Age | Commit message (Collapse) | Author |
|
We have no speed problem with handling floats/doubles in our progress
handling, but that shouldn't prevent us from cleaning up the handling
slightly to avoid unclean casting to ints.
Reported-By: gcc -Wdouble-promotion -Wold-style-cast
|
|
As a follow up to the last commit, let's replace APT_CONST
with APT_PURE everywhere to clean stuff up.
|
|
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
|
|
Opening the file before we drop privileges in the methods allows us to
avoid chowning in the acquire main process which can apply to the wrong
file (imagine Binary scoped settings) and surprises users as their
permission setup is overridden.
There are no security benefits as the file is open, so an evil method
could as before read the contents of the file, but it isn't worse than
before and we avoid permission problems in this setup.
|
|
Weak hashes like filesize can be used by methods for basic checks and
early refusals even if we can't use them for hard security proposes.
Normal apt operations are not affected by this as they fail if no strong
hash is available, but if apt is forced to work with weak-only files or
e.g. in apt-helper context it can have benefits as weak is better than
no hash for the methods.
|
|
This makes it easier to see which headers includes what.
The changes were done by running
git grep -l '#\s*include' \
| grep -E '.(cc|h)$' \
| xargs sed -i -E 's/(^\s*)#(\s*)include/\1#\2 include/'
To modify all include lines by adding a space, and then running
./git-clang-format.sh.
|
|
Having messages being printed on the error stack and confirm them by
commandline flags is an okayish first step, but some frontends will
probably want to have a more interactive feeling here with a proper
question the user can just press yes/no for as for some frontends a
commandline flag makes no sense…
|
|
POSIX.1-2008 gives us a range of *at calls to deal with files
including the unlinkat so we can remove a file from a directory
based on a path to the file relative to the directory.
(In our case here the path we have is just the filename)
We avoid changing directories in this way which e.g. fails if the
directory we started in no longer exists or is otherwise inaccessible.
Closes: 860738
|
|
Since the introduction of by-hash, two differently named
files might have the same real URL. In our case, the files
icons-64x64.tar.gz and icons-128x128.tar.gz of empty tarballs.
APT would try to merge them and end with weird errors because
it completed the first download and enters the second stage for
decompressing and verifying. After that it would queue a new item
to copy the original file to the location, but that copy item would
be in the wrong stage, causing it to use the hashes for the
decompressed item.
Closes: #838441
|
|
If apt renames a file to .FAILED it leaves its namespace and is never
touched again – expect since 1.1~exp4 in which "apt clean" will remove
those files. The usefulness of these files rapidly degrades if you don't
keep the update log itself (together with debug output in the best case)
through and on 99% of all system they will be kept around forever just
to collect dust over time and eat up space.
With this commit an update call will remove all FAILED files of previous
runs, so that the FAILED files you have on disk are always only the ones
related to the last apt run stopping apt from hoarding files.
Closes: 846476
|
|
Note: This is a warning about disabling a security feature. It is
supposed to be scary as we are disabling a security feature and we
can't just be silent about it! Downloads really shouldn't happen
any longer as root to decrease the attack surface – but if a warning
causes that much uproar, consider what an error would do…
The old WARNING message:
| W: Can't drop privileges for downloading as file 'foobar' couldn't be
| accessed by user '_apt'. - pkgAcquire::Run (13: Permission denied)
is frequently (incorrectly) considered to be an error message indicating
that the download didn't happen which isn't the case, it was performed,
but without all the security features enabled we could have used if run
from some other place…
The word "unsandboxed" is chosen as the term 'sandbox(ed)' is a common
encounter in feature lists/changelogs and more people are hopefully able
to make the connection to 'security' than it is the case for 'privilege
dropping' which is more correct, but far less known.
Closes: #813786
LP: #1522675
|
|
In ad9416611ab83f7799f2dcb4bf7f3ef30e9fe6f8 we fall back to asking the
original mirror (e.g. a redirector) if we do not get the expected
result. This works for the indexes, but patches are a different beast
and much simpler. Adding this fallback code here seems like overkill as
they are usually right along their Index file, so actually forward the
relevant settings to the patch items which fixes pdiff support combined
with a redirector and partial mirrors as in such a situation the pdiff
patches would be 404 and the complete index would be downloaded.
|
|
Employ a priority queue instead of a normal queue to hold
the items; and only add items to the running pipeline if
their priority is the same or higher than the priority
of items in the queue.
