[PATCH] LMDB full patch set

Stefan Metzmacher metze at samba.org
Thu May 3 14:46:51 UTC 2018

Am 03.05.2018 um 07:45 schrieb Andrew Bartlett via samba-technical:
> On Thu, 2018-04-12 at 17:48 +0200, Stefan Metzmacher wrote:
>> Hi Andrew,
>> I was also not able to understand if the getpid() based detection for
>> the fork case correct. What is the correct way for lmdb to cleanup
>> after fork? Is close(fd) in lmdb_pvt_destructor() really the only
>> thing? Doesn't lmdb has other state, which we will leak?
>> Thanks for your patience, but there's still a bit of work required,
>> sorry! But given that will be the core of our AD database careful review
>> is required.
> Is it correct to say that as long as we have the same behaviour, that
> is either allowing transparent (when not locked) use across a fork(),
> or denying it for both tdb and mdb, that you are OK with the patches
> now?

Likely:-) But I would need to swap it into my brain again...

> I'm happy with your squashes in general, except around the tests where
> I need to look much more carefully at what is being done there (it was
> fairly deliberately constructed, so I need to double-check that).  

Please check it and maybe already do the squashes.

> This, and the fork() behaviour aside, are you happy with the tree
> otherwise? 

I've lost track, can you rebase the different stages on top of
master, so that I can see what's actually remaining.

I'll try to have another look when you give me the pointers to
rebased branches.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: OpenPGP digital signature
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20180503/4058c14f/signature.sig>

More information about the samba-technical mailing list