Re: linux 2.6.22.7

<karthikeyan.s> writes:

> 
> 
> Hi,
> Had some doubts in your comments.>> ** (process:25057): WARNING **: Unknown
facility 145 (0x91) in tracefile>> /control/interrupts>>>> ** (process:25057):
WARNING **: Unknown facility 211 (0xd3) in tracefile
> >> /cpu>>>> ** ERROR **: file event.c: line 86 (ltt_event_facility): assertion
failed:>> (facility->exists)
> 
> >Could be caused by the "unable to register probe" messages, caused
by>incorrect CONFIG_SMP ifdefs in probes.
> I did not get what you mean by CONFIG_SMP ifdefs in probes. There is no such
thing in those files(ltt/probes/).
> I have been trying root cause this. I noticed that the viewer gets the
> facility id from an offset in the trace file. And the warning comes
> because the viewer does not recognise the event (the value range for
> the events is 0 to 5 (or 4) and it gets 168(0xa8) for facility net)
> Could it be that the problem is there in lttv( it got the wrong offsets
> of the trace data ? ). Or can lttv error be ruled out and just focus on
> the kernel side? I also looked at 0.9.6 and i did not find CONFIG_SMP
> ifdefs.
> And yes, what happens when the CONFIG_SMP ifdef is incorrect?
> I am not bothered much about the buffer overflow issue right now, but i think
i can take care of it by allocating more buffers.
> 
> 
> 
> 
> _______________________________________________
> Ltt-dev mailing list
> Ltt-dev <at> listserv.shafik.org
> http://listserv.shafik.org/mailman/listinfo/ltt-dev
> 
Hi all, 
    I am also facing the same problem as karthikeyan faced. I tried to apply the
patch manually 
in file ltt/probe/ltt-probe-kernel.c
#ifdef CONFIG_SMP
                result = _marker_set_probe(probe_array[eID].flags,
                                probe_array[eID].name,
                                probe_array[eID].format,
                                ltt_trace, &probe_array[eID]);
                if (!result)
                        printk(KERN_INFO "LTT unable to register probe %s\n",
                                probe_array[eID].name);
#endif
 If i put this line 
LTT unable to register probe kernel_sched_wait_task
LTT unable to register probe kernel_sched_migrate_task
and 
LTT : cpu : 22954 events lost in cpu channel (cpu 0).
this error has gone but then gcc complain that result is not used at all. It may
mean that this probes are not used(I am not sure). I don't know weather it is
correct or not.

 and will running the lttv -m textdump -t trace/control  i got the following error.
** (process:14833): CRITICAL **: cannot open trace trace/control/
Segmentation fault


Thanks 
Bhavesh