The priorities are designed for a 3 stage pipeline system:
In stage 1, all Release files and .diff/Index files are fetched. This
allows us to determine what files remain to be fetched, and thus
ensures a usable progress reporting.
In stage 2, all Pdiff patches are fetched, so we can apply them
in parallel with fetching other files in stage 3.
In stage 3, all other files are fetched (complete index files
such as Contents, Packages).
Performance improvements, mainly from fetching the pdiff patches
before complete files, so they can be applied in parallel:
For the 01 Sep 2016 03:35:23 UTC -> 02 Sep 2016 09:25:37 update
of Debian unstable and testing with Contents and appstream for
amd64 and i386, update time reduced from 37 seconds to 24-28
seconds.
Previously, apt would first download new DEP11 icon tarballs
and metadata files, causing the CPU to be idle. By fetching
the diffs in stage 2, we can now patch our contents and Packages
files while we are downloading the DEP11 stuff.
|
|
|
|
This is needed on BSD where root's default group is wheel, not
root.
|
|
The C.UTF-8 locale is not portable, so we need to use C, otherwise
we crash on other systems. We can use std::locale::classic() for
that, which might also be a bit cheaper than using locale("C").
|
|
Improve-Upon: 2e2865ae53a65c00dd55a892d5b48458f3110366
Reported-By: Julian Andres Klode
Gbp-Dch: Ignore
|
|
The bugreport shows a segfault caused by the code not doing the correct
magical dance to remove an item from inside a queue in all cases. We
could try hard to fix this, but it is actually better and also easier to
perform these checks (which cause instant failure) earlier so that they
haven't entered queue(s) yet, which in return makes cleanup trivial.
The result is that we actually end up failing "too early" as if we
wouldn't be careful download errors would be logged before that process
was even started. Not a problem for the acquire system, but likely to
confuse users and programs alike if they see the download process
producing errors before apt was technically allowed to do an acquire
(it didn't, so no violation, but it looks like it to the untrained eye).
Closes: 835195
|
|
This time it is the formatting of floating numbers in progress
reporting with a radix charater potentially not being dot.
Followup of 7303e11ff28f920a6277c159aa46f80c007350bb. Regression of
b58e2c7c56b1416a343e81f9f80cb1f02c128e25 in so far as it exchanging
very effected with slightly less effected code.
LP: 1611010
|
|
Setting the C++ locale via std::locale::global(std::locale("")); which
would otherwise default to the default C locale (aka: unaffected by
setlocale) effects the formatting of numeric types in IO streams, which
for output for humans is perfectly sensible, but breaks our many text
interfaces used and parsed by us and others without expecting the
numbers to be formatted.
Closes: #825396
|
|
libapt allows to configure compressors to be used by its system via
configuration implemented in 03bef78461c6f443187b60799402624326843396,
but that was never really documented and also only partly working, which
also explains why the tests weren't using it…
|
|
Progress reporting used an "upper bound" on files we might get, expect
that this wasn't correct in case pdiff entered the picture. So instead
of calculating a value which is perhaps incorrect, we just accept that
we can't tell how many files we are going to download and just keep at
0% until we know. Additionally, if we have pdiffs we wait until we got
these (sub)index files, too.
That could all be done better by downloading all Release files first and
planing with them in hand accordingly, but one step at a time.
|
|
Queues feeding workers like rred are created in a random pattern to get
a few of them to run in parallel – but if we already have an idling queue
we don't need to assign it to a (potentially new) random queue as that
saves us the (agruably small) overhead of starting up a new queue,
avoids adding jobs to an already busy queue while others idle and as
a bonus reduces the size of debug logs a bit.
We also keep starting new queues now until we reach our limit before
we assign work at random to them, which should give us a more effective
utilisation overall compared to potentially adding work to busy queues
while we haven't reached our queue limit yet.
|
|
For the non-pdiff case, we have can have accurate progress
reporting because after fetching the {,In}Release files we know
how many IndexFiles will be fetched and what size they have.
