Rafael.Vanoni at Sun.COM wrote:

> Li, Aubrey wrote:
>> Hi Rafael,
>> 
>> Rafael.Vanoni wrote:
>> 
>>> Li, Aubrey wrote:
>>>> Rafael Vanoni wrote:
>>>> 
>>>>> The problem was that the xcalls probe fires N times within
>>>>> xc_common, and the script was zeroing the variable with the
>>>>> address of the xcall function after the first one.
>>>>> 
>>>>> This export contains all the changes from the previous with this
>>>>> issue fixed. Let me know what you think.
>>>>> 
>>>> It looks like dtp_events.c is missing in your patch.
>>>> Can you add it and re-send the patch?
>>>> 
>>>> Thanks,
>>>> -Aubrey
>>> Ah, sorry about that. Here's the export that fixes the xcall problem
>>> plus a diff for the --cpu option. You'll need to import the
>>> patch before
>>> applying the diff.
>>> 
>> 
>> 7770 looks good except to keep the same behavior as the previous
>> version, I think we need to add predicate "/pid != $pid/" to either
>> "fbt::xc_common:entry" or "sysinfo:::xcalls". 
>> 
>> What do you think? If you agree with that, please commit 7770 with
>> this slight change. 
> 
> How about adding the predicate to the default (and -c option) but not
> for verbose mode ? 

That makes sense, and if you think it's necessary. But I think it's a bit 
complicated
and not worth doing.

> 
> BTW, I found a bug in the sparc script, turns out
> fbt::init_mondo:entry
> doesn't exist. I'll have it fixed in the morning PST time.
> 
Well, you have to be on your own about the SPARC implementation, :-)

Thanks,
-Aubrey

Reply via email to