Currently, samba built-in DNS client is UDP-only (it does
support TCP mode, but not all components use it). In
particular, when winbind queries list of AD DCs (SRV
record) which can be quite large, it uses UDP-only query.
And at the same time, samba DNS client does not support
EDNS0 at all, so the reply is limited to standard DNS
packet size which is 512 bytes.
Add minimal EDNS0 OPT record to ADDITIONAL section when
sending a DNS request out, indicating we can accept
DNS packets up to 4Kb in size.
Since DNSSEC is in wide use today and DNSSEC requires
EDNS0, it is okay to assume ENDS0 is widely supported
these days, so there should be no regressions when
enabling EDNS0.
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Gbp-Pq: Name edns0.patch
As has been said several times in
https://bugzilla.samba.org/show_bug.cgi?id=13622 ,
the check 'bandsize > SIZE_MAX/nbands' is useless. But it
is also wrong, in 2 ways: first, nbands might be 0 (when
no bands has been allocated yet), and second, there's no
point in comparing this with SIZE_MAX, since size_t on 32bit
platforms is a 32bit integer, while bandsize is off_t which
is 64bits (samba always enables LFS).
This check causes the module to fail when bandsize*nbands
exceeds 32bits, which has been reported for example at
https://bugs.debian.org/974868 .
Whole thing can't overflow because it is already guarded
by time_machine_max_size. Or at the very least, by current
disk sizes... :)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Gbp-Pq: Name fruit-disable-useless-size_t-overflow-check.patch
Lots of samba libraries has incomplete dependencies listed
in wscript files. This usually is not a problem since the
link line includes dependencies of their dependencies of
their dependencies, and somewhere down that line all immediate
dependencies which are missing are actually present. But
sometimes this becomes a problem when a library does not
declare direct dependency on at least one private library
which it actually uses: in case no private library is
listed as direct dependency, private library directory is
not put into RUNPATH of the resulting binary, so the binary
can not find its own dependencies.
Fix a few such places, including one library which is a part
of public abi (libsmbldap).
Michael Tokarev [Tue, 26 Apr 2022 13:14:38 +0000 (16:14 +0300)]
testparm: do not fail if /run/samba does not exist
testparm explicitly fails if $piddir or $lockdir does not exist.
However, the daemons which actually use these directories, will
create it on demand, there is no need to fail even simple testparm
operations if the dirs are not there.
This change lets to (pre)configure samba without bothering to
pre-create the directories which are overwise needed only to fulfil
testparm criteria.
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
Gbp-Pq: Name testparm-do-not-fail-if-pid-dir-does-not-exist.patch
Michael Tokarev [Fri, 8 Apr 2022 08:50:21 +0000 (11:50 +0300)]
disable setuid configure checks
For some strange reason, when running reprotest test on salsa-ci,
which apparently is running as root, - on the *second* build only
the configure fails (after successfully built package the first
time). The configure test tries to change gid and verifies it
actually changed (not that the syscall exist), - and that fails.
Since it is extremely uncommon to configure the build process as
root, salsa-ci test environment details are quite deep down the line,
and we know the syscall actually works, just disable the probe,
pretending we are not root.
lib/replace/replace.h header defines ZERO_STRUCT macro
which uses memset_s() function (which is similar to
memset() but can not be optimized out by the compiler).
Glibc has bzero() with similar property, while memset_s()
have is implemented in lib/replace/replace.c, - this way,
some binaries needlessly link with libreplace-samba4 just
to get rep_memset_s() symbol. By using bzero() instead,
this endless linkage is eliminated, so we can package,
for example, libldb (which uses ZERO_STRUCT) without it
linking to libreplace-samba4.
Note: actually using explicit_bzero() so it is not optimized
out by the compiler - this is the original goal of using
memset_s().
CTDB uses /var/log/ctdb/ directory for the default log files. With
syslog disabled, systemd journal is not able to correctly inform
errors happening during service initialization.
Upstream community creates generic config files to be used by different
distributions, so this change makes no big difference to be accepted by
upstream.
With this patch the end user will be able to identify initialization
errors by executing:
systemctl status ctdb.service
or to follow ctdb logs by executing:
journalctl -f -u ctdb
Signed-off-by: Rafael David Tinoco <rafaeldtinoco@ubuntu.com>
-
Author: Rafael David Tinoco <rafaeldtinoco@ubuntu.com>
Bug-Debian: https://bugs.debian.org/929931
Bug-Ubuntu: https://bugs.launchpad.net/bugs/722201
Last-Update: 2022-03-24
Gbp-Pq: Name ctdb-config-enable-syslog-by-default.patch
Upstream defines nfs related service names based on the Linux
distribution. This patch fixes the names for Debian and derivatives.
Update by Andreas Hasenack <andreas@canonical.com> (LP: #1961840):
Use nfsconf(8) if it's available, instead of parsing the old config
files in /etc/default/nfs-*
Jeroen Dekkers [Wed, 8 May 2024 10:23:57 +0000 (13:23 +0300)]
Add so version number to private libraries for dpkg-shlibdeps
We also want dpkg-shlibdeps to generate correct dependency information
for the private libraries in our binary packages, but dpkg-shlibdeps
only works when the library has a version number.
Michael Tokarev [Thu, 3 Nov 2022 17:49:33 +0000 (20:49 +0300)]
hurd compatibility changes
Hurd does not define PIPE_BUF, so lib/tevent/testsuite.c fails to compile
(yes, this file is used as part of *samba* testsuite, not tevent testsuite).
Define it to a safe minimal value like 512 bytes.
Hurd does not provide SA_NOCLDWAIT define, so lib/util/tests/tfork.c does
not compile. This is only needed during testing to omit zombie process
generation, which has only cosmetic effect. Define it to be 0.
Michael Tokarev [Sat, 2 Apr 2022 16:01:55 +0000 (19:01 +0300)]
ensure libsmbclient.h is being used with LFS enabled
Bug-Debian: https://bugs.debian.org/221618
We build samba with LFS (Large File Support) even on 32bits.
This means some types like off_t are 64-bit wide, again,
even on a 32bit host. libsmbclient.h uses off_t in function
prototypes, and thes prototypes muct match those which were
used at samba compile time - if some other source includes
libsmbclient.h without LFS, it'll get wrong prototypes and
the resulting binary will most likely crash when using
libsmbclient functions.
Detect and error-out this at compile time.
We can not do anything with this in the public header since
it is alredy too late to redefine things, since we can't
guarantee we're the first header a program #includes, and
at the time this libsmbclient.h is included, off_t can
already be defined so our (re)define of _FILE_OFFSET_BITS
does nothing already.
Patching libsmbclient.h to use off64_t means client program
should change their off_t to off64_t too when storing
file offsets returning from libsmbclient, so this is not
an option too.
With this change, we will error out even if the user source
does not use any off_t-related functions. Namely, it was ok
to #include <libsmbclient.h> and use smbc_open/smbc_read/
smbc_write/smbc_close without _F_O_B=64, - neither of these
functions uses off_t. smbc_lseek and others doesn't work,
but if a program does not use them anyway, whole thing will
just work even without enabling LFS. Ideally we can probably
check each individual function which is being affected, by
replacing it with #error if sizeof(off_t) < 8. But this
requires quite some hackery...
projects / samba.git / log