[ltt-dev] lttv crashed in tracefile.c:1582

Dany Madden dany4madden at gmail.com
Wed Mar 31 17:12:20 EDT 2010


On Wed, Mar 31, 2010 at 1:56 PM, Mathieu Desnoyers <
compudj at krystal.dyndns.org> wrote:

> * Mathieu Desnoyers (compudj at krystal.dyndns.org) wrote:
> > * Dany Madden (dany4madden at gmail.com) wrote:
> > > On Wed, Mar 31, 2010 at 10:24 AM, Mathieu Desnoyers <
> > > compudj at krystal.dyndns.org> wrote:
> > >
> > > > * Dany Madden (dany4madden at gmail.com) wrote:
> > > > > Hello,
> > > > >
> > > > > Please take a look at this lttv error with the mainline kernel
> > > > > 2.6.32.9 on ppc64.
> > > > >
> > > > > lttng: patch-2.6.32.9-lttng-0.198
> > > > > lttv-0.12.30-02102010
> > > > > ltt-control-0.79-01022010
> > > > >
> > > > > LTTV crashed with this error:
> > > > >
> > > > > ERROR:tracefile.c:1582:ltt_update_event_size: assertion failed:
> (info !=
> > > > NULL)
> > > > > /usr/bin/lttv: line 15:  2696 Aborted                 $0.real $*
> > > > >
> > > > >
> > > > > CONFIG_LTT=y
> > > > > CONFIG_LTT_FILTER=m
> > > > > # CONFIG_HAVE_LTT_DUMP_TABLES is not set
> > > > > CONFIG_LTT_RELAY=y
> > > > > CONFIG_LTT_RELAY_LOCKLESS=y
> > > > > # CONFIG_LTT_RELAY_IRQOFF is not set
> > > > > # CONFIG_LTT_RELAY_LOCKED is not set
> > > > > CONFIG_LTT_SERIALIZE=y
> > > > > CONFIG_LTT_FAST_SERIALIZE=y
> > > > > CONFIG_LTT_TRACEPROBES=m
> > > > > CONFIG_LTT_TRACE_CONTROL=m
> > > > > CONFIG_LTT_TRACER=y
> > > > > # CONFIG_LTT_ALIGNMENT is not set
> > > > > CONFIG_LTT_CHECK_ARCH_EFFICIENT_UNALIGNED_ACCESS=y
> > > > > # CONFIG_LTT_DEBUG_EVENT_SIZE is not set
> > > > > CONFIG_LTT_USERSPACE_EVENT=m
> > > > > CONFIG_LTT_VMCORE=y
> > > > > CONFIG_LTT_STATEDUMP=m
> > > > > CONFIG_LTT_KPROBES=m
> > > > > # CONFIG_LTT_ASCII is not set
> > > > >
> > > > >
> > > > > > modprobe ltt-trace-control
> > > > > > modprobe ipc-trace
> > > > > > modprobe kernel-trace
> > > > > > modprobe mm-trace
> > > > > > modprobe fs-trace
> > > > > > modprobe syscall-trace
> > > > > > modprobe trap-trace
> > > > >
> > > > > > ltt-armall
> > > > > > lttctl -C -w /tmp/trace1 trace1
> > > > > > lttctl -D trace1
> > > > > > lttv -m textDump -t /tmp/trace1
> > > > > **
> > > > > ERROR:tracefile.c:1582:ltt_update_event_size: assertion failed:
> (info !=
> > > > NULL)
> > > > > /usr/bin/lttv: line 15:  2696 Aborted                 $0.real $*
> > > > >
> > > > > This post,
> > > >
> http://lists.casi.polymtl.ca/pipermail/ltt-dev/2010-February/002189.html,
> > > > > describes the same error on 32 bit system. By not having the
> > > > > LTT_ALIGNMENT set, shouldn't it just work in my case?
> > > > >
> > > > > Thanks in advance for your help!
> > > >
> > > > A few questions:
> > > >
> > > > - the trace is taken on a ppc64 machine. Is is viewed on a ppc64 too,
> or
> > > >  on a different arch size/endianness ?
> > > >
> > >
> > > Trace collected and viewed on the same ppc64 system.
> > >
> > >
> > > >
> > > > - Can you provide some information about the content of the trace
> > > >  directory ? (files, file size)
> > > >
> > >
> > > The files in the trace directory are all the same size??
> > >
> > > > /tmp/trace3 # ls -l
> > > total 2448
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 fs_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 fs_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 fs_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 fs_3
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 input_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 input_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 input_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 input_3
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 ipc_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 ipc_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 ipc_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 ipc_3
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 kernel_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 kernel_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 kernel_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 kernel_3
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 metadata_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 metadata_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 metadata_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 metadata_3
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 mm_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 mm_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 mm_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 mm_3
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 module_state_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 module_state_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 module_state_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 module_state_3
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 softirq_state_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 softirq_state_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 softirq_state_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 softirq_state_3
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 swap_state_0
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 swap_state_1
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 swap_state_2
> > > -rwxr-xr-x 1 root root 65536 2010-03-31 19:32 swap_state_3
> > >
> >
> > Oh, that's a pretty good hint. Although I tried to make everything as
> > generic as possible, I think LTTV assumes that the page size is 4k.
> > However, recent ppc64 kernels have a 64k page size. So I'm fairly sure
> > that this is where the problem comes from.
> >
> > Can you send me (privately) a tarball of your trace so I can see what I
> > can do with LTTV to support it ?
>
> OK, got it. Well, your metadata channels are all empty (looked with
> hexedit). So clearly LTTV cannot make sense of the data in the other
> channels.
>
> So let's try the simple solutions first: try to load all the
> LTTng-related modules you have (I see that a few were missing from your
> list), and only then run ltt-armall. If that does not work, try to
> compile LTTng as "built in".
>

