Until "credential: refuse to operate when missing host or protocol",
Git's credential handling code interpreted URLs with empty scheme to
mean "give me credentials matching this host for any protocol".
Luckily libcurl does not recognize such URLs (it tries to look for a
protocol named "" and fails). Just in case that changes, let's reject
them within Git as well. This way, credential_from_url is guaranteed to
always produce a "struct credential" with protocol and host set.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
libcurl permits making requests without a URL scheme specified. In
this case, it guesses the URL from the hostname, so I can run
git ls-remote http::ftp.example.com/path/to/repo
and it would make an FTP request.
Any user intentionally using such a URL is likely to have made a typo.
Unfortunately, credential_from_url is not able to determine the host and
protocol in order to determine appropriate credentials to send, and
until "credential: refuse to operate when missing host or protocol",
this resulted in another host's credentials being leaked to the named
host.
Teach credential_from_url_gently to consider such a URL to be invalid
so that fsck can detect and block gitmodules files with such URLs,
allowing server operators to avoid serving them to downstream users
running older versions of Git.
This also means that when such URLs are passed on the command line, Git
will print a clearer error so affected users can switch to the simpler
URL that explicitly specifies the host and protocol they intend.
One subtlety: .gitmodules files can contain relative URLs, representing
a URL relative to the URL they were cloned from. The relative URL
resolver used for .gitmodules can follow ".." components out of the path
part and past the host part of a URL, meaning that such a relative URL
can be used to traverse from a https://foo.example.com/innocent
superproject to a https::attacker.example.com/exploit submodule.
Fortunately a leading ':' in the first path component after a series of
leading './' and '../' components is unlikely to show up in other
contexts, so we can catch this by detecting that pattern.
Reported-by: Jeff King <peff@peff.net>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Reviewed-by: Jeff King <peff@peff.net>
When we try to initialize credential loading by URL and find that the
URL is invalid, we set all fields to NULL in order to avoid acting on
malicious input. Later when we request credentials, we diagonse the
erroneous input:
fatal: refusing to work with credential missing host field
This is problematic in two ways:
- The message doesn't tell the user *why* we are missing the host
field, so they can't tell from this message alone how to recover.
There can be intervening messages after the original warning of
bad input, so the user may not have the context to put two and two
together.
- The error only occurs when we actually need to get a credential. If
the URL permits anonymous access, the only encouragement the user gets
to correct their bogus URL is a quiet warning.
This is inconsistent with the check we perform in fsck, where any use
of such a URL as a submodule is an error.
When we see such a bogus URL, let's not try to be nice and continue
without helpers. Instead, die() immediately. This is simpler and
obviously safe. And there's very little chance of disrupting a normal
workflow.
It's _possible_ that somebody has a legitimate URL with a raw newline in
it. It already wouldn't work with credential helpers, so this patch
steps that up from an inconvenience to "we will refuse to work with it
at all". If such a case does exist, we should figure out a way to work
with it (especially if the newline is only in the path component, which
we normally don't even pass to helpers). But until we see a real report,
we're better off being defensive.
Reported-by: Carlo Arenas <carenas@gmail.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
The credential helper protocol was designed to be very flexible: the
fields it takes as input are treated as a pattern, and any missing
fields are taken as wildcards. This allows unusual things like:
echo protocol=https | git credential reject
to delete all stored https credentials (assuming the helpers themselves
treat the input that way). But when helpers are invoked automatically by
Git, this flexibility works against us. If for whatever reason we don't
have a "host" field, then we'd match _any_ host. When you're filling a
credential to send to a remote server, this is almost certainly not what
you want.
Prevent this at the layer that writes to the credential helper. Add a
check to the credential API that the host and protocol are always passed
in, and add an assertion to the credential_write function that speaks
credential helper protocol to be doubly sure.
There are a few ways this can be triggered in practice:
- the "git credential" command passes along arbitrary credential
parameters it reads from stdin.
