Improve multi-byte access performance when UNALIGNED availability is unknown by Songdoeon · Pull Request #16207 · netty/netty

Songdoeon · 2026-01-31T15:07:47Z

Motivation:

When the JVM cannot report UNALIGNED support (e.g., on some ARM platforms or in restricted environments), Netty falls back to an architecture-based guess. If that guess yields UNALIGNED = false, multi-byte reads (getLong,
getInt, getShort) degrade to byte-by-byte access (8× getByte) even though the JIT could still optimize at runtime. There is currently no distinction between "the platform is known to not support unaligned access" and "we
simply don't know."

Modification:

Introduce UNALIGNED_AVAILABLE flag to distinguish whether UNALIGNED info came from the runtime (true) or from an architecture-based guess (false). When unaligned info is unavailable, use VarHandle instead of 8× getByte so
HotSpot can emit the optimal access strategy at runtime.

PlatformDependent0: add UNALIGNED_AVAILABLE flag; add io.netty.unalignedAccess=true|false|unavailable system property.
PlatformDependent: expose isUnalignedAvailable(); allow VarHandle initialization even when Unsafe is available.
UnsafeByteBufUtil: use VarHandle for multi-byte byte[] operations (getShort/getInt/getLong, setShort/setInt/setLong and their LE variants) when unaligned info is unavailable.
UnpooledUnsafeDirectByteBuf: use VarHandle for multi-byte ByteBuffer operations (_getShort/_getInt/_getLong, _setShort/_setInt/_setLong and their LE variants) when unaligned info is unavailable.
ByteBufUtil: enable SWAR indexOf/lastIndexOf via VarHandle when unaligned info is unavailable.

Three-tier access strategy:

Condition	Strategy	Path
UNALIGNED = true (known supported)	Unsafe direct access	single `Unsafe.getLong()`
UNALIGNED info unavailable (unknown)	VarHandle	JIT chooses optimal access
UNALIGNED = false (known unsupported)	8× getByte	safe byte-wise assembly

Result:

Fixes #15781.

Benchmark (JDK 24.0.1, Apple Silicon, JMH 1.36):

Benchmark Summary

Benchmark	false (8× getByte)	unavailable (VarHandle)	Improvement
getLongDirect	2.144 ns/op	1.764 ns/op	17.7% faster
getLongHeap	2.152 ns/op	1.119 ns/op	48.0% faster

Sorry for the delay in submitting this PR.
If anything looks incorrect or needs adjustment, I will update it promptly.
Thanks for the review.

…known

franz1981 · 2026-01-31T15:47:59Z

Rebase @Songdoeon please 🙏

franz1981 · 2026-01-31T15:50:44Z

This issue has a much wider scope as all multi byes ops are affected so please address these as well and don't tie it to just index of

…known

…m/Songdoeon/netty into Poor_access_performance_UNALIGNED

Songdoeon · 2026-02-01T06:49:12Z

Thanks

updated to cover all multi-byte get/set paths instead of only getLong/indexOf and adjusted the PR description accordingly.

normanmaurer · 2026-02-01T09:06:12Z

buffer/src/main/java/io/netty/buffer/UnpooledUnsafeDirectByteBuf.java

    @Override
    protected void _setLong(int index, long value) {
+        if (!PlatformDependent.isUnaligned() &&
+                !PlatformDependent.isUnalignedAvailable() && PlatformDependent.hasVarHandle()) {


Why not combine these once in a static boolean to simplify ?

private static final USE_VAR_HANDLE = !PlatformDependent.isUnaligned() && !PlatformDependent.isUnalignedAvailable() && PlatformDependent.hasVarHandle(); ... ... if (USE_VAR_HANDLE) { .... }

Thanks, I extracted the repeated condition into a static final boolean USE_VAR_HANDLE
and applied it across all short/int/long get/set paths in this class.

microbench/src/main/java/io/netty/buffer/UnalignedAccessUnavailableBenchmark.java

microbench/src/main/java/io/netty/microbench/search/UnalignedIndexOfBenchmark.java

franz1981 · 2026-02-02T15:10:55Z

buffer/src/main/java/io/netty/buffer/UnpooledUnsafeDirectByteBuf.java

@@ -31,6 +31,11 @@
 */


Let me try to understand what is going on here, please confirm:

Netty detect presence of Unsafe - which will make it to allocate Unsafe-handled buffers

BUT...unalignment is not detected properly, so it can use VarHandle for multi-bytes accesses

This looks a bit odd, but kind-of make sense.
And for SWAR, instead, I see that this benefit the non-Unsafe case too, or I'm misreading it?
Please confirm 🙏
And thanks for this!

