DFS block cache: fix lock issue and support parallel index loading
This change is a fix to http://git.eclipse.org/r/c/jgit/jgit/+/183562
that was reverted in http://git.eclipse.org/r/c/jgit/jgit/+/184978
due to deadlocks. Separate locks in DfsBlockFile are removed to rely
on getting value from DfsBlockCache with region locking in place.
With this change bitmap index creation is not blocked on index and
reverse index full initialization in DfsPackFile. Now bitmap index
and index could be read from storage in parallel in separate threads.
Matthias Sohn [Fri, 15 Oct 2021 21:10:12 +0000 (23:10 +0200)]
Merge branch 'stable-5.13'
* stable-5.13:
Fix missing peel-part in lsRefsV2 for loose annotated tags
Fix RevWalk.getMergedInto() ignores annotated tags
Optimize RevWalk.getMergedInto()
reftable: drop code for truncated reads
reftable: pass on invalid object ID in conversion
Update eclipse-jarsigner-plugin to 1.3.2
Fix running benchmarks from bazel
Update eclipse-jarsigner-plugin to 1.3.2
Add org.bouncycastle.bcutil to p2 repository
Matthias Sohn [Fri, 15 Oct 2021 20:58:21 +0000 (22:58 +0200)]
Merge branch 'stable-5.12' into stable-5.13
* stable-5.12:
Fix missing peel-part in lsRefsV2 for loose annotated tags
Fix RevWalk.getMergedInto() ignores annotated tags
Optimize RevWalk.getMergedInto()
reftable: drop code for truncated reads
reftable: pass on invalid object ID in conversion
Update eclipse-jarsigner-plugin to 1.3.2
Fix running benchmarks from bazel
Update eclipse-jarsigner-plugin to 1.3.2
Matthias Sohn [Fri, 15 Oct 2021 20:48:01 +0000 (22:48 +0200)]
Merge branch 'stable-5.11' into stable-5.12
* stable-5.11:
Fix missing peel-part in lsRefsV2 for loose annotated tags
reftable: drop code for truncated reads
reftable: pass on invalid object ID in conversion
Update eclipse-jarsigner-plugin to 1.3.2
Fix running benchmarks from bazel
Update eclipse-jarsigner-plugin to 1.3.2
Matthias Sohn [Fri, 15 Oct 2021 20:45:18 +0000 (22:45 +0200)]
Merge branch 'stable-5.10' into stable-5.11
* stable-5.10:
Fix missing peel-part in lsRefsV2 for loose annotated tags
reftable: drop code for truncated reads
reftable: pass on invalid object ID in conversion
Update eclipse-jarsigner-plugin to 1.3.2
Fix running benchmarks from bazel
Update eclipse-jarsigner-plugin to 1.3.2
Matthias Sohn [Fri, 15 Oct 2021 20:32:00 +0000 (22:32 +0200)]
Merge branch 'stable-5.9' into stable-5.10
* stable-5.9:
Fix missing peel-part in lsRefsV2 for loose annotated tags
reftable: drop code for truncated reads
reftable: pass on invalid object ID in conversion
Update eclipse-jarsigner-plugin to 1.3.2
Fix running benchmarks from bazel
Update eclipse-jarsigner-plugin to 1.3.2
Unfortunately, the existing UploadPackTest didn't reveal this issue
although it provided the test case needed to do so: testV2LsRefsPeel.
This is because The UploadPackTest uses InMemoryRepository which
internally uses Dfs* implementations. The issue is only reproducible
when using the FileRepository.
It is a non-trivial task to refactor the UploadPackTest to work against
both InMemoryRepository and FileRepository and this change is not trying
to do that. This change creates a new test:
UploadPackLsRefsFileRepositoryTest and copies the necesssary code from
the UploadPackTest.
DFS block cache: allow multiple passes for blocks before eviction
Let certain pack extensions that are expensive to load from storage
(e.g. pack index, bitmap index) stay in DFS block cache longer than
others by overriding default cache count through DfsBlockCacheConfig
Don't change default behavior when cache override map is empty. Use int
cacheCount instead of boolean hot for Ref<T>
Find out which branches is 2 merged into:
After we calculated that master contains 2, we can mark 6 as TEMP_MARK
to avoid unwanted walks.