Therefore init the filesize early (in pkgAcqIndex::Init) and
ensure that in Acquire::Pulse() looks at already downloaded
bits when calculating the progress in Acquire::Pulse.
Also improve debug output of Debug::acquire::progress
|
|
pkgAcqChangelog has the default behaviour of downloading a changelog to
a temporary directory (inside /tmp, not /tmp directly), which is cleaned
up on shutdown, but this can be overridden to store the changelog more
permanently – but that caries a permission problem.
For changelog we can 'easily' solve this by always downloading to a
temporary directory and only move it out of there on done.
|
|
Introduced in 9d2a8a7388cf3b0bbbe92f6b0b30a533e1167f40 apt tries to
merge actions like downloading the same (as judged by hashes) file
into doing it once. The implementation was very simple in that it isn't
planing at all. Turns out that it works 90% of the time just fine, but
has issues in more complicated situations in which items can be in
different stages downloading different files emitting potentially the
"wrong" hash – like while pdiffs are worked on we might end up copying
the patch instead of the result file giving us very strange errors in
return. Reverting the change until we can implement a better planing
solution seems to be the best course of action even if its sad.
Closes: 810046
|
|
The maximum parallelization soft limit is the number of CPU
cores * 2 on systems defining _SC_NPROCESSORS_ONLN. The hard
limit in all cases is Acquire::QueueHost::Limit.
|
|
Gbp-Dch: ignore
|
|
We need to pass 0llu instead of 0 as the init value, otherwise
std::accumulate will calculate with ints.
Reported-by: Raphaël Hertzog
|
|
If it does not exist disabled priv dropping as there is nothing
we can drop to. This will unblock people with special chroots
or systems that deleted the "_apt" user.
Closes: #806406
|
|
This fixes a regression caussed by commit
95278287f4e1eeaf5d96749d6fc9bfc53fb400d0
that moved the error detection of RunFds() later into the loop.
However this broke detecting issues like dead acquire methods.
Instead of relying on the global error state (which is bad)
we now pass a boolean value back from RunFds() and break on
false.
Closes: #806406
|
|
In ce1f3a2c we started warning about failing unlinking, which we
consistently do for directories. That isn't a problem as directories
usually aren't in the places we do want to clean up – with the potential
exeception of "lost+found", so lets ignore it like we ignore our own
partial/ subdirectory.
Closes: 805424
|
|
In 0940230d we started dropping privileges for file (and a bit later for
copy, too) with the intend of uniforming this for all methods. The
commit message says that the source will likely fail based on the
compressors already – and there isn't much secret in the repository
content. After all, after apt has run the update everyone can access the
content via apt anyway…
There are sources through which worked before which are mostly
single-deb (and those with the uncompressed files available).
The first one being especially surprising for users maybe, so instead of
failing, we make it so that apt detects that it can't access a source as
_apt and if so doesn't drop (for all sources!) privileges – but we limit
this to file/copy, so the uncompress which might be needed will still
fail – but that failed before this regression.
We display a notice about this, mostly so that if it still fails (e.g.
compressed) the user has some idea what is wrong.
Closes: 805069
|
|
Unlinking /dev/null is bad, we shouldn't do that. Also, we should print
at least a warning if we tried to unlink a file but didn't manage to
pull it of (ignoring the case were the file is /dev/null or doesn't
exist in the first place).
This got triggered by a relatively unlikely to cause problem in
pkgAcquire::Worker::PrepareFiles which would while temporary
uncompressed files (which are set to keep compressed) figure out that to
files are the same and prepare for sharing by deleting them. Bad move.
That also shows why not printing a warning is a bad idea as this hide
the error for in non-root test runs.
Git-Dch: Ignore
|
|
Reported-By: gcc
Understandable: no
Git-Dch: Ignore
|
|
Reported-By: scan-build
Git-Dch: Ignore
|
|
Our error reporting is historically grown into some kind of mess.
A while ago I implemented stacking for the global error which is used in
this commit now to wrap calls to functions which do not report (all)
errors via return, so that only failures in those calls cause a failure
to propergate down the chain rather than failing if anything
(potentially totally unrelated) has failed at some point in the past.
