|
72df1cd8
|
2019-12-03T19:01:00
|
|
test: improve badname verification test
The name of the `add_invalid_filename` function suggests that we
_want_ to add an invalid filename. Rename the function to show that
we expect to _fail_ to add the invalid filename.
|
|
|
f3b28604
|
2019-12-03T18:57:16
|
|
test: ensure treebuilder validate new protection rules
Ensure that the new protection around .git::$INDEX_ALLOCATION rules are
enabled for using the treebuilder when core.protectNTFS is set.
|
|
|
336991db
|
2019-12-03T18:56:31
|
|
test: ensure index adds validate new protection rules
Ensure that the new protection around .git::$INDEX_ALLOCATION rules are
enabled for adding to the index when core.protectNTFS is set.
|
|
|
a3cbd204
|
2019-12-03T18:49:23
|
|
test: improve badname verification test
The name of the `write_invalid_filename` function suggests that we
_want_ to write an invalid filename. Rename the function to show that
we expect to _fail_ to write the invalid filename.
|
|
64c612cc
|
2019-09-18T15:25:02
|
|
Protect against 8.3 "short name" attacks also on Linux/macOS
The Windows Subsystem for Linux (WSL) is getting increasingly popular,
in particular because it makes it _so_ easy to run Linux software on
Windows' files, via the auto-mounted Windows drives (`C:\` is mapped to
`/mnt/c/`, no need to set that up manually).
Unfortunately, files/directories on the Windows drives can be accessed
via their _short names_, if that feature is enabled (which it is on the
`C:` drive by default).
Which means that we have to safeguard even our Linux users against the
short name attacks.
Further, while the default options of CIFS/SMB-mounts seem to disallow
accessing files on network shares via their short names on Linux/macOS,
it _is_ possible to do so with the right options.
So let's just safe-guard against short name attacks _everywhere_.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
d29d4de2
|
2019-09-18T15:08:56
|
|
cl_git_fail: do not report bogus error message
When we expect a checkout operation to fail, but it succeeds, we
actually do not want to see the error messages that were generated in
the meantime for errors that were handled gracefully by the code (e.g.
when an object could not be found in a pack: in this case, the next
backend would have been given a chance to look up the object, and
probably would have found it because the checkout succeeded, after all).
Which means that in the specific case of `cl_git_fail()`, we actually
want to clear the global error state _after_ evaluating the command: we
know that any still-available error would be bogus, seeing as the
command succeeded (unexpectedly).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
b8464342
|
2019-12-03T17:47:31
|
|
path: rename function that detects end of filename
The function `only_spaces_and_dots` used to detect the end of the
filename on win32. Now we look at spaces and dots _before_ the end of
the string _or_ a `:` character, which would signify a win32 alternate
data stream.
Thus, rename the function `ntfs_end_of_filename` to indicate that it
detects the (virtual) end of a filename, that any further characters
would be elided to the given path.
|
|
|
e1832eb2
|
2019-09-18T16:33:18
|
|
path: also guard `.gitmodules` against NTFS Alternate Data Streams
We just safe-guarded `.git` against NTFS Alternate Data Stream-related
attack vectors, and now it is time to do the same for `.gitmodules`.
Note: In the added regression test, we refrain from verifying all kinds
of variations between short names and NTFS Alternate Data Streams: as
the new code disallows _all_ Alternate Data Streams of `.gitmodules`, it
is enough to test one in order to know that all of them are guarded
against.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
3f7851ea
|
2019-09-18T14:32:05
|
|
Disallow NTFS Alternate Data Stream attacks, even on Linux/macOS
A little-known feature of NTFS is that it offers to store metadata in
so-called "Alternate Data Streams" (inspired by Apple's "resource
forks") that are copied together with the file they are associated with.
These Alternate Data Streams can be accessed via `<file name>:<stream
name>:<stream type>`.
Directories, too, have Alternate Data Streams, and they even have a
default stream type `$INDEX_ALLOCATION`. Which means that `abc/` and
`abc::$INDEX_ALLOCATION/` are actually equivalent.
This is of course another attack vector on the Git directory that we
definitely want to prevent.
