| Age | Commit message (Collapse) | Author |
|---|
|
debSystem uses a reference counted lock, so you can acquire it
multiple times in your applications, possibly nested. Nesting
locks causes a fd leak, though, as we only increment the lock
count when we already have locked twice, rather than once, and
hence when we call lock the second time, instead of increasing
the lock count, we open another lock fd.
This fixes the code to check if we have locked at all (> 0).
There is no practical problem here aside from the fd leak, as
closing the new fd releases the lock on the old one due to the
weird semantics of fcntl locks.
(cherry picked from commit 79f012bd09ae99d4c9d63dc0ac960376b5338b32)
(cherry picked from commit 1edcb718293f24ad190703a345f8f868b6e3bcc4)
LP: #1778547
|
|
(cherry picked from commit 4ba9412d545232794bd7838e0174df8fda4152b8)
LP: #1778551
|
|
|
|
We do want to get our autopkgtests triggered by dpkg uploads
in Ubuntu, but this does not happen because we don't have
an explicit dependency on it. Add one.
(cherry picked from commit b1989ab988cf4da47f979933e551b6dda694dcf2)
|
|
|
|
gpg2 generates keyboxes by default and users end up putting either those
or armored files into the trusted.gpg.d directory which apt tools
neither expect nor can really work with without fortifying backward
compatibility (at least under the ".gpg" extension).
A (short) discussion about how to deal with keyboxes happened in
https://lists.debian.org/deity/2017/07/msg00083.html
As the last message in that thread is this changeset lets go ahead
with it and see how it turns out.
The idea is here simply that we check the first octal of a gpg file to
have one of three accepted values. Testing on my machines has always
produced just one of these, but running into those values on invalid
files is reasonabily unlikely to not worry too much.
Closes: #876508
(cherry picked from commit 012932793ba0ea9398a9acd80593bed8e77cfbfc)
|
|
APT connects just fine to any .onion address given, only if the connect
fails somehow it will perform checks on the sanity of which in this case
is checking the length as they are well defined and as the strings are
arbitrary a user typing them easily mistypes which apt should can be
slightly more helpful in figuring out by saying the onion hasn't the
required length.
(cherry picked from commit f3e34838d95132e5f318e85525326decbfb19e36)
|
|
Closes: #879137
(cherry picked from commit f7281c4dbfe77f1d2a4da2f9e309104b0c86c45f)
|
|
tagfile-order.c: Add missing fields from dpkg 1.19
For binary packages, this is Build-Essential; for source packages,
it is Description.
test-bug-718329-...: Ignore control.tar.*, changes in dpkg 1.19
test-apt-extracttemplates: Fix for dpkg 1.19
(cherry picked from commit 404dececf913d3c09368a73ca00aa8172dbf6865)
|
|
(cherry picked from commit 91c90664eebb81bdc9fc228bc0b33eec5674753d)
|
|
1.5.y is the artful release series, so let's test it there.
|
|
|
|
In commit Do not warn about duplicate "legacy" targets, we
we added an if, that changed the .po files...
|
|
|
|
We now wait for being online ourselves, so all we need to wait
on is for services we are using to be online first. This avoids
severe boot slowdowns by other services having specified an
After=network-online.target without a Wants=.
Gbp-Dch: Full
|
|
|
|
This fixes the indentation of struct members, for
example, which were previously indented by 4 spaces.
Gbp-Dch: ignore
|
|
|
|
We do not actually test without these libraries, and it likely
would not build without them due to them being NOTFOUND and CMake
exiting with an error, so let's just mark them as required.
|
|
When udev is not available, the variables for libraries and
include paths are set to NOTFOUND rather than an empty string
and CMake exits with an error. Use a generator expression to
only use these variables when UDEV_FOUND is true.
|
|
|
|
We previously dlopen()ed it, but it seems painful to do that
without any real gain, except for possibly not having libudev
in the address space and not having code #ifdefed for Linux.
The latter means that we are a bit more likely to break stuff
for non-Linux systems now if we play with udev, but at least
we don't end up with it silently breaking because of a libudev
ABI break.
The existing function pointers in the struct were renamed and
kept for compat purposes.
Fixes Debian/apt#48
Also adjust prepare-release to strip [linux-any] from build-depends
for travis.
|
|
When writing a Sources files hashes that were already present
in the .dsc were always copied through (or modified), even if
disabled. Remove them instead when they are disabled, otherwise
we end up with hashes for tarballs and stuff but not for dsc
files (as the dsc obviously does not hash itself).
Also adjust the tests: test-compressed-indexes relied on Files
being present in showsrc, and test-apt-update-weak-hashes expected
the tarball to be downloaded when an archive only has MD5 and we
are requiring SHA256 because that used to work because the tarball
was always included.