I loaded all the ltt modules i can fine in /lib/modules/2.6.32.9 and saw a
different error now.

> lttv -m textDump -t /tmp/trace5

** ERROR **: Undescribed event 0 in channel irq_state
aborting...
/usr/bin/lttv: line 15:  3361 Aborted                 $0.real $*

> lsmod | grep trace
ltt_marker_control      6091  1 ltt_trace_control
ltt_trace_control      21767  0
ipc_trace               2559  0
kernel_trace           20861  0
mm_trace                6960  0
fs_trace                7557  0
syscall_trace           1739  0
trap_trace              1707  0
net_trace              13676  0
block_trace            12110  0
ltt_filter              1966  1 jbd2_trace
jbd2_trace              5880  0
lockdep_trace           3753  0
net_extended_trace      3241  0
pm_trace                2599  0
rcu_trace               2217  0

>  lsmod | grep ltt
ltt_marker_control      6091  1 ltt_trace_control
ltt_trace_control      21767  0
ltt_filter              1966  1 jbd2_trace
ltt_statedump         124464  1 ltt_kprobes
ltt_kprobes            11857  0
ltt_userspace_event     2911  0


>
> Thanks,
>
> Mathieu
>
> >
> > Thanks,
> >
> > Mathieu
> >
> > >
> > > >
> > > > Also, trying to enable "CONFIG_LTT_DEBUG_EVENT_SIZE" might help by
> > > > providing extra self-check on trace metadata consistency.
> > > >
> > > > Rebuilt the kernel with CONFIG_LTT_DEBUG_EVENT_SIZE=y , recollected
> the
> > > trace and viewed it. I still see the same error and did notice anything
> else
> > > different than it was before.
> > >
> > >
> > > > Thanks,
> > > >
> > > > Mathieu
> > > >
> > > > >
> > > > > Dany Madden
> > > >
> > > > > _______________________________________________
> > > > > ltt-dev mailing list
> > > > > ltt-dev at lists.casi.polymtl.ca
> > > > > http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
> > > >
> > > >
> > > > --
> > > > Mathieu Desnoyers
> > > > Operating System Efficiency R&D Consultant
> > > > EfficiOS Inc.
> > > > http://www.efficios.com
> > > >
> >
> > --
> > Mathieu Desnoyers
> > Operating System Efficiency R&D Consultant
> > EfficiOS Inc.
> > http://www.efficios.com
>
> --
> Mathieu Desnoyers
> Operating System Efficiency R&D Consultant
> EfficiOS Inc.
> http://www.efficios.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.casi.polymtl.ca/pipermail/lttng-dev/attachments/20100331/8ec28a5c/attachment-0003.htm>


More information about the lttng-dev mailing list