This causes an encodeBS thunk, and the first evaluation of the string
forces it. From then on, further uses operate on a ByteString. This
avoids converting repeatedly.
This does, as a side effect, make long notes in json output not
be indented. The indentation is only needed to offset them
underneath the display of the file they apply to, so that's ok.
Sponsored-by: Brock Spratlen on Patreon
UnquotedString does not need to be quoted, but still it's possible
it contains something attacker-controlled, which could have an
escape sequence or control character in it. This is a convenient
place to filter out such things, since quoting alrready handles
those in filenames.
Sponsored-by: Luke Shumaker on Patreon
giveup changed to filter out control characters. (It is too low level to
make it use StringContainingQuotedPath.)
error still does not, but it should only be used for internal errors,
where the message is not attacker-controlled.
Changed a lot of existing error to giveup when it is not strictly an
internal error.
Of course, other exceptions can still be thrown, either by code in
git-annex, or a library, that include some attacker-controlled value.
This does not guard against those.
Sponsored-by: Noam Kremen on Patreon
When the filenames are part of the git repository or other files that
might have attacker-controlled names, quote them in error messages.
This is fairly complete, although I didn't do the one in
Utility.DirWatcher.INotify.hs because that doesn't have access to
Git.Filename or Annex.
But it's also quite possible I missed some. And also while scanning for
these, I found giveup used with other things that could be attacker
controlled to contain control characters (eg Keys). So, I'm thinking
it would also be good for giveup to just filter out control characters.
This commit is then not the only line of defence, but just good
formatting when git-annex displays a filename in an error message.
Sponsored-by: Kevin Mueller on Patreon
Added StringContainingQuotedPath, which is used for ActionItemOther.
In the process, checked every ActionItemOther for those containing
filenames, and made them use quoting.
Sponsored-by: Graham Spencer on Patreon
This is by no means complete, but escaping filenames in actionItemDesc does
cover most commands.
Note that for ActionItemBranchFilePath, the value is branch:file, and I
choose to only quote the file part (if necessary). I considered quoting the
whole thing. But, branch names cannot contain control characters, and while
they can contain unicode, git coes not quote unicode when displaying branch
names. So, it would be surprising for git-annex to quote unicode in a
branch name.
The find command is the most obvious command that still needs to be
dealt with. There are probably other places that filenames also get
displayed, eg embedded in error messages.
Some other commands use ActionItemOther with a filename, I think that
ActionItemOther should either be pre-sanitized, or should explicitly not
be used for filenames, so that needs more work.
When --json is used, unicode does not get escaped, but control
characters were already escaped in json.
(Key escaping may turn out to be needed, but I'm ignoring that for now.)
Sponsored-by: unqueued on Patreon
This turns out to be possible after all, because the old one decomposed
a unicode Char to multiple Word8s and encoded those. It should be faster
in some places, particularly in Git.Filename.encodeAlways.
The old version encoded all unicode by default as well as ascii control
characters and also '"'. The new one only encodes ascii control
characters by default.
That old behavior was visible in Utility.Format.format, which did escape
'"' when used in eg git-annex find --format='${escaped_file}\n'
So made sure to keep that working the same. Although the man page only
says it will escape "unusual" characters, so it might be able to be
changed.
Git.Filename.encodeAlways also needs to escape '"' ; that was the
original reason that was escaped.
Types.Transferrer I judge is ok to not escape '"', because the escaped
value is sent in a line-based protocol, which is decoded at the other
end by decode_c. So old git-annex and new will be fine whether that is
escaped or not, the result will be the same.
Note that when asked to escape a double quote, it is escaped to \"
rather than to \042. That's the same behavior as git has. It's
perhaps somehow more of a special case than it needs to be.
Sponsored-by: k0ld on Patreon
This speeds up a few things, notably CmdLine.Seek using Git.Filename
which uses decode_c and this avoids a conversion to String and back,
and probably the ByteString implementation of decode_c is also faster
for simple cases at least than the string version.
encode_c cannot be converted to ByteString (or if it did, it would have
to convert right back to String in order to handle unicode).
Sponsored-by: Brock Spratlen on Patreon
That is a legal url, but parseUrl parses it to "/c:/path"
which is not a valid path on Windows. So as a workaround, use
parseURIPortable everywhere, which removes the leading slash when
run on windows.
Note that if an url is parsed like this and then serialized back
to a string, it will be different from the input. Which could
potentially be a problem, but is probably not in practice.
An alternative way to do it would be to have an uriPathPortable
that fixes up the path after parsing. But it would be harder to
make sure that is used everywhere, since uriPath is also used
when constructing an URI.