Yes that's correct.

Unsafe is used when unaligned support is known. When it cannot be reliably detected, we use VarHandle
for multi-byte accesses so the JVM can choose the optimal strategy instead of forcing the byte-by-byte path.

For SWAR, it’s also enabled in the unknown-unaligned + VarHandle case. In ByteBufUtil:

SWAR_UNALIGNED = UNALIGNED || (!UNALIGNED_AVAILABLE && PlatformDependent.hasVarHandle());

…ndexOfBenchmark.java

…lableBenchmark.java

franz1981

If android is happy, I think this is a great work. Thanks @Songdoeon 💕

normanmaurer

Almost there

buffer/src/main/java/io/netty/buffer/ByteBufUtil.java

buffer/src/main/java/io/netty/buffer/UnsafeByteBufUtil.java

normanmaurer · 2026-02-10T16:36:03Z

@yawkat @chrisvest PTAL as well.

chrisvest

I see others have been confused about the logic of when to do what type of access, and I see that we have 3 separate static final fields controlling this in various places.

I wonder if its possible to collect more of that decision logic in PlatformDependent, so it's only in one place, and then also annotate it with a comment that explain how we decide on the different approaches.

buffer/src/main/java/io/netty/buffer/UnsafeByteBufUtil.java

common/src/main/java/io/netty/util/internal/PlatformDependent.java

Songdoeon · 2026-02-14T08:11:00Z

Thanks for the review! I've simplified the logic based on your feedback:

Removed the isUnalignedAvailable() concept — VarHandle is now always used as a fallback when UNALIGNED is false.
Centralized decision logic into PlatformDependent with two methods: useVarHandleForMultiByteAccess() and canUnalignedAccess().
canUnalignedAccess() covers both native unaligned support and VarHandle-based access where the JVM handles alignment internally.

chrisvest · 2026-02-16T17:36:31Z

Looks much better with the recent changes 👍

normanmaurer · 2026-02-19T07:09:45Z

@Songdoeon thanks!

netty-project-bot · 2026-02-19T07:14:37Z

Could not create automatic backport PR.
Got conflicts when cherry-picking onto 5.0.

…unknown (netty#16207) Motivation: When the JVM cannot report UNALIGNED support (e.g., on some ARM platforms or in restricted environments), Netty falls back to an architecture-based guess. If that guess yields UNALIGNED = false, multi-byte reads (getLong, getInt, getShort) degrade to byte-by-byte access (8× getByte) even though the JIT could still optimize at runtime. There is currently no distinction between "the platform is known to not support unaligned access" and "we simply don't know." Modification: Introduce UNALIGNED_AVAILABLE flag to distinguish whether UNALIGNED info came from the runtime (true) or from an architecture-based guess (false). When unaligned info is unavailable, use VarHandle instead of 8× getByte so HotSpot can emit the optimal access strategy at runtime. - PlatformDependent0: add UNALIGNED_AVAILABLE flag; add io.netty.unalignedAccess=true|false|unavailable system property. - PlatformDependent: expose isUnalignedAvailable(); allow VarHandle initialization even when Unsafe is available. - UnsafeByteBufUtil: use `VarHandle` for multi-byte `byte[]` operations (`getShort/getInt/getLong`, `setShort/setInt/setLong` and their LE variants) when unaligned info is unavailable. - UnpooledUnsafeDirectByteBuf: use `VarHandle` for multi-byte `ByteBuffer` operations (`_getShort/_getInt/_getLong`, `_setShort/_setInt/_setLong` and their LE variants) when unaligned info is unavailable. - ByteBufUtil: enable SWAR indexOf/lastIndexOf via VarHandle when unaligned info is unavailable. Three-tier access strategy: | Condition | Strategy | Path | |---------|---------|------| | UNALIGNED = true (known supported) | Unsafe direct access | single `Unsafe.getLong()` | | UNALIGNED info unavailable (unknown) | VarHandle | JIT chooses optimal access | | UNALIGNED = false (known unsupported) | 8× getByte | safe byte-wise assembly | --- Result: Fixes netty#15781. Benchmark (JDK 24.0.1, Apple Silicon, JMH 1.36): ### Benchmark Summary | Benchmark | false (8× getByte) | unavailable (VarHandle) | Improvement | |----------|--------------------|--------------------------|-------------| | getLongDirect | 2.144 ns/op | 1.764 ns/op | 17.7% faster | | getLongHeap | 2.152 ns/op | 1.119 ns/op | 48.0% faster | Sorry for the delay in submitting this PR. If anything looks incorrect or needs adjustment, I will update it promptly. Thanks for the review. --------- Co-authored-by: Norman Maurer <norman_maurer@apple.com> Co-authored-by: Chris Vest <christianvest_hansen@apple.com> (cherry picked from commit 0ee9723)