This way we can avoid stopping the entire acquire process just because a
single source produced an error for example. It also means that after
the acquire process the cache is generated – even if the acquire
process had failures – as we still have the old good data around we can and
should generate a cache for (again).
There are probably more instances of this hiding, but all these looked
like the easiest to work with and fix with reasonable (aka net-positive)
effects.
|
|
Various smaller improvements so that the check deals better with already
downloaded files, relative paths and other things.
Git-Dch: Ignore
|
|
We had a very similar method previously for our own private usage, but
with some generalisation we can move this check into the acquire system
proper so that all frontends profit from this compatibility change.
As we are disabling a security feature here a warning is issued and
frontends are advised to consider reworking their download logic if
possible.
Note that this is implemented as an all or nothing situation: We can't
just (not) drop privileges for a subset of the files in a fetcher, so in
case you have to download some files with and some without you need to
use two fetchers.
|
|
Git-Dch: ignore
|
|
|
|
Doing this disables the implicit copy assignment operator (among others)
which would cause hovac if used on the classes as it would just copy the
pointer, not the data the d-pointer points to. For most of the classes
we don't need a copy assignment operator anyway and in many classes it
was broken before as many contain a pointer of some sort.
Only for our Cacheset Container interfaces we define an explicit copy
assignment operator which could later be implemented to copy the data
from one d-pointer to the other if we need it.
Git-Dch: Ignore
|
|
Some of them modify the ABI, but given that we prepare a big one
already, these few hardly count for much.
Git-Dch: Ignore
|
|
To have a chance to keep the ABI for a while we need all three to team
up. One of them missing and we might loose, so ensuring that they are
available is a very tedious but needed task once in a while.
Git-Dch: Ignore
|
|
Progress reports once in a while which is a bit to unpredictable for
testcases, so we enforce a steady progress for them in the hope that
this makes the tests (mostly test-apt-progress-fd) a bit more stable.
Git-Dch: Ignore
|
|
It shouldn't be too common, but sometimes people have multiple mirrors
in the sources or otherwise repositories with the same content. Now that
we gracefully can handle multiple requests to the same URI, we can also
fold multiple requests with the same expected hashes into one. Note that
this isn't trying to find oppertunities for merging, but just merges if
it happens to encounter the oppertunity for it.
This is most obvious in the new testcase actually as it needs to delay
the action to give the acquire system enough time to figure out that
they can be merged.
|
|
This is an unlikely event for indexes and co, but it can happen quiet
easily e.g. for changelogs where you want to get the changelogs for
multiple binary package(version)s which happen to all be built from a
single source.
The interesting part is that the Acquire system actually detected this
already and set the item requesting the URI again to StatDone - expect
that this is hardly sufficient: an Item must be Complete=true as well
to be considered truely done and that is only the tip of the ::Done
handling iceberg. So instead of this StatDone hack we allow QItems to be
owned by multiple items and notify all owners about everything now,
so that for the point of each item they got it downloaded just for them.
|
|
We autocreate for a while now the last two directories in /var/lib/apt/lists
(similar for /var/cache/apt/archives) which is very nice for systems having
any of those on tmpfs or other non-persistent storage. This also means
though that this creation is effected by the default umask, so for
people with aggressive umasks like 027 the directories will be created
with 750, which means all non-root users are left out, which is usually
exactly what we want then this umask is set, but the cache and lib
directories contain public knowledge. There isn't any need to protect
them from viewers and they render apt completely useless if not
readable.
|
|
Do the same with less code in apt-get. This especially ensures that the
lock file (and the parent directories) exist before we are trying to
lock. It also means that clean now creates the directories if they are
missing so we returned to a proper clean state now.
Git-Dch: Ignore
|
|
partial files are chowned by the Item baseclass to let the methods work
with them. Now, this baseclass is also responsible for chowning the
files back to root instead of having various deeper levels do this.
The consequence is that all overloaded Failed() methods now call the
Item::Failed base as their first step. The same is done for Done().
The effect is that even in partial files usually don't belong to
_apt anymore, helping sneakernets and reducing possibilities of a bad
method modifying files not belonging to them.
The change is supported by the framework not only supporting being run
as root, but with proper permission management, too, so that privilege
dropping can be tested with them.
|