Closes: #872963
|
|
This fails if no Files field exists anymore, for example, because
the Sources index only contains SHA256 hashes. Instead check all
hashes.
|
|
If a source has a legacy Contents file, and two lines mention
the same archive but different components, a warning would be
issued that is confusing. So, as the field is named Contents-deb-legacy,
let's just not print warnings for fields containing "legacy".
LP: #1697120
Closes: #839259
|
|
There's no real point in pulling it in in the timer already,
and it it somewhat saver to do so in the service.
|
|
Introduce a new helper, apt-helper wait-online that uses
NetworkManager and/or systemd-networkd to wait for them
reporting online, with a time out of 30 seconds; and run
that helper before running the daily update script.
LP: #1699850
Gbp-Dch: Full
|
|
The relevant calling code as well as the implementation for the deb
system was removed 2 years ago with the refactoring of release
information storage (b07aeb1a6e24825e534167a737043441e871de9f).
This commit removes the the unused remains of this change with no
practical effect on anybody (expect codesize) as the methods were
declared as hidden and hence only libapt could have called it.
Gbp-Dch: Ignore
|
|
A libapt user who hasn't initialized _system likely has a reason, so we
shouldn't greet back with a segfault usually deep down in the callstack
for no reason. If the user had intended to pick up information from the
system, _system wouldn't be uninitialized after all.
LP: #1613184
SRU: 1.4.y
|
|
Commit e250a8d8d8ef2f8f8c5e2041f7645c49fba7aa36 implemented the fix and
should have included already this testcase for it.
Gbp-Dch: Ignore
|
|
APT by default logs terminal (term.log) and actions (history.log), but
if either or Dir::Log directly is set to /dev/null it continues to do
so, which isn't too bad – just wasted effort – but term.log is
chmodded to protect it from the general public (as it may contain
otherwise private data the admin entired in the terminal) which
shouldn't happen for /dev/null.
|
|
gpgv: WARNING: This key is not suitable for signing in --compliance=gnupg mode
|
|
Closes: #874293
|
|
Closes: #874285
|
|
Closes: #873914
|
|
Seems we forgot to update the packaging when adding the manual
page. Once we have translations for it, we need to add them
as well...
Closes: #873934
|
|
|
|
It was broken because apt.conf.d was not readable, but that's
where the architecture is defined...
|
|
As a follow up to the last commit, let's replace APT_CONST
with APT_PURE everywhere to clean stuff up.
|
|
Functions marked with the const attribute may not inspect
any global memory. This includes targets of pointers or
references passed as arguments. A pure function however
is free to inspect memory, but may not have any side
effects.
The function StringSplit() was marked as const, but took
two references to strings. When the second one was passed
as a literal as in StringSplit(name, "::") the compiler
cleverly figured out that we only inspect the address of
"::" (since StringSplit is const) and thus optimized away
the "::" content.
While patching out individual broken uses of APT_CONST
would be possible, this is already the second case, and
there might be more, so let's redefine APT_CONST to use
the pure attribute, so we don't end up with the same
situation again in some time.
|
|
|
|
It contained raw text inside a refsect1
Gbp-Dch: ignore
|
|
The version is probably wrong for most, but oh well,
let's just pretend we are introducing them now.
|
|
We now require gcc 7 on the packaging side, and add an appropriate
symbol to our symbols file.
Also adjust prepare-release to ignore g++ version requirements
when setting up build dependencies on CI.
Closes: #871275
|
|
This caused a build failure in the test suite.
|
|
APT clients always noticed if a method isn't supported and nowadays
generate a message of the form:
E: The method driver …/foobar could not be found.
N: Is the package apt-transport-foobar installed?
This only worked if a single source was using such an unavailable method
through as we were registering the failed config the first round and
the second would try to send requests to the not started method, which
wouldn't work and hang instead (+ hiding the error messages as they would
be shown only at the end of the execution).
Closes: 870675
|
|
Regression-Of: 3317ad864c997f4897756c0a2989c4199e9cda62
|
|
If 'apt-ftparchive packages /path/to/files/' (or sources) is used the
files to include in the generated index (on stdout) were included in the
order in which they were discovered, which isn't a very stable order
which could lead to indexes changing without actually changing content
causing needless changes in the repository changing hashsums, pdiffs,
rsyncs, downloads, ….
This does not effect apt-ftparchive calls which already have an order
defined via a filelist (like generate) which will still print in the
order given by the filelist.
Note that a similar effect can be achieved by post-processing index
files with apt-sortpkgs.
Closes: 869557
Thanks: Chris Lamb for initial patch & Stefan Lippers-Hollmann for testing
|
|
The code only used to warn when it came into a situation where
something actually had to be forced. Warn directly after parsing
the command-line instead, that's more accurate.
|
|
|
