Coverity fixes

Release 0.99

Release script

This commit also switches our SOVERSION to be "$MAJOR.$MINOR" instead of
"$MINOR", only. This is in preparation of v1.0, where the previous
scheme would've stopped working in an obvious way.

Give the release a name, "Torschlusspanik" (the fear that time is
running out to act).  Indeed, the time is running out for changes to be
included in v1.0.

azure: fix ARM32 builds by replacing gosu(1)

openssl: fix Valgrind issues in nightly builds

fuzzers: Fix the documentation

In our test case object::cache::fast_thread_rush, we're creating 100
concurrent threads opening a repository and reading objects from it.
This test actually fails on ARM32 with an out-of-memory error, which
isn't entirely unexpected.

Work around the issue by halving the number of threads.

Our nightly builds are currently failing due to our ARM-based jobs.
These jobs crash immediately when entering the Docker container with a
exception thrown by Go's language runtime. As we're able to successfully
builds the Docker images in previous steps, it's unlikely to be a bug in
Docker itself. Instead, this exception is thrown by gosu(1), which is a
Go-based utility to drop privileges and run by our entrypoint.

Fix the issue by dropping gosu(1) in favor of sudo(1).

Our two Docker build instructions for Xenial and Bionic have diverged a
bit. Let's re-synchronize them with each other to make them as similar
as possible.

The build step for our Docker images currently succeeds even if building
the Docker image fails due to missing && chains in the build script. Fix
this by adding them in.

Since migrating to Docker containings for our build and test
infrastructure, we do not use the "setup-linux.sh" script anymore.
Remove it to avoid any confusion.

Some of the commands are now out of date.

As OpenSSL loves using uninitialized bytes as another source of entropy,
we need to mark them as defined so that Valgrind won't complain about
use of these bytes. Traditionally, we've been using the macro
`VALGRIND_MAKE_MEM_DEFINED` provided by Valgrind, but starting with
OpenSSL 1.1 the code doesn't compile anymore due to `struct SSL` having
become opaque. As such, we also can't set it as defined anymore, as we
have no way of knowing its size.

Let's change gears instead by just swapping out the allocator functions
of OpenSSL with our own ones. The twist is that instead of calling
`malloc`, we just call `calloc` to have the bytes initialized
automatically. Next to soothing Valgrind, this approach has the benefit
of being completely agnostic of the memory sanitizer and is neatly
contained at a single place.

Note that we shouldn't do this for non-Valgrind builds. As we cannot
set up memory functions for a given SSL context, only, we need to swap
them at a global context. Furthermore, as it's possible to call
`OPENSSL_set_mem_functions` once only, we'd prevent users of libgit2 to
set up their own allocators.

OpenSSL doesn't initialize bytes on purpose in order to generate
additional entropy. Valgrind isn't too happy about that though, causing
it to generate warninings about various issues regarding use of
uninitialized bytes.

We traditionally had some infrastructure to silence these errors in our
OpenSSL stream implementation, where we invoke the Valgrind macro
`VALGRIND_MAKE_MEMDEFINED` in various callbacks that we provide to
OpenSSL. Naturally, we only include these instructions if a preprocessor
define "VALGRIND" is set, and that in turn is only set if passing
"-DVALGRIND" to CMake. We do that in our usual Azure pipelines, but we
in fact forgot to do this in our nightly build. As a result, we get a
slew of warnings for these nightly builds, but not for our normal
builds.

To fix this, we could just add "-DVALGRIND" to our nightly builds. But
starting with commit d827b11b (tests: execute leak checker via CTest
directly, 2019-06-28), we do have a secondary variable that directs
whether we want to use memory sanitizers for our builds. As such, every
user wishing to use Valgrind for our tests needs to pass both options
"VALGRIND" and "USE_LEAK_CHECKER", which is cumbersome and error prone,
as can be seen by our own builds.

