Re: Another 6.99.31 amd64 panic

[Chavdar Ivanov <ci4ic4%gmail.com@localhost>]

Well, I did the 'git bisect' dance yesterday, to no avail. I never
checked first if 'trunk' was bad, I assumed it as it was from around
the 13th of February, my problems started earlier. I stated 'git
bisect good' from around 25th of January and proceeded with the
process, every time building a clean kernel.

All of them worked just fine.

The overnight build (from the main CVS repo) completed fine. The crash
was gone. Same from the overnight build from releng. The releng kernel
from 24th of February still crashes, so all we can say now is that a
commit between 24th and 26th has sorted it out, and it is not some
transient hardware fault.

Unfortunately I am none the wiser. All is good, though. At least I
learned something about git, which should be useful other time.

Thanks,

Chavdar Ivanov





On 24 February 2014 15:25, Chavdar Ivanov <ci4ic4%gmail.com@localhost> wrote:
> On 24 February 2014 11:43, Nick Hudson <skrll%netbsd.org@localhost> wrote:
>> On 02/24/14 11:17, Chavdar Ivanov wrote:
>>>
>>> I could, but I am not sure about the exact cvs syntax - I usually
>>> blindly update to the latest. It should't take long, as the kernel
>>> from 26th of January works fine. If you would give me a hint how to
>>> specify the date, I'll do it today.  I went through the manual and the
>>> FAQ quickly, but couldn't find the exact date spec to fetch.
>>
>>
>> Hi,
>>
>> I found out recently that the git copy on github is easier to work with when
>> doing bisects.
>>
>>     git clone https://github.com/jsonn/src.git
>>
>> Use git log to find a point that was last known working
>>
>>     git bisect start
>>     git bisect bad [change from above]
>>
>> etc,
>>
>> As it wasn't so long ago maybe the cvs version isn't so bad
>>
>>     cvs update -D 20140126
>>
>> Will get you back to Jan 26th. Finding half way points between dates hurts
>> my head :)
>>
>> Nick
>
> Thanks, I'll do that later.
>
> Chavdar
>
>
>
>
> --
> ----



-- 
----