Verifying hashsums of objects we are reading from the ODB may be costly
as we have to perform an additional hashsum calculation on the object.
Especially when reading large objects, the penalty can be as high as
35%, as can be seen when executing the equivalent of `git cat-file` with
and without verification enabled. To mitigate for this, we add a global
option for libgit2 which enables the developer to turn off the
verification, e.g. when he can be reasonably sure that the objects on
disk won't be corrupted.
The upstream git.git project verifies objects when looking them up from
disk. This avoids scenarios where objects have somehow become corrupt on
disk, e.g. due to hardware failures or bit flips. While our mantra is
usually to follow upstream behavior, we do not do so in this case, as we
never check hashes of objects we have just read from disk.
To fix this, we create a new error class `GIT_EMISMATCH` which denotes
that we have looked up an object with a hashsum mismatch. `odb_read_1`
will then, after having read the object from its backend, hash the
object and compare the resulting hash to the expected hash. If hashes do
not match, it will return an error.
This obviously introduces another computation of checksums and could
potentially impact performance. Note though that we usually perform I/O
operations directly before doing this computation, and as such the
actual overhead should be drowned out by I/O. Running our test suite
seems to confirm this guess. On a Linux system with best-of-five
timings, we had 21.592s with the check enabled and 21.590s with the
ckeck disabled. Note though that our test suite mostly contains very
small blobs only. It is expected that repositories with bigger blobs may
notice an increased hit by this check.
In addition to a new test, we also had to change the
odb::backend::nonrefreshing test suite, which now triggers a hashsum
mismatch when looking up the commit "deadbeef...". This is expected, as
the fake backend allocated inside of the test will return an empty
object for the OID "deadbeef...", which will obviously not hash back to
"deadbeef..." again. We can simply adjust the hash to equal the hash of
the empty object here to fix this test.
We currently have no tests which check whether we fail reading corrupted
objects. Add one which modifies contents of an object stored on disk and
then tries to read the object.
The object::lookup tests do use the "testrepo.git" repository in a
read-only way, so we do not set up the repository as a sandbox but
simply open it. But in a future commit, we will want to test looking up
objects which are corrupted in some way, which requires us to modify the
on-disk data. Doing this in a repository without creating the sandbox
will modify contents of our libgit2 repository, though.
Create the repository in a sandbox to avoid this.
Instead of going through the usual steps of reading a tree recursively
into an index, modifying it and writing it back out as a tree, introduce
a function to perform simple updates more efficiently.
`git_tree_create_updated` avoids reading trees which are not modified
and supports upsert and delete operations. It is not as versatile as
modifying the index, but it makes some common operations much more
efficient.
While no extra header fields are defined for tags, git accepts them by
ignoring them and continuing the search for the message. There are a few
tags like this in the wild which git parses just fine, so we should do
the same.
The callback mechanism makes it awkward to write data from an IO
source; move to `_fromstream()` which lets the caller remain in control,
in the same vein as we prefer iterators over foreach callbacks.
By returning when the count goes to zero rather than below it, setting
`howmany` to 7 in fact writes out the string 6 times.
Correct the termination condition to write out the string the amount of
times we specify.
The pair of `git_blob_create_frombuffer()` and
`git_blob_create_frombuffer_commit()` is meant to replace
`git_blob_create_fromchunks()` by providing a way for a user to write a
new blob when they want filtering or they do not know the size.
This approach allows the caller to retain control over when to add data
to this buffer and a more natural fit into higher-level language's own
stream abstractions instead of having to handle IO wait in the callback.
The in-memory buffer size of 2MB is chosen somewhat arbitrarily to be a
round multiple of usual page sizes and a value where most blobs seem
likely to be either going to be way below or way over that size. It's
also a round number of pages.
This implementation re-uses the helper we have from `_fromchunks()` so
we end up writing everything to disk, but hopefully more efficiently
than with a default filebuf. A later optimisation can be to avoid
writing the in-memory contents to disk, with some extra complexity.
Submodules don't exist in the objectdb and the code is making us try to
look for a blob with its commit id, which is obviously not going to
work.