Instead, let's consolidate this into a single option, removing the old
"-DVALGRIND" one. Instead, let's just add the preprocessor directive if
USE_LEAK_CHECKER equals "valgrind" and remove "-DVALGRIND" from our own
pipelines.

azure: fix misleading messages printed to stderr being

tests: iterator: fix iterator expecting too few items

The current release process is not documented in any way. As a result,
it's not obvious how releases should be done at all, like e.g. which
locations need adjusting.

To fix this, let's introduce a new script that shall from now on be used
to do all releases. As input it gets the tree that shall be released,
the repository in which to do the release, credentials to
authenticate against GitHub and the new version. E.g. executing the
following will create a new release v0.32:

    $ ./script/release.py 0.32.0 --user pks-t --password ****

While the password may currently be your usual GitLab password, it's
recommended to use a personal access token intead.

The script will then perform the following steps:

    1. Verify that "include/git2/version.h" matches the new version.

    2. Verify that "docs/changelog.md" has a section for that new
       version.

    3. Extract the changelog entries for the current release from
       "docs/changelog.md".

    4. Generate two archives in "tar.gz" and "zip" format via "git
       archive" from the tree passed by the user. If no tree was passed,
       we will use "HEAD".

    5. Create the GitHub release using the extracted changelog entries
       as well as tag and name information derived from the version
       passed by the used.

    6. Upload both code archives to that release.

This should cover all steps required for a new release and thus ensures
that nothing is missing that shouldn't be.

Python's PEP 8 specifies that one shall use spaces instead of tabs as
coding style, and we actually honor that currently. Our EditorConfig
does not special-case Python scripts, though, which is why we end up
with our C coding style and thus with tabs.

Special-case "*.py" files to override that default with spaces to fix
this.

The testcase iterator::workdir::filesystem_gunk sets up quite a lot of
directories, which is why it only runs in case GITTEST_INVASIVE_SPEED is
set in the environment. Because we do not run our default CI with this
variable, we didn't notice commit 852c83ee (refs: refuse to delete
HEAD, 2020-01-15) breaking the test as it introduced a new reference to
the "testrepo" repository.

Fix the oversight by increasing the number of expected iterator items.

In order to properly tear down the test environment, we will kill
git-daemon(1) if we've exercised it. As git-daemon(1) is spawned as a
background process, it is still owned by the shell and thus killing it
later on will print a termination message to the shell's stderr, causing
Azure to report it as an error.

Fix this by disowning the background process.

The Docker entrypoint currently creates the libgit2 user with "useradd
--create-home". As we start the Docker container with two volumes
pointing into "/home/libgit2/", the home directory will already exist.
While useradd(1) copes with this just fine, it will print error messages
to stderr which end up as failures in our Azure pipelines.

Fix this by simply removing the "--create-home" parameter.

Without the "--silent" parameter, curl will print a progress meter to
stderr. Azure has the nice feature of interpreting any output to stderr
as errors with a big red warning towards the end of the build. Let's
thus silence curl to not generate any misleading messages.

When building dependencies for our Docker images, we first download the
sources to disk first, unpack them and finally remove the archive again.
This can be sped up by piping the downloading archive into tar(1)
directly to parallelize both tasks. Furthermore, let's silence curl(1)
to not print to status information to stderr, which tends to be
interpreted as errors by Azure Pipelines.

OpenSSL pre-v1.1 required us to set up a locking function to properly
support multithreading. The locking function signature cannot return any
error codes, and as a result we can't do anything if `git_mutex_lock`
fails. To silence static analysis tools, let's just explicitly ignore
its return value by casting it to `void`.

When adding a new entry to our cache where an entry with the same OID
exists already, then we only update the existing entry in case it is
unparsed and the new entry is parsed. Currently, we do not check the
return value of `git_oidmap_set` though when updating the existing
entry. As a result, we will _not_ have updated the existing entry if
`git_oidmap_set` fails, but have decremented its refcount and
incremented the new entry's refcount. Later on, this may likely lead to
dereferencing invalid memory.

