The situation is this: even when we --disable-gtk-doc, if gtk-doc is
actually installed at configure stage, tools like gtkdoc-rebase will be
picked up and run during the install target. That is bad because the
gtk-doc package may have been removed by then, especially during dpb(1)
bulks (we explicitely disable support for it so why should it stay...).
So for now, let's add the following env to configure whenever we use
--disable-gtk-doc, until a bettersolution is found...
CONFIGURE_ENV +=ac_cv_path_GTKDOC_CHECK="" \
ac_cv_path_GTKDOC_REBASE="" \
ac_cv_path_GTKDOC_MKPDF=""
An issue with the use of lease times was found and fixed. Making
certain changes to the end time of an IPv6 lease could cause the
server to abort. Thanks to Glen Eustace of Massey University,
New Zealand for finding this issue.
Changes in version 0.2.2.39 - 2012-09-11
Tor 0.2.2.39 fixes two more opportunities for remotely triggerable
assertions.
o Security fixes:
- Fix an assertion failure in tor_timegm() that could be triggered
by a badly formatted directory object. Bug found by fuzzing with
Radamsa. Fixes bug 6811; bugfix on 0.2.0.20-rc.
- Do not crash when comparing an address with port value 0 to an
address policy. This bug could have been used to cause a remote
assertion failure by or against directory authorities, or to
allow some applications to crash clients. Fixes bug 6690; bugfix
on 0.2.1.10-alpha.
No CVEs for these vulnerabilities yet.
https://kb.isc.org/article/AA-00778
If a record with RDATA in excess of 65535 bytes is loaded into a
nameserver, a subsequent query for that record will cause named to exit
with an assertion failure.
This vulnerability can be exploited remotely against recursive servers
by inducing them to query for records provided by an authoritative
server. It affects authoritative servers if a zone containing this type
of resource record is loaded from file or provided via zone transfer.
with no objections. It relies on a GUI toolkit which hasn't been updated
in 10 years, needs to run as root in order to get tcpdump to parse
capture files, and even then it still doesn't work.
* Fix warnings reported by clang.
* Using -1 is the same as 0, except older libpcap left 0 undefined.
* Fixed non-conflict.
* Always process all waiting packets.
Tested on amd64.
Ok benoit@ (maintainer)
* Improved HTTPS cipher handling and added support for chained certificates.
* Allow the source password to be undefined. There was a corner case,
where a default password would have taken effect. It would require the
admin to remove the 'source-password' from the icecast config to take
effect. Default configs ship with the password set, so this
vulnerability doesn't trigger there.
* Prevent error log injection of control characters by substituting
non-alphanumeric characters with a '.' (CVE-2011-4612). Injection
attempts can be identified via access.log, as that stores URL encoded
requests. Investigation if further logging code needs to have
sanitized output is ongoing.
Tested on amd64.
Reads fine aja@