It's also worth noting that System.FilePath.normalize "/c:/path"
yields "c:/path". The reason I didn't use it is that it also
may change "/" to "\" in the path and I wanted to keep the url
changes minimal. Also noticed that convertToWindowsNativeNamespace
handles "/c:/path" the same as "c:/path".
Sponsored-By: the NIH-funded NICEMAN (ReproNim TR&D3) project
Such an url is not valid; parseURI will fail on it. But git-annex doesn't
actually need to parse the url, because all it needs to do to support
syncing with it is know that it's not a local path, and use git pull and
push.
(Note that there is no good reason for the user to use such an url. An
absolute url is valid and I patched git-remote-gcrypt to support them
years ago. Still, users gonna do anything that tools allow, and
git-remote-gcrypt still supports them.)
Sponsored-by: Jack Hill on Patreon
git hash-object --stdin-paths is a newline protocol so it cannot
support them. It would help to not use absPath, when the problem
is that the repository itself is in a path with a newline. But,
there's a reason it used absPath, which is that
git hash-object --stdin-paths actually chdirs to the top of the
repository on startup! That is not documented, and I think is a bug
in git.
I considered making the path relative to the top of the repo, but
then what if this is a git bug and gets fixed? git-annex would break
horribly.
So instead, keep the absPath, but when the path contains a newline,
fall back to running git hash-object once per file, which avoids
the problem with newlines and --stdin-paths. It will be slower,
but this is an edge case. (Similar slow code paths are already used
elsewhere when dealing with filenames with newlines and other parts
of git that use line-based protocols.)
Sponsored-by: Dartmouth College's Datalad project
Works around this bug in unix-compat:
https://github.com/jacobstanley/unix-compat/issues/56
getFileStatus and other FilePath using functions in unix-compat do not do
UNC conversion on Windows.
Made Utility.RawFilePath use convertToWindowsNativeNamespace to do the
necessary conversion on windows to support long filenames.
Audited all imports of System.PosixCompat.Files to make sure that no
functions that operate on FilePath were imported from it. Instead, use
the equvilants from Utility.RawFilePath. In particular the
re-export of that module in Common had to be removed, which led to lots
of other changes throughout the code.
The changes to Build.Configure, Build.DesktopFile, and Build.TestConfig
make Utility.Directory not be needed to build setup. And so let it use
Utility.RawFilePath, which depends on unix, which cannot be in
setup-depends.
Sponsored-by: Dartmouth College's Datalad project
path to a bare repo when git config is not allowed to list the configs
due to the CVE-2022-24765 fix.
That resulted in a confusing error message, and prevented the nice
message that explains how to mark the repo as safe to use.
Made isBare a tristate so that the case where core.bare is not returned can
be handled.
The handling in updateLocation is to check if the directory
contains config and objects and if so assume it's bare.
Note that if that heuristic is somehow wrong, it would construct a repo
that thinks it's bare but is not. That could cause follow-on problems,
but since git-annex then checks checkRepoConfigInaccessible, and skips
using the repo anyway, a wrong guess should not be a problem.
Sponsored-by: Luke Shumaker on Patreon
Avoids displaying warning about git-annex restage needing to be run in
situations where it does not.
Closing a handle flushes it anyway, so no need for an explict flush. The
handle does get closed twice, but that's fine, the second one does nothing.
Sponsored-by: Dartmouth College's DANDI project
Let GIT_DIR and --git-dir override git's protection against operating in a
repository owned by another user.
This is the same behavior other git commands have.
Sponsored-by: Jarkko Kniivilä on Patreon
When pointer files need to be restaged, they're first written to the
log, and then when the restage operation runs, it reads the log. This
way, if the git-annex process is interrupted before it can do the
restaging, a later git-annex process can do it.
Currently, this lets a git-annex get/drop command be interrupted and
then re-ran, and as long as it gets/drops additional files, it will
clean up after the interrupted command. But more changes are
needed to make it easier to restage after an interrupted process.
Kept using the git queue to run the restage action, even though the
list of files that it builds up for that action is not actually used by
the action. This could perhaps be simplified to make restaging a cleanup
action that gets registered, rather than using the git queue for it. But
I wasn't sure if that would cause visible behavior changes, when eg
dropping a large number of files, currently the git queue flushes
periodically, and so it restages incrementally, rather than all at the
end.
In restagePointerFiles, it reads the restage log twice, once to get
the number of files and size, and a second time to process it.
This seemed better than reading the whole file into memory, since
potentially a huge number of files could be in there. Probably the OS
will cache the file in memory and there will not be much performance
impact. It might be better to keep running tallies in another file
though. But updating that atomically with the log seems hard.