chrisvest · 2026-02-20T23:13:44Z

Forward port PR: #16319

…unknown (#16207) (#16319) Motivation: When the JVM cannot report UNALIGNED support (e.g., on some ARM platforms or in restricted environments), Netty falls back to an architecture-based guess. If that guess yields UNALIGNED = false, multi-byte reads (getLong, getInt, getShort) degrade to byte-by-byte access (8× getByte) even though the JIT could still optimize at runtime. There is currently no distinction between "the platform is known to not support unaligned access" and "we simply don't know." Modification: Introduce UNALIGNED_AVAILABLE flag to distinguish whether UNALIGNED info came from the runtime (true) or from an architecture-based guess (false). When unaligned info is unavailable, use VarHandle instead of 8× getByte so HotSpot can emit the optimal access strategy at runtime. - PlatformDependent0: add UNALIGNED_AVAILABLE flag; add io.netty.unalignedAccess=true|false|unavailable system property. - PlatformDependent: expose isUnalignedAvailable(); allow VarHandle initialization even when Unsafe is available. - UnsafeByteBufUtil: use `VarHandle` for multi-byte `byte[]` operations (`getShort/getInt/getLong`, `setShort/setInt/setLong` and their LE variants) when unaligned info is unavailable. - UnpooledUnsafeDirectByteBuf: use `VarHandle` for multi-byte `ByteBuffer` operations (`_getShort/_getInt/_getLong`, `_setShort/_setInt/_setLong` and their LE variants) when unaligned info is unavailable. - ByteBufUtil: enable SWAR indexOf/lastIndexOf via VarHandle when unaligned info is unavailable. Three-tier access strategy: | Condition | Strategy | Path | |---------|---------|------| | UNALIGNED = true (known supported) | Unsafe direct access | single `Unsafe.getLong()` | | UNALIGNED info unavailable (unknown) | VarHandle | JIT chooses optimal access | | UNALIGNED = false (known unsupported) | 8× getByte | safe byte-wise assembly | --- Result: Fixes #15781. Benchmark (JDK 24.0.1, Apple Silicon, JMH 1.36): ### Benchmark Summary | Benchmark | false (8× getByte) | unavailable (VarHandle) | Improvement | |----------|--------------------|--------------------------|-------------| | getLongDirect | 2.144 ns/op | 1.764 ns/op | 17.7% faster | | getLongHeap | 2.152 ns/op | 1.119 ns/op | 48.0% faster | Sorry for the delay in submitting this PR. If anything looks incorrect or needs adjustment, I will update it promptly. Thanks for the review. --------- Co-authored-by: Norman Maurer <norman_maurer@apple.com> Co-authored-by: Chris Vest <christianvest_hansen@apple.com> (cherry picked from commit 0ee9723) Co-authored-by: DoeonSong <96420547+songdoeon@users.noreply.github.com>