When we want to figure out if 2 is merge into the topic, the traversal
path would be [7, 6] instead of [7, 6, 5, 3, 2].
Test:
This change can significantly improve performance for tags.
On a copy of the Linux repository, the command 'git tag --contains
<commit>' had the following performance improvement:
The current version ignores tags, even though the tag is a type of
the ref.
Follow-up commits I'll fix it.
Change-Id: Ie6295ca4d16070499912af462239e679a97cce47 Signed-off-by: kylezhao <kylezhao@tencent.com> Reviewed-by: Christian Halstrick <christian.halstrick@sap.com> Reviewed-by: Martin Fick <mfick@codeaurora.org>
Matthias Sohn [Wed, 29 Sep 2021 00:10:34 +0000 (02:10 +0200)]
Delete old target platforms and corresponding Orbit releases
We updated the baseline target platform to jgit-4.17 corresponding to
Eclipse 4.17 (2020-09). Delete all target platforms older than this
version and the corresponding Orbit releases.
Matthias Sohn [Wed, 29 Sep 2021 00:03:02 +0000 (02:03 +0200)]
Update tycho to 2.5.0 and target platform to jgit-4.17
jgit-4.17 corresponds to Eclipse 2020-09, which is the first Eclipse
version that required Java 11, and which is JGit's new baseline as of
JGit 6.0.
Remove pack200 which is deprecated in Java 11 and isn't supported by
tycho 2.5.0 anymore. It was specified in JSR 200 (J2SE 1.5) [1],
deprecated in JEP 336 (Java SE 11) [2] and removed in JEP 367 (Java SE
14) [3].
Matthias Sohn [Tue, 21 Sep 2021 21:24:47 +0000 (23:24 +0200)]
Enable using JMH annotation processor on Java>=9
See
- https://issues.apache.org/jira/browse/MCOMPILER-369
- https://stackoverflow.com/questions/53927874/jmh-doesnt-run-inside-a-java-module-unable-to-find-the-resource-meta-inf-ben
Matthias Sohn [Mon, 28 Dec 2020 17:51:57 +0000 (18:51 +0100)]
Enable compiler option --release
This ensures the compiler compiles against the public, supported and
documented API for a specific VM version (here 11) [1]. This also means
that
we don't need EE descriptors in Eclipse anymore in order to ensure that
only supported APIs of the selected Java version can be used.
According to [2] if option --release is used --source and --target
options can't be used.
While we are at it also add default value for all new jdt core options
added in Eclipse 4.21.
The reftable format is a block based format, but allows for variably
sized blocks. This obviously happens for reflog blocks (which are zlib
compressed), but is also accepted for index blocks: In the spec, this
is motivated as
To achieve constant O(1) disk seeks for lookups the index must be
a single level, which is permitted to exceed the file's
configured block size, but not the format's max block size of
15.99 MiB.
Hence, when parsing a block, one cannot be sure of its exact size:
after reading a default-size block (eg. 4kb), the block header may
state that the block is in fact larger.
Before, the code would mark the block as `truncated`, noting
// Its OK during sequential scan for an index block to have been
// partially read and be truncated in-memory. This happens when
// the index block is larger than the file's blockSize. Caller
// will break out of its scan loop once it sees the blockType.
This looks like either
* a remnant of never-implemented functionality. There is no reason to
ever sequentially scan an index block.
* alluding to sequential scan of the data blocks before the index
blocks (eg. scanning refs, which ends when we find the first ref index
block, and we can then ignore the index block).
This comment is followed by code that populates the
restartTbl/restartCnt fields relative to the (possibly truncated)
buffer. If the buffer is truncated, this essentially reads garbage,
leading to OOB array access when using the index block.
Fix this by dropping the truncated logic and issuing a second read if
the first read was short.
Add a test.
We have never observed this failure scenario at Google. We use 64kb
blocksize, which requires us to need fewer index entries. The reftable
spec mentions an Android repo of size 36M. With 64kb blocks, that's
just 562 index entries. Even with historical growth, we are long from
requiring an index whose size exceeds a single block.
When adding the analogous test for seeking refs, there was no failure.
This points to another possibility which is that the code tries to
avoid writing large index blocks for refs.
