msr_range+15>:        jae    0xc0110d54
<clear_msr_range+33>
0xc0110d44 <clear_msr_range+17>:        xor    %eax,%eax
0xc0110d46 <clear_msr_range+19>:        lea    (%edi,%esi,1),%ecx
0xc0110d49 <clear_msr_range+22>:        mov    %eax,%edx
0xc0110d4b <clear_msr_range+24>:        wrmsr
0xc0110d4d <clear_msr_range+26>:        add    $0x1,%edi
0xc0110d50 <clear_msr_range+29>:        cmp    %ebx,%edi
0xc0110d52 <clear_msr_range+31>:        jb     0xc0110d46
<clear_msr_range+19>
0xc0110d54 <clear_msr_range+33>:        pop    %ebx
0xc0110d55 <clear_msr_range+34>:        pop    %esi
0xc0110d56 <clear_msr_range+35>:        pop    %edi
0xc0110d57 <clear_msr_range+36>:        ret

Kernel/HW/something else?

Re: Trouble running oprofile with 2.6.8 & Gentoo

From: John Levon <levon@mo...> - 2004-08-24 16:54

On Mon, Aug 23, 2004 at 06:37:30PM -0700, Zarakin wrote:

> > Also, can you safely boot with nmi_watchdog=2 on this machine?
> 
> No - I got an another Oops in "clear_msr_range" - EIP at <c0110d4b>
> 
> Dump of assembler code for function clear_msr_range:
> 0xc0110d33 <clear_msr_range+0>: push   %edi
> 0xc0110d34 <clear_msr_range+1>: xor    %edi,%edi

Please report this part of the bug to linux-kernel@...

Either there's some general bug or your machine needs blacklisting.

regards
john

Re: Trouble running oprofile with 2.6.8 & Gentoo

From: Zarakin <zarakin@ho...> - 2004-08-25 06:31

> > No - I got an another Oops in "clear_msr_range" - EIP at <c0110d4b>
> >
> > Dump of assembler code for function clear_msr_range:
> > 0xc0110d33 <clear_msr_range+0>: push   %edi
> > 0xc0110d34 <clear_msr_range+1>: xor    %edi,%edi
>
> Please report this part of the bug to linux-kernel@...
>
> Either there's some general bug or your machine needs blacklisting.


"Intel removed MSR 0x3ba/0x3bb (MSR_IQ_ESCR0 and 1) in prescott processor
(family 15 model 3). I'm going to sleep, if nobody beat me I'll try to
provide a patch, see nmi.c:setup_p4_watchdog() --> clear_msr_range(0x3A0,
31);

This probably break oprofile too, patch will be a bit less obvious"
[Mail-list ref:
http://marc.theaimsgroup.com/?l=linux-kernel&m=109341444121606&w=2 ]