…l [skip ci] Bumps [io.netty:netty-all](https://github.com/netty/netty) from 4.2.10.Final to 4.2.12.Final. Release notes *Sourced from [io.netty:netty-all's releases](https://github.com/netty/netty/releases).* > netty-4.2.12.Final > ------------------ > > What's Changed > -------------- > > * Revert "Eliminate redundant bounds checks in CompositeByteBuf accessors" by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16550](https://redirect.github.com/netty/netty/pull/16550) > > **Full Changelog**: <netty/netty@netty-4.2.11.Final...netty-4.2.12.Final> > > netty-4.2.11.Final > ------------------ > > Security > -------- > > * CVE-2026-33871, [HTTP/2 CONTINUATION Frame Flood Denial of Service](GHSA-w9fj-cfpg-grvv) > * CVE-2026-33870, [HTTP Request Smuggling via Chunked Extension Quoted-String Parsing](GHSA-pwqr-wmgm-9rr8) > > What's Changed > -------------- > > * Update to latest JDK 26 EA release by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16230](https://redirect.github.com/netty/netty/pull/16230) > * HTTP3: Allow to support non-standard HTTP3 settings by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16171](https://redirect.github.com/netty/netty/pull/16171) > * Fix Incorrect nanos-to-millis conversion in epoll\_wait EINTR retry loop by [`@adwsingh`](https://github.com/adwsingh) in [netty/netty#16245](https://redirect.github.com/netty/netty/pull/16245) > * Allocate one large segment and slice for each MsgHdrMemory by [`@dreamlike-ocean`](https://github.com/dreamlike-ocean) in [netty/netty#16234](https://redirect.github.com/netty/netty/pull/16234) > * Make RefCntOpenSslContext.deallocate more robust by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16253](https://redirect.github.com/netty/netty/pull/16253) > * Epoll: Fix excessive CPU usage when Channel is only registered but no… by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16250](https://redirect.github.com/netty/netty/pull/16250) > * Update to gcc for arm 10.3-2021.07 by [`@m1ngyuan`](https://github.com/m1ngyuan) in [netty/netty#16255](https://redirect.github.com/netty/netty/pull/16255) > * Add acmeIdentifier extension support to pkitesting by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16256](https://redirect.github.com/netty/netty/pull/16256) > * Update JDK versions to latest patch releases by [`@m1ngyuan`](https://github.com/m1ngyuan) in [netty/netty#16254](https://redirect.github.com/netty/netty/pull/16254) > * Avoid allocation in HttpObjectEncoder.addEncodedLengthHex method by [`@doom369`](https://github.com/doom369) in [netty/netty#16241](https://redirect.github.com/netty/netty/pull/16241) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16269](https://redirect.github.com/netty/netty/pull/16269) > * Revert "Automatic backporting workflow from 4.1 to 4.2" by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16270](https://redirect.github.com/netty/netty/pull/16270) > * HTTP2: Correctly account for padding when decompress by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16264](https://redirect.github.com/netty/netty/pull/16264) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16271](https://redirect.github.com/netty/netty/pull/16271) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16273](https://redirect.github.com/netty/netty/pull/16273) > * Backport PRs must be created with personal access tokens by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16276](https://redirect.github.com/netty/netty/pull/16276) > * Expose QuicSslContextBuilder::sni by [`@ZeroErrors`](https://github.com/ZeroErrors) in [netty/netty#16178](https://redirect.github.com/netty/netty/pull/16178) > * Add more porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16275](https://redirect.github.com/netty/netty/pull/16275) > * Add more porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16283](https://redirect.github.com/netty/netty/pull/16283) > * Remove the unpooled allocator from test permutations by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16282](https://redirect.github.com/netty/netty/pull/16282) > * Some polishing of the porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16288](https://redirect.github.com/netty/netty/pull/16288) > * Allow to set destination connection id when creating a client side QuicheChannel by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16286](https://redirect.github.com/netty/netty/pull/16286) > * Update to latest JDK26 EA build