Matthias Sohn [Thu, 9 Sep 2021 12:41:30 +0000 (14:41 +0200)]
Merge branch 'master' into next
* master: (38 commits)
Revert "DFS block cache: Refactor to enable parallel index loading"
GitServlet: allow to override default error handlers
Silence API error for new interface method ProtocolV2Hook#onObjectInfo
transport: add object-info capability
Ignore IllegalStateException if JVM is already shutting down
Update orbit to R20210825222808 for 2021-09
Update spotbugs-maven-plugin to 4.3.0
Update ant to 1.10.11 also in pom.xml
DFS block cache: add additional stats to DfsReaderIoStats
Update Orbit to S20210817231813
[gpg] Better GPG home directory determination
FS: cleanup use of final modifier
Ensure FS#searchPath only selects executable files
RevWalk: getMergedInto's result is wrong on the second call
DFS block cache: Refactor to enable parallel index loading
[test] Create keystore with the keytool of the running JDK
[gpg] Update to Bouncy Castle 1.69
[test] Create keystore with the keytool of the running JDK
[sshd] Minor code clean-up
Support commit.template config property
...
Matthias Sohn [Wed, 1 Sep 2021 14:31:13 +0000 (16:31 +0200)]
Merge branch 'master' into stable-5.13
* master:
GitServlet: allow to override default error handlers
Silence API error for new interface method ProtocolV2Hook#onObjectInfo
transport: add object-info capability
Ignore IllegalStateException if JVM is already shutting down
Update orbit to R20210825222808 for 2021-09
RevWalk: getMergedInto's result is wrong on the second call
Antonio Barone [Wed, 1 Sep 2021 10:04:07 +0000 (12:04 +0200)]
GitServlet: allow to override default error handlers
GitServlet delegates repository access over HTTP to the GitFilter
servlet.
GitServlet, in turn, can be extended by jgit consumers to provide custom
logic when handling such operations.
This is the case, for example, with Gerrit Code Review, which provides
custom behavior with a GitOverHttpServlet [1].
Among possible customizations, the ability of specifying a custom error
handler for UploadPack and ReceivePack was already introduced in
GitFilter by Idd3b87d6b and I9c708aa5a2, respectively.
However the `setUploadPackErrorHandler` and `setReceivePackErrorHandler`
methods were never added to the GitServlet.
Expose the `setUploadPackErrorHandler` and `setReceivePackErrorHandler`
methods to the GitServlet, so that consumers of the jgit library might
specify custom error handlers.
Sometimes it is useful to obtain metadata associated with an object
without the need to first download it locally. This is specially useful
when using partial clones.
This change implements the object-info capability that allows clients to
query the remote server for object metadata (currently only size). This
is a backport of the same capability that was recently added to the Git
project a2ba162cda (object-info: support for retrieving object info,
2021-04-20).
Signed-off-by: Bruno Albuquerque <bga@google.com>
Change-Id: I4dc9828e1c247f08b0976b8810be92d124366165
Matthias Sohn [Thu, 12 Aug 2021 07:28:04 +0000 (09:28 +0200)]
Ignore IllegalStateException if JVM is already shutting down
Trying to register/unregister a shutdown hook when the JVM is already in
shutdown throws an IllegalStateException. Ignore this exception since we
can't do anything about it.
Thomas Wolf [Tue, 10 Aug 2021 21:26:42 +0000 (23:26 +0200)]
[gpg] Better GPG home directory determination
GPG can use customized directories instead of the standard ~/.gnupg or
%APPDATA%\gnupg directories:
* Environment variable GNUPGHOME can define the location.
* On Windows, a registry key may define the location (but this is
deprecated).
* Portable installations may use a directory defined via a file
"gpgconf.ctl".
* GPG programs may take a --homedir command-line argument, which
overrides anything.
Implement handling of environment variable GNUPGHOME. The other ways of
GPG to get its home directory are outside the reach of JGit. Provide a
system property "jgit.gpg.home" that the user can set in such cases.
Do tilde replacement for the system property and for GNUPGHOME.
Note that on VMS, the default directory would be ~/gnupg (without dot).
This is not accounted for, but a user on VMS could now use either the
system property or GNUPGHOME to direct JGit to the right directory.
Bug: 575327
Change-Id: Id5ea04a85d58dba0c0df7a705777630d36042467 Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
DFS block cache: Refactor to enable parallel index loading
With this change bitmap index creation is not blocked on index and
reverse index full initialization in DfsPackFile. Now bitmap index and index could be
read from storage in parallel in separate threads.