On Windows, we already do this incidentally, by disallowing colons in
file/directory names.
While it looks as if files'/directories' Alternate Data Streams are not
accessible in the Windows Subsystem for Linux, and neither via
CIFS/SMB-mounted network shares in Linux, it _is_ possible to access
them on SMB-mounted network shares on macOS.
Therefore, let's go the extra mile and prevent this particular attack
_everywhere_. To keep things simple, let's just disallow *any* Alternate
Data Stream of `.git`.
This is libgit2's variant of CVE-2019-1352.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
|
|
|
6bd37c34
|
2019-12-04T14:33:14
|
|
Merge pull request #5318 from libgit2/ethomson/release_docs
release.md: note that we do two security releases
|
|
|
ea702a0b
|
2019-12-04T14:25:36
|
|
release.md: note that we do two security releases
Note that for security releases, we update the two most recent major release branches.
|
|
|
ad1548e4
|
2019-12-04T13:51:03
|
|
Merge pull request #5317 from csware/size_t
MSVC: Fix warning C4133 on x64: "function": Incompatible types - from "unsigned long *" to "size_t *"
|
|
bdf96512
|
2019-12-03T21:17:30
|
|
MSVC: Fix warning C4133 on x64: "function": Incompatible types - from "unsigned long *" to "size_t *"
Signed-off-by: Sven Strickroth <email@cs-ware.de>
|
|
|
4fe52df6
|
2019-12-03T18:26:24
|
|
Merge pull request #5316 from libgit2/ethomson/publish_docs_on_master
ci: only push docs from the libgit2/libgit2 repo
|
|
|
9c30fbed
|
2019-12-03T18:09:03
|
|
ci: only push docs from the libgit2/libgit2 repo
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.
|
|
|
36bfc27a
|
2019-12-01T14:38:54
|
|
Merge pull request #5314 from pks-t/pks/dll-main-removal
global: convert to fiber-local storage to fix exit races
|
|
|
d298f9b2
|
2019-12-01T14:11:56
|
|
Merge pull request #5315 from kastiglione/dl/fix-copypaste-in-git_cherrypick_commit-docstring
Fix copy&paste in git_cherrypick_commit docstring
|
|
|
7f6fdb82
|
2019-12-01T14:11:22
|
|
Merge pull request #5312 from pks-t/pks/patch-base85-overflow
patch_parse: fix out-of-bounds reads caused by integer underflow
|
|
b7cf4b9e
|
2019-11-29T14:16:04
|
|
Fix copy&paste in git_cherrypick_commit docstring
|
|
5c6180b5
|
2019-11-29T11:06:11
|
|
global: convert to fiber-local storage to fix exit races
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.
|
|
|
7f20778b
|
2019-11-29T09:14:04
|
|
Merge pull request #5311 from pks-t/pks/clar-trace-warning
tests: fix compiler warning if tracing is disabled
|
|
|
61038425
|
2019-11-29T09:13:33
|
|
Merge pull request #5313 from pks-t/pks/config-invasive
tests: config: only test parsing huge file with GITTEST_INVASIVE_SPEED
|
|
|
361ebbcb
|
2019-11-28T15:36:40
|
|
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.
|
|
|
33e6c402
|
2019-11-28T15:26:36
|
|
patch_parse: fix out-of-bounds reads caused by integer underflow
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.
|
|
|
1d470a71
|
2019-11-28T14:45:15
|
|
tests: fix compiler warning if tracing is disabled
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`.
|
|
|
fb439c97
|
2019-11-28T14:41:58
|
|
Merge pull request #5306 from herrerog/patchid
diff: complete support for git patchid
|
|
|
61176a9b
|
2019-11-28T14:31:16
|
|
Merge pull request #5243 from pks-t/pks/config-optimize-mem
Memory optimizations for config entries
|
|
ece5bb5e
|
2019-11-07T14:10:00
|
|
diff: make patchid computation work with all types of commits.