Fix the issue by checking the return value of `git_oidmap_set`. In case
it fails, we will simply keep the existing stored instead, even though
it's unparsed.

Git worktree's have the ability to be locked in order to spare them from
deletion, e.g. if a worktree is absent due to being located on a
removable disk it is a good idea to lock it. When locking such
worktrees, it is possible to give a locking reason in order to help the
user later on when inspecting status of any such locked trees.

The function `git_worktree_is_locked` serves to read out the locking
status. It currently does not properly report any errors when reading
the reason file, and callers are unexpecting of any negative return
values, too. Fix this by converting callers to expect error codes and
checking the return code of `git_futils_readbuffer`.

When checking whether a path is a valid repository path, we try to read
the "commondir" link file. In the process, we neither confirm that
constructing the file's path succeeded nor do we verify that reading the
file succeeded, which might cause us to verify repositories on an empty
or bogus path later on.

Fix this by checking return values. As the function to verify repos
doesn't currently support returning errors, this commit also refactors
the function to return an error code, passing validity of the repo via
an out parameter instead, and adjusts all existing callers.

While `git_zstream_set_input` cannot fail right now, it might change in
the future if we ever decide to have it check its parameters more
vigorously. Let's thus check whether its return code signals an error.

Initialization of the hashing context may fail on some systems, most
notably on Win32 via the legacy hashing context. As such, we need to
always check the error code of `git_hash_ctx_init`, which is not done
when creating a new indexer.

Fix the issue by adding checks.

When queueing objects we want to push, we call `git_revwalk_hide` to
hide all objects already known to the remote from our revwalk. We do not
check its return value though, where the orginial intent was to ignore
the case where the pushed OID is not a known committish. As
`git_revwalk_hide` can fail due to other reasons like out-of-memory
exceptions, we should still check its return value.

Fix the issue by checking the function's return value, ignoring
errors hinting that it's not a committish. As `git_revwalk__push_commit`
currently clobbers these error codes, we need to adjust it as well in
order to make it available downstream.

When calling `git_note_next`, we end up calling `git_iterator_advance`
but ignore its error code. The intent is that we do not want to return
an error if it returns `GIT_ITEROVER`, as we want to return that value
on the next invocation of `git_note_next`. We should still check for any
other error codes returned by `git_iterator_advance` to catch unexpected
internal errors.

Fix this by checking the function's return value, ignoring
`GIT_ITEROVER`.

We should always verify error codes returned by function calls in our
test suite to not accidentally miss any weird results. Coverity reported
missing checks in several locations, which this commit fixes.

While the blame helper function `hunk_message` accepts a printf-style
format string, we didn't add a compiler attribute to let the compiler
check for correct conversion specifiers. As a result, some users of the
function used wrong specifiers.

Add the GIT_FORMAT_PRINTF attribute to the function and fix resulting
warnings by using the correct specifiers.

transports: http: fix custom headers not being applied

azure: fix Coverity pipeline

In commit b9c5b15a (http: use the new httpclient, 2019-12-22), the HTTP
code got refactored to extract a generic HTTP client that operates
independently of the Git protocol. Part of refactoring was the creation
of a new `git_http_request` struct that encapsulates the generation of
requests. Our Git-specific HTTP transport was converted to use that in
`generate_request`, but during the process we forgot to set up custom
headers for the `git_http_request` and as a result we do not send out
these headers anymore.

Fix the issue by correctly setting up the request's custom headers and
add a test to verify we correctly send them.

There's several issues with our Coverity builds, like e.g. missing wget
in our containers. Simplify our Coverity pipeline and fix these issues.

Back in commit 5a6740e7 (azure: build Docker images as part of the
pipeline, 2019-08-02), we have converted our pipelines to use self-built
Docker images to ease making changes to our Dockerfiles. The commit
didn't adjust our Coverity pipeline, though, so let's do this now.