Re: RFR: JDK-8242888: Convert dynamic proxy to hidden classes
On Sun, 17 Apr 2022 16:17:30 GMT, liach wrote:
> Convert dynamic proxies to hidden classes. Modifies the serialization of
> proxies (requires change in "Java Object Serialization Specification"). Makes
> the proxies hidden in stack traces. Removes duplicate logic in proxy building.
>
> The main compatibility changes and their rationales are:
> 1. Modification to the serialization specification: In the "An instance of
> the class is allocated... The contents restored appropriately" section, I
> propose explicitly state that handling of proxies are unspecified as to allow
> implementation freedom, though I've seen deliberate attempts for proxies to
> implement interfaces with `readResolve` in order to control their
> serialization behavior.
>- This is for the existing generated constructor accessor is
> bytecode-based, which cannot refer to hidden classes.
>- An alternative is to preserve the behavior, where the serialization
> constructor calls `invokespecial` on the closest serializable superclass'
> no-arg constructor, like in #1830 by @DasBrain.
>- My rationale against preservation is such super calls are unsafe and
> should be discouraged in the long term. Calling the existing constructor with
> a dummy argument, as in my implementation, would be more safe.
> 2. The disappearance of proxies in stack traces.
>- Same behavior exists in lambda expressions: For instance, in
> `((Runnable) () -> { throw new Error(); }).run();`, the `run` method,
> implemented by the lambda, will not appear in the stack trace, and isn't too
> problematic.
>
> A summary of the rest of the changes:
> 1. Merged the two passes of determining module and package of the proxy into
> one. This reduced a lot of code and allowed anchor class (for hidden class
> creation) selection be done together as well.
> 2. Exposed internal API for obtaining a full-privileged lookup to the rest of
> `java.base`. This API is intended for implementation of legacy (pre
> `MethodHandles.Lookup`) caller sensitive public APIs so they don't need more
> complex tricks to obtain proper permissions as lookups.
> 3. Implements [8229959](https://bugs.openjdk.java.net/browse/JDK-8229959):
> passes methods computed by proxy generator as class data to the hidden proxy
> class to reduce generated proxy class size and improve performance.
>
> In addition, since this change is somewhat large, should we keep the old
> proxy generator as well and have it toggled through a command-line flag (like
> the old v49 proxy generator or the old reflection implementation)?
>
> Please feel free to comment or review. This change definitely requires a CSR,
> but I have yet to determine what specifications should be changed.
I now think that this should be split into probably four separate patches:
1. Merge the 2 passes of scanning interfaces, and store the
exported/non-exported package within the classloader value info for dynamic
modules
2. Use the JLIA access to obtain lookup than spin an accessor in every proxy
class
3. Update serialization constructor generation to use method handles (as in
#1830) for hidden classes, don't change the serialization specs for now, as it
affects security.
And finally, convert proxies to hidden classes like in this pr. There will be
an option (most likely command line) to restore old proxies. And in the new
proxies, method objects can be passed as class data, like in the current patch.
The jshell test change will be in the final conversion patch.
-
PR: https://git.openjdk.java.net/jdk/pull/8278
RFR: 8283022: com/sun/crypto/provider/Cipher/AEAD/GCMBufferTest.java failing with -Xcomp after 8273297
When input length provided to the intrinsic is 8192, only 7680 bytes are processed as the intrinsic operates on multiples of 768 bytes. In implGCMCrypt(ByteBuffer src, ByteBuffer dst) method, dst.put(bout, 0, PARALLEL_LEN) statement caused the ciphertext mismatch as PARALLEL_LEN was set to 8192. Since the intrinsic only processed 7680 bytes, the rest output was incorrect. - Commit messages: - Changed PARALLEL_LEN to 7680 as intrinsic processes multiples of 768 bytes Changes: https://git.openjdk.java.net/jdk/pull/8280/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk=8280=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8283022 Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod Patch: https://git.openjdk.java.net/jdk/pull/8280.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/8280/head:pull/8280 PR: https://git.openjdk.java.net/jdk/pull/8280
RFR: 8284935: Improve debug in java.security.jgss
Hi, May I have the simple update reviewed? In the java.security.jgss module implementation, some of the debug information could be generated even if debugging is not enabled, which is not resource friendly. Thanks, Xuelei - Commit messages: - 8284935: Improve debug in java.security.jgss Changes: https://git.openjdk.java.net/jdk/pull/8279/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk=8279=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8284935 Stats: 98 lines in 7 files changed: 40 ins; 16 del; 42 mod Patch: https://git.openjdk.java.net/jdk/pull/8279.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/8279/head:pull/8279 PR: https://git.openjdk.java.net/jdk/pull/8279
Re: RFR: JDK-8242888: Convert dynamic proxy to hidden classes
We re-implemented a subset of Java Serialization, prior to the creation
Java serialization filters. Field types are read ahead from the stream
and invariant's validated during construction for failure atomicity (a
special constructor signature is used for deserialization), there are
also stream limits that require periodical stream resets, to avoid
million laugh style attacks. Also the source of the serialized data must
be authenticated before permission is granted for parsing serial data.