Skip the test if the user wants to insert a submodule.
Path validation may be influenced by `core.protectHFS` and
`core.protectNTFS` configuration settings, thus treebuilders
can take a repository to influence their configuration.
There are some combination of objects and target types which we know
cannot be fulfilled. Return EINVALIDSPEC for those to signify that there
is a mismatch in the user-provided data and what the object model is
capable of satisfying.
If we start at a tag and in the course of peeling find out that we
cannot reach a particular type, we return EPEEL.
The old `allocfmt` is of no use to callers, as they are not able to free
the returned buffer. Export a new API that returns a static string that
doesn't need to be freed.
Finding a filename in a vector means we need to resort it every time we
want to read from it, which includes every time we want to write to it
as well, as we want to find duplicate keys.
A hash-map fits what we want to do much more accurately, as we do not
care about sorting, but just the particular filename.
We still keep removed entries around, as the interface let you assume
they were going to be around until the treebuilder is cleared or freed,
but in this case that involves an append to a vector in the filter case,
which can now fail.
The only time we care about sorting is when we write out the tree, so
let's make that the only time we do any sorting.
This adds in missing calls to `git_buf_sanitize` and fixes a
number of places where `git_buf` APIs could inadvertently write
NUL terminator bytes into invalid buffers. This also changes the
behavior of `git_buf_sanitize` to NUL terminate a buffer if it can
and of `git_buf_shorten` to do nothing if it can.
Adds tests of filtering code with zeroed (i.e. unsanitized) buffer
which was previously triggering a segfault.
Diff and status do not want core.safecrlf to actually raise an
error regardless of the setting, so this extends the filter API
with an additional options flags parameter and adds a flag so that
filters can be applied with GIT_FILTER_OPT_ALLOW_UNSAFE, indicating
that unsafe filter application should be downgraded from a failure
to a warning.
The current version of the commit creation and amend function are unsafe
to use when passing the update_ref parameter, as they do not check that
the reference at the moment of update points to what the user expects.
Make sure that we're moving history forward when we ask the library to
update the reference for us by checking that the first parent of the new
commit is the current value of the reference. We also make sure that the
ref we're updating hasn't moved between the read and the write.
Similarly, when amending a commit, make sure that the current tip of the
branch is the commit we're amending.
This finds a short id string that will unambiguously select the
given object, starting with the core.abbrev length (usually 7)
and growing until it is no longer ambiguous.
This adds an API to amend an existing commit, basically a shorthand
for creating a new commit filling in missing parameters from the
values of an existing commit. As part of this, I also added a new
"sys" API to create a commit using a callback to get the parents.
This allowed me to rewrite all the other commit creation APIs so
that temporary allocations are no longer needed.
This wasn't being tested and since it has a callback, I fixed it
even though the return value of this callback is not treated like
any of the other callbacks in the API.
The callback to supply data chunks could return a negative value
to stop creation of the blob, but we were neither using GIT_EUSER
nor propagating the return value. This makes things use the new
behavior of returning the negative value back to the user.
This changes the behavior of callbacks so that the callback error
code is not converted into GIT_EUSER and instead we propagate the
return value through to the caller. Instead of using the
giterr_capture and giterr_restore functions, we now rely on all
functions to pass back the return value from a callback.
To avoid having a return value with no error message, the user
can call the public giterr_set_str or some such function to set
an error message. There is a new helper 'giterr_set_callback'
that functions can invoke after making a callback which ensures
that some error message was set in case the callback did not set
one.
In places where the sign of the callback return value is
meaningful (e.g. positive to skip, negative to abort), only the
negative values are returned back to the caller, obviously, since
the other values allow for continuing the loop.
The hardest parts of this were in the checkout code where positive
return values were overloaded as meaningful values for checkout.
I fixed this by adding an output parameter to many of the internal
checkout functions and removing the overload. This added some
code, but it is probably a better implementation.
There is some funkiness in the network code where user provided
callbacks could be returning a positive or a negative value and
we want to rely on that to cancel the loop. There are still a
couple places where an user error might get turned into GIT_EUSER
there, I think, though none exercised by the tests.
