Users may fork libgit2 and run libgit2's CI on that, which is
delightful!  However, if they do, we'll fail the documentation publish
phase, which is correct (because we don't allow them to publish _their_
version of the docs) but regrettable (since it should not fail).

Only run the documentation publish phase when we merge branches into the
libgit2/libgit2 repo.

global: convert to fiber-local storage to fix exit races

Fix copy&paste in git_cherrypick_commit docstring

patch_parse: fix out-of-bounds reads caused by integer underflow

On Windows platforms, we automatically clean up the thread-local storage
upon detaching a thread via `DllMain()`. The thing is that this happens
for every thread of applications that link against the libgit2 DLL, even
those that don't have anything to do with libgit2 itself. As a result,
we cannot assume that these unsuspecting threads make use of our
`git_libgit2_init()` and `git_libgit2_shutdow()` reference counting,
which may lead to racy situations:

    Thread 1                    Thread 2

    git_libgit2_shutdown()
                                DllMain(DETACH_THREAD)
                                git__free_tls_data()
    git_atomic_dec() == 0
    git__free_tls_data()
    TlsFree(_tls_index)
                                TlsGetValue(_tls_index)

Due to the second thread never having executed `git_libgit2_init()`, the
first thread will clean up TLS data and as a result also free the
`_tls_index` variable. When detaching the second thread, we
unconditionally access the now-free'd `_tls_index` variable, which is
obviously not going to work out well.

Fix the issue by converting the code to use fiber-local storage instead
of thread-local storage. While FLS will behave the exact same as TLS if
no fibers are in use, it does allow us to specify a destructor similar
to the one that is accepted by pthread_key_create(3P). Like this, we do
not have to manually free indices anymore, but will let the FLS handle
calling the destructor. This allows us to get rid of `DllMain()`
completely, as we only used it to keep track of when threads were
exiting and results in an overall simplification of TLS cleanup.

tests: fix compiler warning if tracing is disabled

tests: config: only test parsing huge file with GITTEST_INVASIVE_SPEED

The test in config::stress::huge_section_with_many_values takes quite a
long time to execute. Hide it behind the GITTEST_INVASIVE_SPEED
environment varibale to not needlessly blow up execution time of tests.
As this environment variable is being set by the continuous integration,
we will execute it regularly anyway.

The patch format for binary files is a simple Base85 encoding with a
length byte as prefix that encodes the current line's length. For each
line, we thus check whether the line's actual length matches its
expected length in order to not faultily apply a truncated patch. This
also acts as a check to verify that we're not reading outside of the
line's string:

	if (encoded_len > ctx->parse_ctx.line_len - 1) {
		error = git_parse_err(...);
		goto done;
	}

There is the possibility for an integer underflow, though. Given a line
with a single prefix byte, only, `line_len` will be zero when reaching
this check. As a result, subtracting one from that will result in an
integer underflow, causing us to assume that there's a wealth of bytes
available later on. Naturally, this may result in an out-of-bounds read.

Fix the issue by checking both `encoded_len` and `line_len` for a
non-zero value. The binary format doesn't make use of zero-length lines
anyway, so we need to know that there are both encoded bytes and
remaining characters available at all.

This patch also adds a test that works based on the last error message.
Checking error messages is usually too tightly coupled, but in fact
parsing the patch failed even before the change. Thus the only
possibility is to use e.g. Valgrind, but that'd result in us not
catching issues when run without Valgrind. As a result, using the error
message is considered a viable tradeoff as we know that we didn't start
decoding Base85 in the first place.

If building libgit2's test suite with tracing disabled, then the
compiler will emit a warning due to the unused `message_prefix`
function. Fix the issue by wrapping the whole file into ifdef's for
`GIT_TRACE` and providing separate empty function implementations for
both `cl_global_trace_register` and `cl_global_trace_disable`.

diff: complete support for git patchid

Memory optimizations for config entries

Current implementation of patchid is not computing a correct patchid
when given a patch where, for example, a new file is added or removed.
Some more corner cases need to be handled to have same behavior as git
patch-id command.
Add some more tests to cover those corner cases.

Signed-off-by: Gregory Herrero <gregory.herrero@oracle.com>

ssh: include sha256 host key hash when supported

Various examples shape-ups

Improve trace support in tests

Move `git_off_t` to `git_object_size_t`

Add compat typdef for git_attr_t

Some libraries haven't updated to git_attr_value_t and break.  Adding
the comapt typedef as suggested.

Only show test trace execution when the CLAR_TRACE_TESTS environment
variable is set.  This reduces the noise during tracing.

Tracing is meant to be extremely low-impact when not enabled.  We
currently ship no tracing calls in libgit2, but if / when we do, the
tracing infrastructure is created to skip tracing as quickly as
possible.  It should compile to a simple test when tracing is off.

Thus, there's on reason to not enable it by default.

CI Build Updates

Prefer `off64_t` internally.

Use int64_t internally for type visibility.

Prefer `off64_t` to `git_off_t` internally for visibility.

Prefer `off64_t` to `git_off_t` for internal visibility.

64 bit types are always 64 bit.

Provide usage hints to valgrind.  We trust the data coming back from
OpenSSL to have been properly initialized.  (And if it has not, it's an
OpenSSL bug, not a libgit2 bug.)

We previously took the `VALGRIND` option to CMake as a hint to disable
mmap.  Remove that; it's broken.  Now use it to pass on the `VALGRIND`
definition so that sources can provide valgrind hints.

valgrind will warn that OpenSSL will use undefined data in connect/read
when talking to certain other TLS stacks.  Thankfully, this only seems
to occur when gcc is the compiler, so hopefully valgrind is just
misunderstanding an optimization.  Regardless, suppress this warning.

We currently talk to Azure Repos for executing an online test
(online::clone::path_whitespace).  Add a simpler test to talk to Azure
Repos to make it obvious that strange test failures are not likely the
whitespace in the path, but actually a function of talking to Azure
Repos itself.

Our docker builds are getting expensive, let's cache some of this.

The valgrind in the PPA is broken and ignores `--exit-errorcode`.
Build and install our own.