by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16295](https://redirect.github.com/netty/netty/pull/16295) > * Add javadoc to clarify responsibility of the user when generating the remote connection id by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16293](https://redirect.github.com/netty/netty/pull/16293) > * Make the build run faster by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16290](https://redirect.github.com/netty/netty/pull/16290) > * Fix IDE warnings in SslHandler by [`@doom369`](https://github.com/doom369) in [netty/netty#16237](https://redirect.github.com/netty/netty/pull/16237) > * Decrease Long allocations and map.put calls in ReferenceCountedOpenSllEngine in handshake() method by [`@doom369`](https://github.com/doom369) in [netty/netty#16242](https://redirect.github.com/netty/netty/pull/16242) > * Support boringssl SSLCredential API by [`@jmcrawford45`](https://github.com/jmcrawford45) in [netty/netty#15919](https://redirect.github.com/netty/netty/pull/15919) > * Fix high-order bit aliasing in HttpUtil.validateToken by [`@furkanvarol`](https://github.com/furkanvarol) in [netty/netty#16279](https://redirect.github.com/netty/netty/pull/16279) > * Improve multi-byte access performance when UNALIGNED availability is unknown by [`@Songdoeon`](https://github.com/Songdoeon) in [netty/netty#16207](https://redirect.github.com/netty/netty/pull/16207) > * Avoid unnecessary SSL.getVersion() call and string allocation in ReferenceCountedOpenSslEngine by [`@doom369`](https://github.com/doom369) in [netty/netty#16278](https://redirect.github.com/netty/netty/pull/16278) > * Support more branch freedom for auto-porting by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16300](https://redirect.github.com/netty/netty/pull/16300) > * fix: the precedence of + is higher than >> by [`@cuiweixie`](https://github.com/cuiweixie) in [netty/netty#16312](https://redirect.github.com/netty/netty/pull/16312) > * AdaptiveByteBufAllocator: make sure byteBuf.capacity() not greater than byteBuf.maxCapacity() by [`@laosijikaichele`](https://github.com/laosijikaichele) in [netty/netty#16309](https://redirect.github.com/netty/netty/pull/16309) > * Fix flaky PooledByteBufAllocatorTest by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16313](https://redirect.github.com/netty/netty/pull/16313) > * Fix pooled arena accounting tests by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16321](https://redirect.github.com/netty/netty/pull/16321) ... (truncated) Commits * [`67ce541`](netty/netty@67ce541) [maven-release-plugin] prepare release netty-4.2.12.Final * [`7074624`](netty/netty@7074624) Revert "Eliminate redundant bounds checks in CompositeByteBuf accessors" ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`c3b0a43`](netty/netty@c3b0a43) [maven-release-plugin] prepare for next development iteration * [`c94a818`](netty/netty@c94a818) [maven-release-plugin] prepare release netty-4.2.11.Final * [`3b76df1`](netty/netty@3b76df1) Merge commit from fork * [`aae944a`](netty/netty@aae944a) Auto-port 4.2: Limit the number of Continuation frames per HTTP2 Headers ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`6001499`](netty/netty@6001499) Eliminate redundant bounds checks in CompositeByteBuf accessors ([#16525](https://redirect.github.com/netty/netty/issues/16525)) * [`a7fbb6f`](netty/netty@a7fbb6f) JdkZlibDecoder: accumulate decompressed output before firing channelRead ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`7937553`](netty/netty@7937553) Enforce io.netty.maxDirectMemory accounting on all Java versions ([#16489](https://redirect.github.com/netty/netty/issues/16489)) * [`893ea2e`](netty/netty@893ea2e) Allocate less in QueryStringDecoder.addParam for typical use case ([#16527](https://redirect.github.com/netty/netty/issues/16527)) * Additional commits viewable in [compare view](netty/netty@netty-4.2.10.Final...netty-4.2.12.Final) [![Dependabot compatibility score](https://dependabot-badges.githubapp.com/badges/compatibility\_score?dependency-name=io.netty:netty-all&package-manager=maven&previous-version=4.2.10.Final&new-version=4.2.12.Final)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores) Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`. [//]: # (dependabot-automerge-start) [//]: # (dependabot-automerge-end) --- Dependabot commands and options You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot show ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