Update PackBitmapIndexV1 constructor to get packIndex and reverseIndex
suppliers instead of actual objects. Inner class IdxPositionBitmap was added to initialize pack objects after bitmap
index is fully read from storage. Update DfsPackFile and PackBitmapIndex
accordingly.
Thomas Wolf [Thu, 29 Jul 2021 11:12:40 +0000 (13:12 +0200)]
Merge branch 'stable-5.12'
* stable-5.12:
[test] Create keystore with the keytool of the running JDK
Fix garbage collection failing to delete pack file
ReachabilityCheckerTestCase: fix reachable from self test case
Change-Id: I4b54f4850819736144edb784617ee902f491ffd6 Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
Thomas Wolf [Sun, 25 Jul 2021 13:44:35 +0000 (15:44 +0200)]
[test] Create keystore with the keytool of the running JDK
Call keytool with the absolute path of "java.home". Otherwise a keytool
for a different, maybe even newer Java version might be picked up, and
then the keystore may not be readable by the JVM used to run the tests.
Thomas Wolf [Sun, 25 Jul 2021 13:44:35 +0000 (15:44 +0200)]
[test] Create keystore with the keytool of the running JDK
Call keytool with the absolute path of "java.home". Otherwise a keytool
for a different, maybe even newer Java version might be picked up, and
then the keystore may not be readable by the JVM used to run the tests.
Change-Id: Iea77024947a34267f008847d81312fe0abadc615 Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
Adds functionality to read the git commit.template property. The
template content is read either via a default encoding or, if present,
via encoding specified by i18n.commitEncoding property.
Matthias Sohn [Mon, 19 Jul 2021 10:32:26 +0000 (12:32 +0200)]
Relax version range for hamcrest and assertj
We currently cannot use hamcrest 2.2 and assertj 3.20.2 (which requires
hamcrest 2.2) in egit tests since other Eclipse dependencies (e.g.
swtbot) require hamcrest 1.1.
Hence relax version range for these components in jgit so that jgit
tests also work when using the egit target platform.
Matthias Sohn [Wed, 14 Jul 2021 20:30:06 +0000 (22:30 +0200)]
Update orbit to I20210713220109
update
- org.apache.commons.compress to 1.20.0.v20210713-192
- org.bouncycastle.bcpg to 1.69.0.v20210713-1924
- org.bouncycastle.bcpkix to 1.69.0.v20210713-1924
- org.bouncycastle.bcprov to 1.69.0.v20210713-1924
In bazel build don't expose bouncycastle to org.eclipse.jgit since it's
not used there anymore since code depending on bouncycastle was moved to
org.eclipse.jgit.gpg.bc.
Matthias Sohn [Sun, 11 Jul 2021 12:21:40 +0000 (14:21 +0200)]
Update orbit to I20210711110031
and update
- assertj to 3.20.2.v20210706-1104
- hamcrest to 2.2.0.v20210711-0821
- classes which were in org.hamcrest.core 1.3 and org.hamcrest.library
1.3 were all moved to org.hamcrest in 2.2
- the annotation org.hamcrest.Factory was removed and is no longer
needed
- junit 4.13 requires hamcrest-core and hamcrest-library 1.3 therefore
keep them in the target platform
Thomas Wolf [Tue, 29 Jun 2021 21:08:02 +0000 (23:08 +0200)]
[sshd] Distinguish key type and signature algorithm for host key
Since the introduction of the rsa-sha2-512 and rsa-sha2-256 signature
types, the key type for RSA is no longer automatically the signature
algorithm. We re-order the list for the host key proposal such that
keys we already have are preferred; this minimizes warnings about new
host keys. When doing so, put all of rsa-sha2-512, rsa-sha2-256, and
ssh-rsa at the front, in that order, not just ssh-rsa.
This ensures that we do prefer RSA keys if we already have an RSA host
key, but at the same time we still prefer the stronger signature
algorithms over the weaker and deprecated SHA1-based ssh-rsa signature.
It also helps avoid a bug found in some Github versions where the Github
SSH server uses a rsa-sha2-512 signature even though ssh-rsa was
negotiated.[1]
projects / jgit.git / log