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>
|
|
|
0b5540b9
|
2019-11-28T13:56:54
|
|
Merge pull request #5307 from palmin/hash_sha256
ssh: include sha256 host key hash when supported
|
|
|
dfea0713
|
2019-11-28T13:51:40
|
|
Merge pull request #5272 from tiennou/examples/cli-ification
Various examples shape-ups
|
|
|
b63ad958
|
2019-11-28T13:19:50
|
|
Merge pull request #5309 from libgit2/ethomson/trace
Improve trace support in tests
|
|
|
0e5243b7
|
2019-11-28T12:42:36
|
|
Merge pull request #5123 from libgit2/ethomson/off_t
Move `git_off_t` to `git_object_size_t`
|
|
|
7198d345
|
2019-11-28T15:12:05
|
|
Merge pull request #5310 from lberk/compat-git-attr-t
Add compat typdef for git_attr_t
|
|
5ace4ccf
|
2019-11-27T22:40:17
|
|
Move deprecated git_attr_t typedef to previous attribute section
|
|
|
3739a15c
|
2019-11-27T21:31:25
|
|
Add attr.h include
|
|
|
aea049b6
|
2019-11-27T20:05:32
|
|
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.
|
|
|
b7f70bc2
|
2019-11-27T12:36:17
|
|
tests: optionally show test execution tracing
Only show test trace execution when the CLAR_TRACE_TESTS environment
variable is set. This reduces the noise during tracing.
|
|
|
85efe896
|
2019-11-27T12:34:10
|
|
tests: display trace level with prefix in tests
|
|
|
625a3a49
|
2019-11-27T12:29:34
|
|
trace: enable tracing by default
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.
|
|
|
7805122b
|
2019-11-27T14:22:27
|
|
Merge pull request #5308 from libgit2/ethomson/cifix
CI Build Updates
|
|
|
6460e8ab
|
2019-06-23T18:13:29
|
|
internal: use off64_t instead of git_off_t
Prefer `off64_t` internally.
|
|
|
05237ee5
|
2019-06-23T17:20:17
|
|
integer: use int64_t's for checks
Use int64_t internally for type visibility.
|
|
|
ee0c8618
|
2019-06-23T17:19:31
|
|
offmap: store off64_t's instead of git_off_t's
Prefer `off64_t` to `git_off_t` internally for visibility.
|
|
|
8be12026
|
2019-06-23T17:09:22
|
|
mmap: use a 64-bit signed type `off64_t` for mmap
Prefer `off64_t` to `git_off_t` for internal visibility.
|
|
|
7e1cc296
|
2019-11-25T13:17:42
|
|
mmap: remove unnecessary assertion
64 bit types are always 64 bit.
|
|
|
c863b3c8
|
2019-11-24T16:49:23
|
|
ci: enable the VALGRIND flag on builds
|
|
|
cb77423f
|
2019-11-24T16:22:31
|
|
valgrind: add valgrind hints in OpenSSL
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.
|
|
|
2ad3eb3e
|
2019-11-24T15:59:26
|
|
valgrind: add suppressions for undefined use
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.
|
|
|
0005c77a
|
2019-11-24T15:49:49
|
|
test: add an azure repos test
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.
|
|
|
b8e00b98
|
2019-11-23T21:17:15
|
|
ci: cache docker layers
Our docker builds are getting expensive, let's cache some of this.
|
|
|
6df3ec4a
|
2019-11-23T21:14:32
|
|
valgrind: suppress libssh2_rsa_sha1_sign leaks
|
|
|
c64b7aaa
|
2019-11-23T20:38:30
|
|
ci: build our own valgrind
The valgrind in the PPA is broken and ignores `--exit-errorcode`.
Build and install our own.
|
|
|
7adc32d5
|
2019-11-23T13:02:29
|
|
valgrind: suppress kexinit leaks
|
|
|
fd831275
|
2019-11-23T12:40:46
|
|
ci: build shared libssh2
|
|
|
84807884
|
2019-11-23T12:40:02
|
|
ci: break dockerfile into stages
Use a multi-stage docker build so that we can cache early stages and not
need to download the apt-provided dependencies during every build (when
only later stages change).
|
|
|
7a3d04dc
|
2019-11-23T12:14:23
|
|
ci: don't delete the apt cache
Deleting the apt cache can be helpful for reducing the size of a
container, but since we don't push it anywhere, it only hinders our
ability to debug problems while working on the container. Keep it.