Also note that it's possible for calcRestageLog to see a different file
than streamRestageLog does. More files may be added to the log in
between. That is ok, it will only cause the filterprocessfaster heuristic to
operate with slightly out of date information, so it may make the wrong
choice for the files that got added and be a little slower than ideal.
Sponsored-by: Dartmouth College's DANDI project
Fix a reversion that prevented git-annex from working in a repository when
--git-dir or GIT_DIR is specified to relocate the git directory to
somewhere else. (Introduced in version 10.20220525)
checkRepoConfigInaccessible could still run git config --list, just passing
--git-dir. It seems not necessary, because I know that passing --git-dir
bypasses git's check for repo ownership. I suppose it might be that git
eventually changes to check something about the ownership of the working
tree, so passing --git-dir without --work-tree would still be worth doing.
But for now this is the simple fix.
Sponsored-by: Nicholas Golder-Manning on Patreon
When accessing a git remote over http needs a git credential prompt for a
password, cache it for the lifetime of the git-annex process, rather than
repeatedly prompting.
The git-lfs special remote already caches the credential when discovering
the endpoint. And presumably commands like git pull do as well, since they
may download multiple urls from a remote.
The TMVar CredentialCache is read, so two concurrent calls to
getBasicAuthFromCredential will both prompt for a credential.
There would already be two concurrent password prompts in such a case,
and existing uses of `prompt` probably avoid it. Anyway, it's no worse
than before.
This should not change the behavior of it, unless there are multiple top
directories, and then it should behave the same as if there was a single
top directory that was actually above the directory to be created.
Sponsored-by: Dartmouth College's Datalad project
This caused git to complain that filter-process failed and kill it with
signal 15. Because it wrote an extra flushPkt for an empty file, which
git did not expect, and so git saw an unexpected response to the next
request.
Luckily, filter-process is only used by default in v9 and up, and v8 is
still the default. Also, git had to be updating an empty file, followed
by another file, which is a fairly unlikely situation. And git restarts
filter-process after this happens and uses it to filter the rest of the
files. So this isn't a crippling bug.
Sponsored-by: Luke Shumaker on Patreon
Some small wins, almost certianly swamped by the system calls, but still
worthwhile progress on the RawFilePath conversion.
Sponsored-by: Erik Bjäreholt on Patreon
This is so, if there's some other failure that triggers it, --debug will
show what went wrong. See https://github.com/datalad/datalad/issues/6708
Sponsored-by: Dartmouth College's Datalad project
Deal with git's recent changes to fix CVE-2022-24765, which prevent using
git in a repository owned by someone else.
That makes git config --list not list the repo's configs, only global
configs. So annex.uuid and annex.version are not visible to git-annex.
It displayed a message about that, which is not right for this situation.
Detect the situation and display a better message, similar to the one other
git commands display.
Also, git-annex init when run in that situation would overwrite annex.uuid
with a new one, since it couldn't see the old one. Add a check to prevent
it running too in this situation. It may be that this fix has security
implications, if a config set by the malicious user who owns the repo
causes git or git-annex to run code. I don't think any git-annex configs
get run by git-annex init. It may be that some git config of a command
does get run by one of the git commands that git-annex init runs. ("git
status" is the command that prompted the CVE-2022-24765, since
core.fsmonitor can cause it to run a command). Since I don't know how
to exploit this, I'm not treating it as a security fix for now.
Note that passing --git-dir makes git bypass the security check. git-annex
does pass --git-dir to most calls to git, which it does to avoid needing
chdir to the directory containing a git repository when accessing a remote.
So, it's possible that somewhere in git-annex it gets as far as running git
with --git-dir, and git reads some configs that are unsafe (what
CVE-2022-24765 is about). This seems unlikely, it would have to be part of
git-annex that runs in git repositories that have no (visible) annex.uuid,
and git-annex init is the only one that I can think of that then goes on to
run git, as discussed earlier. But I've not fully ruled out there being
others..
The git developers seem mostly worried about "git status" or a similar
command implicitly run by a shell prompt, not an explicit use of git in
such a repository. For example, Ævar Arnfjörð Bjarma wrote:
> * There are other bits of config that also point to executable things,
> e.g. core.editor, aliases etc, but nothing has been found yet that
> provides the "at a distance" effect that the core.fsmonitor vector
> does.
>
> I.e. a user is unlikely to go to /tmp/some-crap/here and run "git
> commit", but they (or their shell prompt) might run "git status", and
> if you have a /tmp/.git ...
Sponsored-by: Jarkko Kniivilä on Patreon
Avoid treating refs/annex/last-index or other refs that are not commit
objects as evidence of repository corruption.