…ip ci] Bumps `netty.version` from 4.2.10.Final to 4.2.12.Final. Updates `io.netty:netty-transport` from 4.2.10.Final to 4.2.12.Final Release notes *Sourced from [io.netty:netty-transport's releases](https://github.com/netty/netty/releases).* > netty-4.2.12.Final > ------------------ > > What's Changed > -------------- > > * Revert "Eliminate redundant bounds checks in CompositeByteBuf accessors" by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16550](https://redirect.github.com/netty/netty/pull/16550) > > **Full Changelog**: <netty/netty@netty-4.2.11.Final...netty-4.2.12.Final> > > netty-4.2.11.Final > ------------------ > > Security > -------- > > * CVE-2026-33871, [HTTP/2 CONTINUATION Frame Flood Denial of Service](GHSA-w9fj-cfpg-grvv) > * CVE-2026-33870, [HTTP Request Smuggling via Chunked Extension Quoted-String Parsing](GHSA-pwqr-wmgm-9rr8) > > What's Changed > -------------- > > * Update to latest JDK 26 EA release by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16230](https://redirect.github.com/netty/netty/pull/16230) > * HTTP3: Allow to support non-standard HTTP3 settings by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16171](https://redirect.github.com/netty/netty/pull/16171) > * Fix Incorrect nanos-to-millis conversion in epoll\_wait EINTR retry loop by [`@adwsingh`](https://github.com/adwsingh) in [netty/netty#16245](https://redirect.github.com/netty/netty/pull/16245) > * Allocate one large segment and slice for each MsgHdrMemory by [`@dreamlike-ocean`](https://github.com/dreamlike-ocean) in [netty/netty#16234](https://redirect.github.com/netty/netty/pull/16234) > * Make RefCntOpenSslContext.deallocate more robust by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16253](https://redirect.github.com/netty/netty/pull/16253) > * Epoll: Fix excessive CPU usage when Channel is only registered but no… by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16250](https://redirect.github.com/netty/netty/pull/16250) > * Update to gcc for arm 10.3-2021.07 by [`@m1ngyuan`](https://github.com/m1ngyuan) in [netty/netty#16255](https://redirect.github.com/netty/netty/pull/16255) > * Add acmeIdentifier extension support to pkitesting by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16256](https://redirect.github.com/netty/netty/pull/16256) > * Update JDK versions to latest patch releases by [`@m1ngyuan`](https://github.com/m1ngyuan) in [netty/netty#16254](https://redirect.github.com/netty/netty/pull/16254) > * Avoid allocation in HttpObjectEncoder.addEncodedLengthHex method by [`@doom369`](https://github.com/doom369) in [netty/netty#16241](https://redirect.github.com/netty/netty/pull/16241) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16269](https://redirect.github.com/netty/netty/pull/16269) > * Revert "Automatic backporting workflow from 4.1 to 4.2" by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16270](https://redirect.github.com/netty/netty/pull/16270) > * HTTP2: Correctly account for padding when decompress by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16264](https://redirect.github.com/netty/netty/pull/16264) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16271](https://redirect.github.com/netty/netty/pull/16271) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16273](https://redirect.github.com/netty/netty/pull/16273) > * Backport PRs must be created with personal access tokens by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16276](https://redirect.github.com/netty/netty/pull/16276) > * Expose QuicSslContextBuilder::sni by [`@ZeroErrors`](https://github.com/ZeroErrors) in [netty/netty#16178](https://redirect.github.com/netty/netty/pull/16178) > * Add more porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16275](https://redirect.github.com/netty/netty/pull/16275) > * Add more porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16283](https://redirect.github.com/netty/netty/pull/16283) > * Remove the unpooled allocator from test permutations by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16282](https://redirect.github.com/netty/netty/pull/16282) > * Some polishing of the porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16288](https://redirect.github.com/netty/netty/pull/16288) > * Allow to set destination connection id when creating a client side QuicheChannel by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16286](https://redirect.github.com/netty/netty/pull/16286) > * Update to latest JDK26 EA build by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16295](https://redirect.github.com/netty/netty/pull/16295) > * Add javadoc to clarify responsibility of the user when generating the remote connection id by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16293](https://redirect.github.com/netty/netty/pull/16293) > * Make the build run faster by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16290](https://redirect.github.com/netty/netty/pull/16290) > * Fix IDE warnings in SslHandler by [`@doom369`](https://github.com/doom369) in [netty/netty#16237](https://redirect.github.com/netty/netty/pull/16237) > * Decrease Long allocations and map.put calls in ReferenceCountedOpenSllEngine in handshake() method by [`@doom369`](https://github.com/doom369) in [netty/netty#16242](https://redirect.github.com/netty/netty/pull/16242) > * Support