- until the previous patch, when the host field of a URL is empty, we
would leave it unset (rather than setting it to the empty string)
- a URL like "example.com/foo.git" is treated by curl as if "http://"
was present, but our parser sees it as a non-URL and leaves all
fields unset
- the recent fix for URLs with embedded newlines blanks the URL but
otherwise continues. Rather than having the desired effect of
looking up no credential at all, many helpers will return _any_
credential
Our earlier test for an embedded newline didn't catch this because it
only checked that the credential was cleared, but didn't configure an
actual helper. Configuring the "verbatim" helper in the test would show
that it is invoked (it's obviously a silly helper which doesn't look at
its input, but the point is that it shouldn't be run at all). Since
we're switching this case to die(), we don't need to bother with a
helper. We can see the new behavior just by checking that the operation
fails.
We'll add new tests covering partial input as well (these can be
triggered through various means with url-parsing, but it's simpler to
just check them directly, as we know we are covered even if the url
parser changes behavior in the future).
[jn: changed to die() instead of logging and showing a manual
username/password prompt]
Reported-by: Carlo Arenas <carenas@gmail.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
We may feed a URL like "cert:///path/to/cert.pem" into the credential
machinery to get the key for a client-side certificate. That
credential has no hostname field, which is about to be disallowed (to
avoid confusion with protocols where a helper _would_ expect a
hostname).
This means as of the next patch, credential helpers won't work for
unlocking certs. Let's fix that by doing two things:
- when we parse a url with an empty host, set the host field to the
empty string (asking only to match stored entries with an empty
host) rather than NULL (asking to match _any_ host).
- when we build a cert:// credential by hand, similarly assign an
empty string
It's the latter that is more likely to impact real users in practice,
since it's what's used for http connections. But we don't have good
infrastructure to test it.
The url-parsing version will help anybody using git-credential in a
script, and is easy to test.
Signed-off-by: Jeff King <peff@peff.net>
Reviewed-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
The credential protocol can't represent newlines in values, but URLs can
embed percent-encoded newlines in various components. A previous commit
taught the low-level writing routines to die() when encountering this,
but we can be a little friendlier to the user by detecting them earlier
and handling them gracefully.
This patch teaches credential_from_url() to notice such components,
issue a warning, and blank the credential (which will generally result
in prompting the user for a username and password). We blank the whole
credential in this case. Another option would be to blank only the
invalid component. However, we're probably better off not feeding a
partially-parsed URL result to a credential helper. We don't know how a
given helper would handle it, so we're better off to err on the side of
matching nothing rather than something unexpected.
The die() call in credential_write() is _probably_ impossible to reach
after this patch. Values should end up in credential structs only by URL
parsing (which is covered here), or by reading credential protocol input
(which by definition cannot read a newline into a value). But we should
definitely keep the low-level check, as it's our final and most accurate
line of defense against protocol injection attacks. Arguably it could
become a BUG(), but it probably doesn't matter much either way.
Note that the public interface of credential_from_url() grows a little
more than we need here. We'll use the extra flexibility in a future
patch to help fsck catch these cases.
The credential protocol that we use to speak to helpers can't represent
values with newlines in them. This was an intentional design choice to
keep the protocol simple, since none of the values we pass should
generally have newlines.
However, if we _do_ encounter a newline in a value, we blindly transmit
it in credential_write(). Such values may break the protocol syntax, or
worse, inject new valid lines into the protocol stream.
The most likely way for a newline to end up in a credential struct is by
decoding a URL with a percent-encoded newline. However, since the bug
occurs at the moment we write the value to the protocol, we'll catch it
there. That should leave no possibility of accidentally missing a code
path that can trigger the problem.
At this level of the code we have little choice but to die(). However,
since we'd not ever expect to see this case outside of a malicious URL,
that's an acceptable outcome.
Reported-by: Felix Wilhelm <fwilhelm@google.com>
In some cases, a user will want to use a specific credential helper for
a wildcard pattern, such as https://*.corp.example.com. We have code
that handles this already with the urlmatch code, so let's use that
instead of our custom code.