|
|
|
f592c737
|
2019-11-23T11:55:50
|
|
ci: don't install libssh2 since we build it
|
|
|
5dc1be8d
|
2019-11-23T11:25:56
|
|
valgrind: suppress uninitialized reads in libcrypto
libcrypto will read uninitialized memory as entropy. Suppress warnings
from this behavior.
|
|
|
767990e9
|
2019-11-23T11:25:38
|
|
ci: show distribution information
The lsb-release command is missing on our images; just show the
information from the file instead of relying on it.
|
|
|
91ba65af
|
2019-11-23T10:58:38
|
|
ci: provide a default for xcode generator
Provide a sane default for `CMAKE_GENERATOR` in the build script so that
it can be invoked without having to set that in the environment.
|
|
|
f94c9276
|
2019-10-27T22:20:38
|
|
example: use `git_object_size_t` for object size
|
|
|
4dffa295
|
2019-06-23T18:09:00
|
|
blame: use a size_t for the buffer
|
|
|
6c13cf6d
|
2019-11-22T15:18:54
|
|
filestamp: use `uint64_t` for object size
Instead of using a signed type (`off_t`) use an unsigned `uint64_t` for
the size of the files.
|
|
|
fefefd1d
|
2019-06-23T16:42:14
|
|
odb: use `git_object_size_t` for object size
Instead of using a signed type (`off_t`) use a new `git_object_size_t`
for the sizes of objects.
|
|
|
fb2198db
|
2019-06-23T16:23:59
|
|
futils_filesize: use `uint64_t` for object size
Instead of using a signed type (`off_t`) use `uint64_t` for the maximum
size of files.
|
|
|
4334b177
|
2019-06-23T15:43:38
|
|
blob: use `git_object_size_t` for object size
Instead of using a signed type (`off_t`) use a new `git_object_size_t`
for the sizes of objects.
|
|
|
bed9fc6b
|
2019-06-23T15:16:47
|
|
odb: use `git_object_size_t` for object size
Instead of using a signed type (`off_t`) use a new `git_object_size_t`
for the sizes of objects.
|
|
|
9b04d0be
|
2019-11-22T15:04:09
|
|
types: introduce `git_object_size_t`
Introduce `git_object_size_t`, an unsigned type that we can use for the
maximum size of git objects.
|
|
48c3f7e1
|
2019-11-20T11:21:14
|
|
ssh: include sha256 host key hash when supported
|
|
|
048e94ad
|
2019-11-07T14:13:14
|
|
patch_parse: correct parsing of patch containing not shown binary data.
When not shown binary data is added or removed in a patch, patch parser
is currently returning 'error -1 - corrupt git binary header at line 4'.
Fix it by correctly handling case where binary data is added/removed.
Signed-off-by: Gregory Herrero <gregory.herrero@oracle.com>
|
|
|
b921964b
|
2019-11-07T13:08:51
|
|
diff_print: add support for GIT_DIFF_FORMAT_PATCH_ID.
Git is generating patch-id using a stripped down version of a patch
where hunk header and index information are not present.
Signed-off-by: Gregory Herrero <gregory.herrero@oracle.com>
|
|
|
accd7848
|
2019-11-07T13:02:38
|
|
diff_print: add a new 'print_index' flag when printing diff.
Add a new 'print_index' flag to let the caller decide whether or not
'index <oid>..<oid>' should be printed.
Since patch id needs not to have index when hashing a patch, it will be
useful soon.
Signed-off-by: Gregory Herrero <gregory.herrero@oracle.com>
|
|
|
47dd665a
|
2019-11-16T15:21:56
|
|
Merge pull request #5303 from pks-t/pks/patch-path-in-body-only
patch_parse: use paths from "---"/"+++" lines for binary patches
|
|
|
cb6bc6f2
|
2019-11-16T15:17:54
|
|
Merge pull request #5285 from pcpthm/winhttp-308
Follow 308 redirect in WinHTTP transport
|
|
|
541b8fc5
|
2019-11-16T15:12:52
|
|
Merge pull request #5302 from tiennou/fix/p_lstat-errno
fileops: correct error return on p_lstat failures when mkdir
|
|
|
de7659cc
|
2019-11-10T18:44:56
|
|
patch_parse: use paths from "---"/"+++" lines for binary patches
For some patches, it is not possible to derive the old and new file
paths from the patch header's first line, most importantly when they
contain spaces. In such a case, we derive both paths from the "---" and
"+++" lines, which allow for non-ambiguous parsing. We fail to use these
paths when parsing binary patches without data, though, as we always
expect the header paths to be filled in.