We also removed the ability to deserialize object graphs with circular
links.
This is used for an Remove invocation framework called JERI (Jini
extensible remote invocation). Serialization has been given a public
API to allow extensibility, that is to allow other serialization
protocols to be used via configuration, using a Serialization layer.
Service and service discovery architecture makes use of JERI, for
marshaling object parameters securely over the network, for remote
method calls on services, via remote proxy's.
The way that JERI serializes proxy's has changed significantly since
Jini, instead of marshaling proxy's, a bootstrap proxy (local code only)
is marshaled first, the client first authenticates the connection, the
bootstrap proxy is used to provide information for dynamic downloading
of any required jar files, or wiring of dependencies prior to the
unmarshaling of the service proxy. Unlike Java RMI, which uses
RMIClassLoading for class resolution, each Jeri Endpoint is assigned a
ClassLoader for class resolution during deserialization / object
unmarshaling, this solves the OSGi deserialzation class resolution problem.
https://github.com/pfirmstone/JGDMS/blob/trunk/JGDMS/jgdms-platform/src/main/java/org/apache/river/api/io/ProxySerializer.java
https://github.com/pfirmstone/JGDMS/blob/c1edf5892306f24f8f97f459f499dec54984b08f/JGDMS/jgdms-platform/src/main/java/org/apache/river/api/io/AtomicMarshalInputStream.java#L2263
https://github.com/pfirmstone/JGDMS/blob/c1edf5892306f24f8f97f459f499dec54984b08f/JGDMS/jgdms-platform/src/main/java/org/apache/river/api/io/AtomicMarshalInputStream.java#L853
Would be nice to do some testing with the changes.
--
Regards,
Peter Firmstone
0498 286 363
Zeus Project Services Pty Ltd.
On 18/04/2022 2:24 am, liach wrote:
Convert dynamic proxies to hidden classes. Modifies the serialization of proxies
(requires change in "Java Object Serialization Specification"). Makes the
proxies hidden in stack traces. Removes duplicate logic in proxy building.
The main compatibility changes and their rationales are:
1. Modification to the serialization specification: In the "An instance of the class
is allocated... The contents restored appropriately" section, I propose explicitly
state that handling of proxies are unspecified as to allow implementation freedom, though
I've seen deliberate attempts for proxies to implement interfaces with `readResolve` in
order to control their serialization behavior.
- This is for the existing generated constructor accessor is
bytecode-based, which cannot refer to hidden classes.
- An alternative is to preserve the behavior, where the serialization
constructor calls `invokespecial` on the closest serializable superclass'
no-arg constructor, like in #1830 by @DasBrain.
- My rationale against preservation is such super calls are unsafe and
should be discouraged in the long term. Calling the existing constructor with a
dummy argument, as in my implementation, would be more safe.
2. The disappearance of proxies in stack traces.
- Same behavior exists in lambda expressions: For instance, in `((Runnable) ()
-> { throw new Error(); }).run();`, the `run` method, implemented by the
lambda, will not appear in the stack trace, and isn't too problematic.
A summary of the rest of the changes:
1. Merged the two passes of determining module and package of the proxy into
one. This reduced a lot of code and allowed anchor class (for hidden class
creation) selection be done together as well.
2. Exposed internal API for obtaining a full-privileged lookup to the rest of
`java.base`. This API is intended for implementation of legacy (pre
`MethodHandles.Lookup`) caller sensitive public APIs so they don't need more
complex tricks to obtain proper permissions as lookups.
3. Implements [8229959](https://bugs.openjdk.java.net/browse/JDK-8229959):
passes methods computed by proxy generator as class data to the hidden proxy
class to reduce generated proxy class size and improve performance.
In addition, since this change is somewhat large, should we keep the old proxy
generator as well and have it toggled through a command-line flag (like the old
v49 proxy generator or the old reflection implementation)?