boringssl SSLCredential API by [`@jmcrawford45`](https://github.com/jmcrawford45) in [netty/netty#15919](https://redirect.github.com/netty/netty/pull/15919) > * Fix high-order bit aliasing in HttpUtil.validateToken by [`@furkanvarol`](https://github.com/furkanvarol) in [netty/netty#16279](https://redirect.github.com/netty/netty/pull/16279) > * Improve multi-byte access performance when UNALIGNED availability is unknown by [`@Songdoeon`](https://github.com/Songdoeon) in [netty/netty#16207](https://redirect.github.com/netty/netty/pull/16207) > * Avoid unnecessary SSL.getVersion() call and string allocation in ReferenceCountedOpenSslEngine by [`@doom369`](https://github.com/doom369) in [netty/netty#16278](https://redirect.github.com/netty/netty/pull/16278) > * Support more branch freedom for auto-porting by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16300](https://redirect.github.com/netty/netty/pull/16300) > * fix: the precedence of + is higher than >> by [`@cuiweixie`](https://github.com/cuiweixie) in [netty/netty#16312](https://redirect.github.com/netty/netty/pull/16312) > * AdaptiveByteBufAllocator: make sure byteBuf.capacity() not greater than byteBuf.maxCapacity() by [`@laosijikaichele`](https://github.com/laosijikaichele) in [netty/netty#16309](https://redirect.github.com/netty/netty/pull/16309) > * Fix flaky PooledByteBufAllocatorTest by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16313](https://redirect.github.com/netty/netty/pull/16313) > * Fix pooled arena accounting tests by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16321](https://redirect.github.com/netty/netty/pull/16321) ... (truncated) Commits * [`67ce541`](netty/netty@67ce541) [maven-release-plugin] prepare release netty-4.2.12.Final * [`7074624`](netty/netty@7074624) Revert "Eliminate redundant bounds checks in CompositeByteBuf accessors" ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`c3b0a43`](netty/netty@c3b0a43) [maven-release-plugin] prepare for next development iteration * [`c94a818`](netty/netty@c94a818) [maven-release-plugin] prepare release netty-4.2.11.Final * [`3b76df1`](netty/netty@3b76df1) Merge commit from fork * [`aae944a`](netty/netty@aae944a) Auto-port 4.2: Limit the number of Continuation frames per HTTP2 Headers ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`6001499`](netty/netty@6001499) Eliminate redundant bounds checks in CompositeByteBuf accessors ([#16525](https://redirect.github.com/netty/netty/issues/16525)) * [`a7fbb6f`](netty/netty@a7fbb6f) JdkZlibDecoder: accumulate decompressed output before firing channelRead ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`7937553`](netty/netty@7937553) Enforce io.netty.maxDirectMemory accounting on all Java versions ([#16489](https://redirect.github.com/netty/netty/issues/16489)) * [`893ea2e`](netty/netty@893ea2e) Allocate less in QueryStringDecoder.addParam for typical use case ([#16527](https://redirect.github.com/netty/netty/issues/16527)) * Additional commits viewable in [compare view](netty/netty@netty-4.2.10.Final...netty-4.2.12.Final) Updates `io.netty:netty-codec` from 4.2.10.Final to 4.2.12.Final Release notes *Sourced from [io.netty:netty-codec's releases](https://github.com/netty/netty/releases).* > netty-4.2.12.Final > ------------------ > > What's Changed > -------------- > > * Revert "Eliminate redundant bounds checks in CompositeByteBuf accessors" by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16550](https://redirect.github.com/netty/netty/pull/16550) > > **Full Changelog**: <netty/netty@netty-4.2.11.Final...netty-4.2.12.Final> > > netty-4.2.11.Final > ------------------ > > Security > -------- > > * CVE-2026-33871, [HTTP/2 CONTINUATION Frame Flood Denial of Service](GHSA-w9fj-cfpg-grvv) > * CVE-2026-33870, [HTTP Request Smuggling via Chunked Extension Quoted-String Parsing](GHSA-pwqr-wmgm-9rr8) > > What's Changed > -------------- > > * Update to latest JDK 26 EA release by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16230](https://redirect.github.com/netty/netty/pull/16230) > * HTTP3: Allow to support non-standard HTTP3 settings by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16171](https://redirect.github.com/netty/netty/pull/16171) > * Fix Incorrect nanos-to-millis conversion in epoll\_wait EINTR retry loop by [`@adwsingh`](https://github.com/adwsingh) in [netty/netty#16245](https://redirect.github.com/netty/netty/pull/16245) > * Allocate one large segment and slice for each MsgHdrMemory by [`@dreamlike-ocean`](https://github.com/dreamlike-ocean) in [netty/netty#16234](https://redirect.github.com/netty/netty/pull/16234) > * Make RefCntOpenSslContext.deallocate more robust by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16253](https://redirect.github.com/netty/netty/pull/16253) > * Epoll: Fix excessive CPU usage when Channel is only registered but no… by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16250](https://redirect.github.com/netty/netty/pull/16250) > * Update to gcc for arm 10.3-2021.07 by [`@m1ngyuan`](https://github.com/m1ngyuan) in [netty/netty#16255](https://redirect.github.com/netty/netty/pull/16255) > * Add acmeIdentifier extension support to pkitesting by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16256](https://redirect.github.com/netty/netty/pull/16256) > * Update