Fix this by using the "---"/"+++" paths by default and only fall back to
header paths if they aren't set. If neither of those paths are set, we
just return an error. Add two tests to verify this behaviour, one of
which would have previously caused a segfault.
|
|
0c2b0206
|
2019-11-09T09:41:13
|
|
fileops: correct error return on p_lstat failures when mkdir
IIRC I got a strange return once from lstat, which translated in a weird
error class/message being reported. As a safety measure, enforce a -1 return in
that case.
|
|
|
01ea911b
|
2019-11-06T06:04:55
|
|
Merge pull request #5299 from pks-t/pks/config-mem-snapshots
config_mem: implement support for snapshots
|
|
|
a3d8a437
|
2019-11-06T06:04:37
|
|
Merge pull request #5298 from pks-t/pks/patch-whitespace-only-paths
patch_parse: fix segfault when header path contains whitespace only
|
|
|
fe42557a
|
2019-11-06T11:08:52
|
|
examples: buff up rev-list by adding OID support
This allows the example to be used as a quick revwalk test harness.
|
|
|
313908f9
|
2019-11-06T11:08:49
|
|
examples: normalize decls and usage of options structs
|
|
|
4a4ad2bc
|
2019-11-06T11:08:45
|
|
examples: add comments to add.c
|
|
|
d4a593ef
|
2019-11-06T11:17:52
|
|
examples: modernize add code
|
|
|
c9a09b91
|
2019-11-06T11:16:47
|
|
examples: extract argument conversion helper
|
|
|
204a464f
|
2019-11-06T11:08:39
|
|
examples: fixup for-each-ref style
|
|
|
c924f36a
|
2019-11-06T11:08:35
|
|
examples: keep track of whether we processed a "--" arg
|
|
|
025a9357
|
2019-11-06T11:08:30
|
|
examples: move "args" to its own header
|
|
|
745ccc8a
|
2019-11-06T11:08:26
|
|
examples: remove duplicate includes from common.c
|
|
|
cd5e33fb
|
2019-11-06T11:08:23
|
|
global: DRY includes of assert.h
|
|
|
882220bf
|
2019-11-06T11:08:19
|
|
examples: add missing include barriers
|
|
|
4867523e
|
2019-11-06T11:08:15
|
|
examples: add *.h files to IDEs
|
|
|
146e5bf7
|
2019-11-06T07:27:35
|
|
config_mem: implement support for snapshots
Similar as in commit dadbb33b6 (Fix crash if snapshotting a
config_snapshot, 2019-11-01), let's implement snapshots for in-memory
configuration entries. As this deletes more code than it adds, it
doesn't make any sense to not allow for this and allows users to treat
config backends mostly the same.
|
|
|
de543e29
|
2019-11-05T22:44:27
|
|
patch_parse: fix segfault when header path contains whitespace only
When parsing header paths from a patch, we reject any patches with empty
paths as malformed patches. We perform the check whether a path is empty
before sanitizing it, though, which may lead to a path becoming empty
after the check, e.g. if we have trimmed whitespace. This may lead to a
segfault later when any part of our patching logic actually references
such a path, which may then be a `NULL` pointer.
Fix the issue by performing the check after sanitizing. Add tests to
catch the issue as they would have produced a segfault previosuly.
|
|
|
b7dcea04
|
2019-09-26T15:06:12
|
|
config_entries: micro-optimize storage of multivars
Multivars are configuration entries that have many values for the same
name; we can thus micro-optimize this case by just retaining the name of
the first configuration entry and freeing all the others, letting them
point to the string of the first entry.