Please feel free to comment or review. This change definitely requires a CSR,
but I have yet to determine what specifications should be changed.
-
Commit messages:
- Change proxy
RFR: JDK-8242888: Convert dynamic proxy to hidden classes
Convert dynamic proxies to hidden classes. Modifies the serialization of
proxies (requires change in "Java Object Serialization Specification"). Makes
the proxies hidden in stack traces. Removes duplicate logic in proxy building.
The main compatibility changes and their rationales are:
1. Modification to the serialization specification: In the "An instance of the
class is allocated... The contents restored appropriately" section, I propose
explicitly state that handling of proxies are unspecified as to allow
implementation freedom, though I've seen deliberate attempts for proxies to
implement interfaces with `readResolve` in order to control their serialization
behavior.
- This is for the existing generated constructor accessor is bytecode-based,
which cannot refer to hidden classes.
- An alternative is to preserve the behavior, where the serialization
constructor calls `invokespecial` on the closest serializable superclass'
no-arg constructor, like in #1830 by @DasBrain.
- My rationale against preservation is such super calls are unsafe and
should be discouraged in the long term. Calling the existing constructor with a
dummy argument, as in my implementation, would be more safe.
2. The disappearance of proxies in stack traces.
- Same behavior exists in lambda expressions: For instance, in `((Runnable)
() -> { throw new Error(); }).run();`, the `run` method, implemented by the
lambda, will not appear in the stack trace, and isn't too problematic.
A summary of the rest of the changes:
1. Merged the two passes of determining module and package of the proxy into
one. This reduced a lot of code and allowed anchor class (for hidden class
creation) selection be done together as well.
2. Exposed internal API for obtaining a full-privileged lookup to the rest of
`java.base`. This API is intended for implementation of legacy (pre
`MethodHandles.Lookup`) caller sensitive public APIs so they don't need more
complex tricks to obtain proper permissions as lookups.
3. Implements [8229959](https://bugs.openjdk.java.net/browse/JDK-8229959):
passes methods computed by proxy generator as class data to the hidden proxy
class to reduce generated proxy class size and improve performance.
In addition, since this change is somewhat large, should we keep the old proxy
generator as well and have it toggled through a command-line flag (like the old
v49 proxy generator or the old reflection implementation)?
Please feel free to comment or review. This change definitely requires a CSR,
but I have yet to determine what specifications should be changed.
-
Commit messages:
- Change proxy serialization to work with hidden classes (require spec update)
- Formatting
- Move to hidden class and methods in class data
- Implement anchors and remove proxyClassLookup factory
Changes: https://git.openjdk.java.net/jdk/pull/8278/files
Webrev: https://webrevs.openjdk.java.net/?repo=jdk=8278=00
Issue: https://bugs.openjdk.java.net/browse/JDK-8242888
Stats: 400 lines in 6 files changed: 122 ins; 222 del; 56 mod
Patch: https://git.openjdk.java.net/jdk/pull/8278.diff
Fetch: git fetch https://git.openjdk.java.net/jdk pull/8278/head:pull/8278
PR: https://git.openjdk.java.net/jdk/pull/8278
Re: RFR: 8284926: Share the certificate NamedGroup in SignatureScheme::getSignerOfPreferableAlgorithm
On Sat, 16 Apr 2022 13:59:25 GMT, John Jiang wrote: > It would not to generate the certificate's ECParameterSpec and NamedGroup > multiple times in method `SignatureScheme::getSignerOfPreferableAlgorithm`. src/java.base/share/classes/sun/security/ssl/SignatureScheme.java line 479: > 477: NamedGroup namedGroup = params != null > 478: ? NamedGroup.valueOf(params) : null; > 479: X509Possession.java might be a better place for the update, by caching the named group of the private key. - PR: https://git.openjdk.java.net/jdk/pull/8271
RFR: 8284933: Improve debug in jdk.crypto.cryptoki
Hi, May I have the simple update reviewed? In the jdk.crypto.cryptoki module implementation, some of the debug information could be calculated even if the debug is not enabled, which is not resource friendly. Thanks, Xuelei - Commit messages: - 8284933: Improve debug in jdk.crypto.cryptoki Changes: https://git.openjdk.java.net/jdk/pull/8277/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk=8277=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8284933 Stats: 34 lines in 1 file changed: 20 ins; 2 del; 12 mod Patch: https://git.openjdk.java.net/jdk/pull/8277.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/8277/head:pull/8277 PR: https://git.openjdk.java.net/jdk/pull/8277