Since the urlmatch code is a superset of our current matching in terms
of capabilities, there shouldn't be any cases of things that matched
previously that don't match now. However, in addition to wildcard
matching, we now use partial path matching, which can cause slightly
different behavior in the case that a helper applies to the prefix
(considering path components) of the remote URL. While different, this
is probably the behavior people were wanting anyway.
Since we're using the urlmatch code, we need to encode the components
we've gotten into a URL to match, so add a function to percent-encode
data and format the URL with it. We now also no longer need to the
custom code to match URLs, so let's remove it.
Additionally, the urlmatch code always looks for the best match, whereas
we want all matches for credential helpers to preserve existing
behavior. Let's add an optional field, select_fn, that lets us control
which items we want (in this case, all of them) and default it to the
best-match code that already exists for other users.
Signed-off-by: brian m. carlson <bk2204@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Everywhere else in the codebase, we use the rule that the last matching
configuration option is the one that takes effect. This is helpful
because it allows more specific configuration settings (e.g., per-repo
configuration) to override less specific settings (e.g., per-user
configuration).
However, in the credential code, we didn't honor this setting, and
instead picked the first setting we had, and stuck with it. This was
likely to ensure we picked the value from the URL, which we want to
honor over the configuration.
It's possible to do both, though, so let's check if the value is the one
we've gotten over our protocol connection, which if present will have
come from the URL, and keep it if so. Otherwise, let's overwrite the
value with the latest version we've got from the configuration, so we
keep the last configuration value.
Signed-off-by: brian m. carlson <bk2204@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The credential subsystem can trigger SIGPIPE when writing to an
external helper if that helper closes its stdin before reading the
whole input. Normally this is rare, since helpers would need to read
that input to make a decision about how to respond, but:
1. It's reasonable to configure a helper which only handles "get"
while ignoring "store". Such a handler might not read stdin
for "store", thereby rapidly closing stdin upon helper exit.
2. A broken or misbehaving helper might exit immediately. That's an
error, but it's not reasonable for it to take down the parent Git
process with SIGPIPE.
Even with such a helper, seeing this problem should be rare. Getting
SIGPIPE requires the helper racily exiting before we've written the
fairly small credential output.
Signed-off-by: Erik E Brady <brady@cisco.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A common pattern to free a piece of memory and assign NULL to the
pointer that used to point at it has been replaced with a new
FREE_AND_NULL() macro.
* ab/free-and-null:
*.[ch] refactoring: make use of the FREE_AND_NULL() macro
coccinelle: make use of the "expression" FREE_AND_NULL() rule
coccinelle: add a rule to make "expression" code use FREE_AND_NULL()
coccinelle: make use of the "type" FREE_AND_NULL() rule
coccinelle: add a rule to make "type" code use FREE_AND_NULL()
git-compat-util: add a FREE_AND_NULL() wrapper around free(ptr); ptr = NULL
Replace occurrences of `free(ptr); ptr = NULL` which weren't caught by
the coccinelle rule. These fall into two categories:
- free/NULL assignments one after the other which coccinelle all put
on one line, which is functionally equivalent code, but very ugly.
- manually spotted occurrences where the NULL assignment isn't right
after the free() call.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Apply the result of the just-added coccinelle rule. This manually
excludes a few occurrences, mostly things that resulted in many
FREE_AND_NULL() on one line, that'll be manually fixed in a subsequent
change.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Stop including config.h by default in cache.h. Instead only include
config.h in those files which require use of the config system.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Sine the credential.helper key is a multi-valued config
list, there's no way to "unset" a helper once it's been set.
So if your system /etc/gitconfig sets one, you can never
avoid running it, but only add your own helpers on top.
Since an empty value for credential.helper is nonsensical
(it would just try to run "git-credential-"), we can assume
nobody is using it. Let's define it to reset the helper
list, letting you override lower-priority instances which
have come before.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The strbuf_getline() interface allows a byte other than LF or NUL as
the line terminator, but this is only because I wrote these
codepaths anticipating that there might be a value other than NUL
and LF that could be useful when I introduced line_termination long
time ago. No useful caller that uses other value has emerged.
