[DGD] objregwatch flagged

[Raymond Jennings]

The bug apparently manifested itself in part by corrupting the symbol table
so that calls to _Q_next and _Q_prev got deranged, which in turn tripped
the watcher when the klib fed it the corrupted information.

Naturally, given my history with DGD, a segfault was just around the
corner, and the core dump showed obviously bogus data.


On Sun, Jul 21, 2013 at 9:10 PM, Raymond Jennings <shentino at gmail.com>wrote:

> Well it turns out that it was a false alarm.
>
> Turns out there was an object upgrade bug in DGD that Felix said may have
> had something to do with this problem.
>
> He's committed a fix to the repo.
>
> Since it appears the Klib was working fine all along I'm closing this bug
> report.
>
>
> On Mon, Jul 8, 2013 at 7:26 PM, Raymond Jennings <shentino at gmail.com>wrote:
>
>> hit enter too soon again.  Thanks to felix for the logging idea.
>>
>> I've been sleepless for a few days and on a major coffee binge due to a
>> real life crisis.
>>
>>
>> On Mon, Jul 8, 2013 at 7:06 PM, Raymond Jennings <shentino at gmail.com>wrote:
>>
>>> As for the original problem, since the only times I've been able to
>>> reproduce it are when I was doing something funny with dgd.
>>>
>>> The code looks solid in both my fork and dworkin's original version, so
>>> I don't know what is going on.
>>>
>>> I'm considering the bug unreproducible for now, and adding some logging
>>> bits to see if I can catch it again, but in the meantime I'll settle for
>>> warning that something is broken somehwere and it's probably my fault.
>>>  It's unacceptably vague, but it's the best I can do.
>>>
>>>
>>> On Mon, Jul 8, 2013 at 5:30 PM, Raymond Jennings <shentino at gmail.com>wrote:
>>>
>>>> I do maintain my fork as a whole by default though, in that if
>>>> something is broken in the original klib and nobody else has the time to
>>>> fix it, I probably will wind up fixing it anyway and offering the patch to
>>>> anyone who wants it, including felix.
>>>>
>>>> Since there are multiple repositories maintained by different people,
>>>> I'm not really sure if anyone could be called the "official" maintainer for
>>>> the kernel library.
>>>>
>>>> Then again, that's one of the beautiful things about open source anyway.
>>>>
>>>>
>>>
>>
>