The attached test case is an extreme example that demonstrates this. It
contains a section name that is approximately 500kB in size with 20.000
entries "a=b". Without the optimization, this would require at least
20000*500kB bytes, which is around 10GB. With this patch, it only
requires 500kB+20000*1B=20500kB.
The obvious culprit here is the section header, which we repeatedly
include in each of the configuration entry's names. This makes it very
easier for an adversary to provide a small configuration file that
disproportionally blows up in memory during processing and is thus a
feasible way for a denial-of-service attack. Unfortunately, we cannot
fix the root cause by e.g. having a separate "section" field that may
easily be deduplicated due to the `git_config_entry` structure being
part of our public API. So this micro-optimization is the best we can do
for now.
|
|
|
62320860
|
2019-09-26T14:43:19
|
|
config_entries: only keep track of a single entry list
Whenever adding a configuration entry to the config entries structure,
we allocate two list heads:
- The first list head is added to the global list of config entries
in order to be able to iterate over configuration entries in the
order they were originally added.
- The second list head is added to the map of entries in order to
efficiently look up an entry by its name. If no entry with the
same name exists in the map, then we add the new entry to the map
directly. Otherwise, we append the new entry's list head to the
pre-existing entry's list in order to keep track of multivars.
While the former usecase is perfectly sound, the second usecase can be
optimized. The only reason why we keep track of multivar entries in
another separate list is to be able to determine whether an entry is
unique or not by seeing whether its `next` pointer is set. So we keep
track of a complete list of multivar entries just to have a single bit
of information of whether it has other multivar entries with the same
entry name.
We can completely get rid of this secondary list by just adding a
`first` field to the list structure itself. When executing
`git_config_entries_append`, we will then simply check whether the
configuration map already has an entry with the same name -- if so, we
will set the `first` to zero to indicate that it is not the initial
entry anymore. Instead of a second list head in the map, we can thus now
directly store the list head of the first global list inside of the map
and just refer to that bit.
Note that the more obvious solution would be to store a `unique` field
instead of a `first` field. But as we will only ever inspect the `first`
field of the _last_ entry that has been moved into the map, these are
semantically equivalent in that case.
Having a `first` field also allows for a minor optimization: for
multivar values, we can free the `name` field of all entries that are
_not_ first and have them point to the name of the first entry instead.
|
|
|
8a88701e
|
2019-09-26T13:37:18
|
|
config_entries: mark local functions as static
Some functions which are only used in "config_entries.c" are not marked
as static, which is being fixed by this very commit.
|
|
|
5d773a18
|
2019-11-05T13:04:10
|
|
Merge pull request #5282 from pks-t/pks/config-file-iterator-race
config_file: fix race when creating an iterator
|
|
|
56b203a5
|
2019-10-24T12:20:27
|
|
config_file: keep reference to config entries when creating iterator
When creating a configuration file iterator, then we first refresh the
backend and then afterwards duplicate all refreshed configuration
entries into the iterator in order to avoid seeing any concurrent
modifications of the entries while iterating. The duplication of entries
is not guarded, though, as we do not increase the refcount of the
entries that we duplicate right now. This opens us up for a race, as
another thread may concurrently refresh the repository configuration and
thus swap out the current set of entries. As we didn't increase the
refcount, this may lead to the entries being free'd while we iterate
over them in the first thread.
Fix the issue by properly handling the lifecycle of the backend's
entries via `config_file_entries_take` and `git_config_entries_free`,
respectively.
|
|
|
0927156a
|
2019-10-24T12:32:11
|
|
config_file: refactor taking entries ref to return an error code
The function to take a reference to the config file's config entries
currently returns the reference via return value. Due to this, it's
harder than necessary to integrate into our typical coding style, as one
needs to make sure that a proper error code is set before erroring out
from the caller. This bites us in `config_file_delete`, where we call
`goto out` directly when `config_file_entries_take` returns `NULL`, but
we actually forget to set up the error code and thus return success.
Fix the issue by refactoring the function to return an error code and
pass the reference via an out-pointer.
|
|
|
db301087
|
2019-10-24T12:17:02
|
|
config_file: remove unused includes
|