By now, it is clear that the interface is overly broad without a
good reason. Many codepaths have hardcoded preference to read
either LF terminated or NUL terminated records from their input, and
then call strbuf_getline() with LF or NUL as the third parameter.
This step introduces two thin wrappers around strbuf_getline(),
namely, strbuf_getline_lf() and strbuf_getline_nul(), and
mechanically rewrites these call sites to call either one of
them. The changes contained in this patch are:
* introduction of these two functions in strbuf.[ch]
* mechanical conversion of all callers to strbuf_getline() with
either '\n' or '\0' as the third parameter to instead call the
respective thin wrapper.
After this step, output from "git grep 'strbuf_getline('" would
become a lot smaller. An interim goal of this series is to make
this an empty set, so that we can have strbuf_getline_crlf() take
over the shorter name strbuf_getline().
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Credential helpers are asked in turn until one of them give
positive response, which is cumbersome to turn off when you need to
run Git in an automated setting. The credential helper interface
learned to allow a helper to say "stop, don't ask other helpers."
Also GIT_TERMINAL_PROMPT environment can be set to false to disable
our built-in prompt mechanism for passwords.
* jk/credential-quit:
prompt: respect GIT_TERMINAL_PROMPT to disable terminal prompts
credential: let helpers tell us to quit
When we are trying to fill a credential, we loop over the
set of defined credential-helpers, then fall back to running
askpass, and then finally prompt on the terminal. Helpers
which cannot find a credential are free to tell us nothing,
but they cannot currently ask us to stop prompting.
This patch lets them provide a "quit" attribute, which asks
us to stop the process entirely (avoiding running more
helpers, as well as the askpass/terminal prompt).
This has a few possible uses:
1. A helper which prompts the user itself (e.g., in a
dialog) can provide a "cancel" button to the user to
stop further prompts.
2. Some helpers may know that prompting cannot possibly
work. For example, if their role is to broker a ticket
from an external auth system and that auth system
cannot be contacted, there is no point in continuing
(we need a ticket to authenticate, and the user cannot
provide one by typing it in).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Most struct child_process variables are cleared using memset first after
declaration. Provide a macro, CHILD_PROCESS_INIT, that can be used to
initialize them statically instead. That's shorter, doesn't require a
function call and is slightly more readable (especially given that we
already have STRBUF_INIT, ARGV_ARRAY_INIT etc.).
Helped-by: Johannes Sixt <j6t@kdbg.org>
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The skip_prefix() function returns a pointer to the content
past the prefix, or NULL if the prefix was not found. While
this is nice and simple, in practice it makes it hard to use
for two reasons:
1. When you want to conditionally skip or keep the string
as-is, you have to introduce a temporary variable.
For example:
tmp = skip_prefix(buf, "foo");
if (tmp)
buf = tmp;
2. It is verbose to check the outcome in a conditional, as
you need extra parentheses to silence compiler
warnings. For example:
if ((cp = skip_prefix(buf, "foo"))
/* do something with cp */
Both of these make it harder to use for long if-chains, and
we tend to use starts_with() instead. However, the first line
of "do something" is often to then skip forward in buf past
the prefix, either using a magic constant or with an extra
strlen(3) (which is generally computed at compile time, but
means we are repeating ourselves).
This patch refactors skip_prefix() to return a simple boolean,
and to provide the pointer value as an out-parameter. If the
prefix is not found, the out-parameter is untouched. This
lets you write:
if (skip_prefix(arg, "foo ", &arg))
do_foo(arg);
else if (skip_prefix(arg, "bar ", &arg))
do_bar(arg);
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The git-credential command requires that you feed it a
broken-down credential, which means that the client needs to
parse a URL itself. Since we have our own URL-parsing
routines, we can easily allow the caller to just give us the
URL as-is, saving them some code.
Signed-off-by: Jeff King <peff@peff.net>
Acked-by: Matthieu Moy <Matthieu.Moy@imag.fr>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Instead of outputing only the username and password, print all the
attributes, even those that already appeared in the input.
