Some of our headers use APT_COMPILING_APT trickery to avoid exposing too
broadly details we don't want external clients to know and make use of.
The flip-side is that this can lead to different compilation units
seeing different definitions if they aren't all using the same config.
APT is not the place this information should be stored at, but it is a
good place to experiment and see what will be (not) needed in the future
for a proper implementation higher up the stack.
This is why "BarbarianArchitectures" is chosen instead of a more neutral
and/or sensible "VeryForeign" and isn't readily exported in the API to
other clients for this PoC as a to be drawn up standard will likely
require potentially incompatible changes. Having a then outdated and
slightly different implementation block a "good" name would be bad.
The functionality itself mostly exists (ignoring bugs) since the
introduction of MultiArch as we always had the risk of encountering
packages of architectures not known to dpkg (forced onto the system,
potentially before MultiArch) we had to deal with somehow and other
edge cases.
All this commit really does is allowing what could previously only be
achieved with editing sources.list and some conf options via a single
config option: -o APT::BarbarianArchitectures=foo,bar
The JSON encoder only looked at the top state, but did not
pop it, so if we nested objects, we got stuck in whatever
the last state we pushed aside was, so in our example, we
wrongly get a comma inserted _after_ key "b":
{"a":[{}],
"b":,[{}]
}
This allows us to correctly encode strings containing quotation
marks, escape characters and control characters.
The test case is a bit nasty because it embeds private-cachefile.cc
for linkage reasons.
MaybeAddAuth() here tells us that it refused to use the credentials
for an http source; but that caused the test suite to fail at a later
stage because we checked if there were any errors/warning. Strangely,
this is only triggered with LTO enabled.
Actually check that the warning is being set and then reject it.
Our configuration files are not security relevant, but having a parser
which avoids crashing on them even if they are seriously messed up is
not a bad idea anyway. It is also a good opportunity to brush up the
code a bit avoiding a few small string copies with our string_view.
strtoul(l) surprises us with parsing negative values which should not
exist in the places we use to parse them, so we can just downright
refuse them rather than trying to work with them by having them promoted
to huge positive values.
This has no attack surface though as the loop is to end very soon anyhow
and the method only used while reading CD-ROM mountpoints which seems
like a very unlikely attack vector…
For \xff and friends with the highest bit set and hence being a negative
value on signed char systems the wrong encoding is produced as we run
into undefined behaviour accessing negative array indexes.
We can avoid this problem simply by using an unsigned data type.
This allows patterns like ~nalpha~nbeta and ~nalpha!~nbeta to
work like they do in APT.
Also add a comment to remind readers that everything in START
should be in short too.
Cc: stable >= 2.0
Our testcases had their own implementation of GetTempFile with the
feature of a temporary file with a choosen suffix. Merging this into
GetTempFile lets us drop this duplicate and hence test more our code
rather than testing our helpers for test implementation.
And then hashsums_test had another implementation… and extracttar wasn't
even trying to use a real tempfile… one GetTempFile to rule them all!
That also ensures that these tempfiles are created in a temporary
directory rather than the current directory which is a nice touch and
tries a little harder to clean up those tempfiles.
No sensible file should include these, but even insensible files do not
gain unfair advantages with it as this parser does not deal with
security critical files before they haven't passed other checks like
signatures or hashsums.
The problem is that the parser accepts and parses empty tag names
correctly, but does not store the data parsed which will effect later
passes over the data resulting e.g. in the following tag containing
the name and value of the previous (empty) tag, its own tagname and its
own value or a crash due to an attempt to access invalid memory
depending on who passes over the data and what is done with it.
This commit fixes both, the incidient of the crash reported by
Anatoly Trosinenko who reproduced it via apt-sortpkgs:
| $ cat /tmp/Packages-null
| 0:
| PACKAGE:0
|
| :
| PACKAGE:
|
| PACKAGE::
| $ apt-sortpkgs /tmp/Packages-null
and the deeper parsing issue shown by the included testcase.
Reported-By: Anatoly Trosinenko <anatoly.trosinenko@gmail.com>
References: 8710a36a01
we have to stop parsing on space so that things like ~ramd64 | ~rall
work correctly.
aptitude does not stop parsing on ?, but we'll do as it gets very
confusing otherwise if you write stuff like ~ramd64?name(foo), and
it resolves to ?and(?architecture(amd64?name), (foo))...
Also make pattern detector in cacheset and private's list accept
such patterns. We probably should just try to parse and see if it
is a (start of a) pattern.
David Kalnischkies
Julian Andres Klode