Maintained by: NLnet Labs

[Unbound-users] Unbound 1.5.3rc1 pre-release

W.C.A. Wijngaards
Wed Mar 4 08:15:41 CET 2015


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Brad, Heiner,

On 03/04/2015 03:20 AM, Brad Smith wrote:
> On 03/03/15 16:42, Heiner Kallweit wrote:
>> Am 03.03.2015 um 15:51 schrieb W.C.A. Wijngaards:
>>> Hi,
>>> 
>>> Unbound 1.5.3rc1 maintainers prerelease is available: 
>>> http://www.unbound.net/downloads/unbound-1.5.3rc1.tar.gz sha1
>>> 8a440a7e9c65da89465d0740254b8c955b83a91a sha256
>>> 9dbe3a89e4f8a16eb55f0dd6634f5a1112fc05303ca466056fbdfc84ad9bc98f
>>>
>>>
>>> 
This release fixes an issue where reload (sighup) could cause unbound
>>> to exit, because of permission acquisition changes inserted in
>>> 1.5.2.
>>> 
>>> Bug Fixes: - - [bugzilla: 647 ] Fix #647 crash in 1.5.2 because
>>> pwd.db no longer accessible after reload. - - [bugzilla: 645 ]
>>> Fix #645 Portability to Solaris 10, use AF_LOCAL. - -
>>> [bugzilla: 646 ] Fix #646 Portability to Solaris, -lrt for 
>>> getentropy_solaris. - - Use the getrandom syscall introduced in
>>> Linux 3.17 (from Heiner Kallweit).
>>> 
>>> Best regards, Wouter
>> With regard to the proposed getrandom syscall patch Brad Smith
>> sent this comment to me and the mailing list: "The getentropy()
>> code for Linux within the Unbound tree is old. The upstream code
>> from the OpenBSD tree has already dealt with using the new system
>> call 6 months ago. So Wouter just needs to re-sync with the code
>> from the OpenBSD tree." Did you consider this? Or would it simply
>> have caused too much regression testing effort and you preferred
>> a small extension for now?

I did not want portability failures.  It would get silly with the fix
to an issue requiring another fix and so on.  The getrandom code, and
also the unix-socket code (a recent change with nothing to do with the
getrandom code) are causing portability issues.

> I sent a diff to Wouter to sync in those changes from upstream but 
> haven't heard anything back from him. I also attached the diff
> here for you Heiner.

I'll pick it up for next version.  Thanks for the diff :-)

Best regards,
   Wouter

> Note: I don't use Linux so if you can test that would probably be 
> quite useful to Wouter.
> 
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1

iQIcBAEBAgAGBQJU9rEdAAoJEJ9vHC1+BF+N/KAP/0/sjGHJVeetMslr/5F947kx
Rt0CCpYQ1rHe/Qbi9wfDtHKH4mKGjajPNBT8ZGukYuyxYihqgahEUlz8vPcyf92a
/TgZBiHio1lH+vHx6yrso5MeFey0Gr3aP//X6bJHZFwD+PS59TUFQ33zPdTv1XlO
CnZzkhuV//GAtzvq/8Ztn0g8tKwdcIU+xqzT/wC57gPbXyd3COf9I2eHJahpb05A
TrAMkN2Ztuw0y5X24zaDyEt4GotnQ28uABmeTJpvqy0+PDMeWEwcKOk2GzlXOQWY
ncd3l+XrCUlu3OuS5P1cm7pYe9vxeY/jsK/4+Hb4N9CdxCnSLnsYeBOg9scEEqTH
j4f5p5agzlogHL931kUC+QHGtn/QY0GhFhIUIA2X4wD+/z2qcvsEgHGeFeq2Du5x
xK7omKg65EOuSGxsUh1NRspUr7SlbYmuDA7bxw6VPQCdX7gJ4elYFXxFrLA9NDf6
zYQzvle7jVy/e1cpzmZzSZI8SY0Ml3m1YHKNbevJAwTPdxJFwZwoS4AF5r2PS+VX
bBhfaICGiFAwC5O2i6UVce35bw9xQlVwqCWaYyZ6yJxT/BiAvqdS9b+wjFZKg887
h3drVuzyiOn78qv9j7lMvt1WF5YZ/ph5JS0z1rE0wU1vrhs2AvISAXU8qfzzqcCm
k7y2JS37+cl8utM07P1B
=hru6
-----END PGP SIGNATURE-----