This is closer to what the C API does, and allows one to take the exact
output of "git credential fill" as input to "git credential approve" or
"git credential reject".
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We use git_getpass to retrieve the username and password
from the terminal. However, git_getpass will not echo the
username as the user types. We can fix this by using the
more generic git_prompt, which underlies git_getpass but
lets us specify an "echo" option.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is currently in connect.c, but really has nothing to
do with the git protocol itself. Let's make a new source
file all about prompting the user, which will make it
cleaner to refactor.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When parsing a URL into a credential struct, we carefully
record each part of the URL, including the path on the
remote host, and use the result as part of the credential
context.
This had two practical implications:
1. Credential helpers which store a credential for later
access are likely to use the "path" portion as part of
the storage key. That means that a request to
https://example.com/foo.git
would not use the same credential that was stored in an
earlier request for:
https://example.com/bar.git
2. The prompt shown to the user includes all relevant
context, including the path.
In most cases, however, users will have a single password
per host. The behavior in (1) will be inconvenient, and the
prompt in (2) will be overly long.
This patch introduces a config option to toggle the
relevance of http paths. When turned on, we use the path as
before. When turned off, we drop the path component from the
context: helpers don't see it, and it does not appear in the
prompt.
This is nothing you couldn't do with a clever credential
helper at the start of your stack, like:
[credential "http://"]
helper = "!f() { grep -v ^path= ; }; f"
helper = your_real_helper
But doing this:
[credential]
useHttpPath = false
is way easier and more readable. Furthermore, since most
users will want the "off" behavior, that is the new default.
Users who want it "on" can set the variable (either for all
credentials, or just for a subset using
credential.*.useHttpPath).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Credential helpers can help users avoid having to type their
username and password over and over. However, some users may
not want a helper for their password, or they may be running
a helper which caches for a short time. In this case, it is
convenient to provide the non-secret username portion of
their credential via config.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The functionality for credential storage helpers is already
there; we just need to give the users a way to turn it on.
This patch provides a "credential.helper" configuration
variable which allows the user to provide one or more helper
strings.
Rather than simply matching credential.helper, we will also
compare URLs in subsection headings to the current context.
This means you can apply configuration to a subset of
credentials. For example:
[credential "https://example.com"]
helper = foo
would match a request for "https://example.com/foo.git", but
not one for "https://kernel.org/foo.git".
This is overkill for the "helper" variable, since users are
unlikely to want different helpers for different sites (and
since helpers run arbitrary code, they could do the matching
themselves anyway).
However, future patches will add new config variables where
this extra feature will be more useful.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
All of the components of a credential struct can be found in
a URL. For example, the URL:
http://foo:bar@example.com/repo.git
contains:
protocol=http
host=example.com
path=repo.git
username=foo
password=bar
We want to be able to turn URLs into broken-down credential
structs so that we know two things:
1. Which parts of the username/password we still need
2. What the context of the request is (for prompting or
as a key for storing credentials).
This code is based on http_auth_init in http.c, but needed a
few modifications in order to get all of the components that
the credential object is interested in.
Once the http code is switched over to the credential API,
then http_auth_init can just go away.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There are a few places in git that need to get a username
and password credential from the user; the most notable one
is HTTP authentication for smart-http pushing.
Right now the only choices for providing credentials are to
put them plaintext into your ~/.netrc, or to have git prompt
you (either on the terminal or via an askpass program). The
former is not very secure, and the latter is not very
convenient.
Unfortunately, there is no "always best" solution for
password management. The details will depend on the tradeoff
you want between security and convenience, as well as how
git can integrate with other security systems (e.g., many
operating systems provide a keychain or password wallet for
single sign-on).
This patch provides an abstract notion of credentials as a
data item, and provides three basic operations:
- fill (i.e., acquire from external storage or from the
user)
- approve (mark a credential as "working" for further
storage)
- reject (mark a credential as "not working", so it can
be removed from storage)
These operations can be backed by external helper processes
that interact with system- or user-specific secure storage.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
