[ltt-dev] Problems with merging patch-2.6.30-rc2-lttng-0.126

[gregory.haskins@gmail.com (Gregory Haskins)]

Mathieu Desnoyers wrote:
> * Gregory Haskins (ghaskins at novell.com) wrote:
>   
>> Mathieu Desnoyers wrote:
>>     
>>> * Gregory Haskins (ghaskins at novell.com) wrote:
>>>   
>>>       
>>>> Mathieu Desnoyers wrote:
>>>>     
>>>>         
>>>>> * Gregory Haskins (ghaskins at novell.com) wrote:
>>>>>   
>>>>>       
>>>>>           
>>>>>> Hi all,
>>>>>>   I am trying to apply 2.6.30-rc2-lttng-0.126 to a clean 2.6.30-rc2 and
>>>>>> I am getting lots of rejects.  I went through one time and resolved all
>>>>>> the rejects, but the resulting kernel did trace properly so I perhaps
>>>>>> had some fuzz issues.  What kernel base should I be using for this series?
>>>>>>
>>>>>>     
>>>>>>         
>>>>>>             
>>>>> Hrm, weird, I may have messed up the packaging. I just released LTTng
>>>>> 0.127 which should not have such conflicts.
>>>>>   
>>>>>       
>>>>>           
>>>> Hi Mathieu,
>>>>  I am seeing more or less the same thing with 0.127.  The series has a
>>>> ton of conflicts against 30-rc2 and even itself.  Its as if the patches
>>>> are based on a different tree. 
>>>>
>>>> Its probably operator error, but I cannot see what I am doing wrong. 
>>>> Its the same procedure I have always used to bring your tree in.
>>>>
>>>>     
>>>>         
>>> Hmmmmm... it works here... (just tested it again to make sure)
>>>
>>> Can you give me the exact sequence of operation you use to apply the
>>> patchset ? (getting kernel.org tarballs/git, unpacking the lttng
>>> patchset, using quilt...)
>>>   
>>>       
>> Sure, I am using git as the base, and then stgit "import" to pull the
>> series in.  Here is a dump of my history from my current attempt:
>>
>>  1070  git clone -l linux-2.6 linux-2.6-ltt
>>  1071  cd linux-2.6-ltt/
>>  1072  wget
>> http://ltt.polymtl.ca/files/lttng/patch-2.6.30-rc2-lttng-0.127.tar.bz2
>>  1073  tar -jxvf patch-2.6.30-rc2-lttng-0.127.tar.bz2
>>  1074  git branch
>>  1075  git checkout -b master v2.6.30-rc2
>>  1076  git log
>>  1077  stg init
>>  1078  stg import --ignore -s patch-2.6.30-rc2-lttng-0.127/series
>>
>> I have confirmed that Linus' 2.6.30-rc2 tag is the ancestor that
>> immediately precedes the first LTT patch.  I have used this same
>> procedure for years successfully, including working with LTT.  I'm sure 
>> I am doing something wrong, but I can't see it.
>>
>>     
>
> Nah, it's my bad. I always check&refresh patches with "fuzz", but I
> forgot to refresh them this time. Can you try with LTTng 0.128 ?
>
>   
FYI: I had a build error in 0.126 through 0.128 where the
smp_processor_id call in include/sched_events_types.h was undefined with
the attached config.

Reverting "LTTng instrumentation scheduler fix task migration" fixed the
problem (though it left a somewhat ominous warning about the now changed
parameters to the trace-point which I didnt bother fixing).  Hopefully
this is enough of a hint about what is wrong, but ping me if you need
more info.

After I did this, I confirmed that 0.128 does indeed trace again! 
Thanks again, Mathieu.

-Greg

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ltt.config
URL: <http://lists.casi.polymtl.ca/pipermail/lttng-dev/attachments/20090423/45f48448/attachment-0002.asc>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 266 bytes
Desc: OpenPGP digital signature
URL: <http://lists.casi.polymtl.ca/pipermail/lttng-dev/attachments/20090423/45f48448/attachment-0003.pgp>