[DGD] 1.2.133

[Felix A. Croes]

bart at wotf.org wrote:

> Works well for me, but turned up something funny.
>
> As it turns out, there was a condition in which my kernel did a dump_state()
> without doing a swapout() first. That results in a corrupt dumpfile and is
> something I'm sure you metion somewhere (I recall reading about it). Makes you
> wonder why this wasn't a problem earlier on, well, I triggered the condition
> while trying the quick reboot feature, and forced a statedump, something which
> is normally not done manually.
>
> It made me wonder however why dump_state() doesn't do this itself if its
> basicly mandatory. I could quite see why it shouldn't if there were legitimate
> uses of dump_state() without a swapout(), but I can't come up with one :)

Creating a state dump does not require swapout(), though it used to
be the case that you periodically had to do a swapout to allow DGD to
perform full garbage collection (which is now fully automatic).


> So, am I again missing something here?

You have found a bug, either in DGD or in your own changes.  My own
state dumps seem to work okay without swapouts.

Regards,
Dworkin