The repair code checks to find bad refs by trying to run `git log` on
them, and assumes that no output means something is broken. But git log
on a tree object is empty.
This was worth fixing generally, not as a special case, since it's certainly
possible that other things store tree or other objects in refs.
Sponsored-by: Max Thoursie on Patreon
Implemented by making Git.Queue have a FlushAction, which can accumulate
along with another action on files, and runs only once the other action has
run.
This lets git-annex unlock queue up git update-index actions, without
conflicting with the restagePointerFiles FlushActions.
In a repository with filter-process enabled, git-annex unlock will
often not take any more time than before, though it may when the files are
large. Either way, it should always slow down less than git-annex status
speeds up.
When filter-process is not enabled, git-annex unlock will slow down as much
as git status speeds up.
Sponsored-by: Jochen Bartl on Patreon
So that eg, addurl of several large files that take time to download will
update the index for each file, rather than deferring the index updates to
the end.
In cases like an add of many smallish files, where a new file is being
added every few seconds. In that case, the queue will still build up a
lot of changes which are flushed at once, for best performance. Since
the default queue size is 10240, often it only gets flushed once at the
end, same as before. (Notice that updateQueue updated _lastchanged
when adding a new item to the queue without flushing it; that is
necessary to avoid it flushing the queue every 5 minutes in this case.)
But, when it takes more than a 5 minutes to add a file, the overhead of
updating the index immediately is probably small, so do it after each
file. This avoids git-annex potentially taking a very very long time
indeed to stage newly added files, which can be annoying to the user who
would like to get on with doing something with the files it's already
added, eg using git mv to rename them to a better name.
This is only likely to cause a problem if it takes say, 30 seconds to
update the index; doing an extra 30 seconds of work after every 5
minute file add would be less optimal. Normally, updating the index takes
significantly less time than that. On a SSD with 100k files it takes
less than 1 second, and the index write time is bound by disk read and
write so is not too much worse on a hard drive. So I hope this will not
impact users, although if it does turn out to, the time limit could be
made configurable.
A perhaps better way to do it would be to have a background worker
thread that wakes up every 60 seconds or so and flushes the queue.
That is made somewhat difficult because the queue can contain Annex
actions and so this would add a new source of concurrency issues.
So I'm trying to avoid that approach if possible.
Sponsored-by: Erik Bjäreholt on Patreon
This version of git -- or its new default "ort" resolver -- handles such
a conflict by staging two files, one with the original name and the other
named file~ref. Use unmergedSiblingFile when the latter is detected.
(It doesn't do that when the conflict is between a directory and a file
or symlink though, so see previous commit for how that case is handled.)
The sibling file has to be deleted separately, because cleanConflictCruft
may not delete it -- that only handles files that are annex links,
but the sibling file may be the non-annexed file side of the conflict.
The graftin code had assumed that, when the other side of a conclict
is a symlink, the file in the work tree will contain the non-annexed
content that we want it to contain. But that is not the case with the new
git; the file may be the annex link and needs to be replaced with the
content, while the annex link will be written as a -variant file.
(The weird doesDirectoryExist check in graftin turns out to still be
needed, test suite failed when I tried to remove it.)
Test suite passes with new git with ort resolver default. Have not tried it
with old git or other defaults.
Sponsored-by: Noam Kremen on Patreon
Based on my earlier benchmark, I have a rough cost model for how
expensive it is for git-annex smudge to be run on a file, vs
how expensive it is for a gigabyte of a file's content to be read and
piped through to filter-process.
So, using that cost model, it can decide if using filter-process will
be more or less expensive than running the smudge filter on the files to
be restaged.
It turned out to be *really* annoying to temporarily disable
filter-process. I did find a way, but urk, this is horrible. Notice
that, if it's interrupted with it disabled, it will remain disabled
until the next time restagePointerFile runs. Which could be some time
later. If the user runs `git add` or `git checkout` on a lot of small
files before that, they will see slower than expected performance.
(This commit also deletes where I wrote down the benchmark results
earlier.)
Sponsored-by: Noam Kremen on Patreon
filter-process: New command that can make git add/checkout faster when
there are a lot of unlocked annexed files or non-annexed files, but that
also makes git add of large annexed files slower.
Use it by running: git
config filter.annex.process 'git-annex filter-process'
Fully tested and working, but I have not benchmarked it at all.
And, incremental hashing is not done when git add uses it, so extra work is
done in that case.
Sponsored-by: Mark Reidenbach on Patreon
This module is not used yet, but the plan is to use it for smudge/clean
filtering, at least as an option. In some circumstances, using this
interface may perform better than the interface git-annex is currently
using.
Sponsored-by: Brock Spratlen on Patreon