JDK versions to latest patch releases by [`@m1ngyuan`](https://github.com/m1ngyuan) in [netty/netty#16254](https://redirect.github.com/netty/netty/pull/16254) > * Avoid allocation in HttpObjectEncoder.addEncodedLengthHex method by [`@doom369`](https://github.com/doom369) in [netty/netty#16241](https://redirect.github.com/netty/netty/pull/16241) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16269](https://redirect.github.com/netty/netty/pull/16269) > * Revert "Automatic backporting workflow from 4.1 to 4.2" by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16270](https://redirect.github.com/netty/netty/pull/16270) > * HTTP2: Correctly account for padding when decompress by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16264](https://redirect.github.com/netty/netty/pull/16264) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16271](https://redirect.github.com/netty/netty/pull/16271) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16273](https://redirect.github.com/netty/netty/pull/16273) > * Backport PRs must be created with personal access tokens by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16276](https://redirect.github.com/netty/netty/pull/16276) > * Expose QuicSslContextBuilder::sni by [`@ZeroErrors`](https://github.com/ZeroErrors) in [netty/netty#16178](https://redirect.github.com/netty/netty/pull/16178) > * Add more porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16275](https://redirect.github.com/netty/netty/pull/16275) > * Add more porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16283](https://redirect.github.com/netty/netty/pull/16283) > * Remove the unpooled allocator from test permutations by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16282](https://redirect.github.com/netty/netty/pull/16282) > * Some polishing of the porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16288](https://redirect.github.com/netty/netty/pull/16288) > * Allow to set destination connection id when creating a client side QuicheChannel by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16286](https://redirect.github.com/netty/netty/pull/16286) > * Update to latest JDK26 EA build by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16295](https://redirect.github.com/netty/netty/pull/16295) > * Add javadoc to clarify responsibility of the user when generating the remote connection id by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16293](https://redirect.github.com/netty/netty/pull/16293) > * Make the build run faster by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16290](https://redirect.github.com/netty/netty/pull/16290) > * Fix IDE warnings in SslHandler by [`@doom369`](https://github.com/doom369) in [netty/netty#16237](https://redirect.github.com/netty/netty/pull/16237) > * Decrease Long allocations and map.put calls in ReferenceCountedOpenSllEngine in handshake() method by [`@doom369`](https://github.com/doom369) in [netty/netty#16242](https://redirect.github.com/netty/netty/pull/16242) > * Support boringssl SSLCredential API by [`@jmcrawford45`](https://github.com/jmcrawford45) in [netty/netty#15919](https://redirect.github.com/netty/netty/pull/15919) > * Fix high-order bit aliasing in HttpUtil.validateToken by [`@furkanvarol`](https://github.com/furkanvarol) in [netty/netty#16279](https://redirect.github.com/netty/netty/pull/16279) > * Improve multi-byte access performance when UNALIGNED availability is unknown by [`@Songdoeon`](https://github.com/Songdoeon) in [netty/netty#16207](https://redirect.github.com/netty/netty/pull/16207) > * Avoid unnecessary SSL.getVersion() call and string allocation in ReferenceCountedOpenSslEngine by [`@doom369`](https://github.com/doom369) in [netty/netty#16278](https://redirect.github.com/netty/netty/pull/16278) > * Support more branch freedom for auto-porting by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16300](https://redirect.github.com/netty/netty/pull/16300) > * fix: the precedence of + is higher than >> by [`@cuiweixie`](https://github.com/cuiweixie) in [netty/netty#16312](https://redirect.github.com/netty/netty/pull/16312) > * AdaptiveByteBufAllocator: make sure byteBuf.capacity() not greater than byteBuf.maxCapacity() by [`@laosijikaichele`](https://github.com/laosijikaichele) in [netty/netty#16309](https://redirect.github.com/netty/netty/pull/16309) > * Fix flaky PooledByteBufAllocatorTest by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16313](https://redirect.github.com/netty/netty/pull/16313) > * Fix pooled arena accounting tests by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16321](https://redirect.github.com/netty/netty/pull/16321) ... (truncated) Commits * [`67ce541`](netty/netty@67ce541) [maven-release-plugin] prepare release netty-4.2.12.Final * [`7074624`](netty/netty@7074624) Revert "Eliminate redundant bounds checks in CompositeByteBuf accessors" ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`c3b0a43`](netty/netty@c3b0a43) [maven-release-plugin] prepare for next development iteration * [`c94a818`](netty/netty@c94a818) [maven-release-plugin] prepare release netty-4.2.11.Final * [`3b76df1`](netty/netty@3b76df1) Merge commit from fork * [`aae944a`](netty/netty@aae944a) Auto-port 4.2: Limit the number of Continuation frames per HTTP2 Headers ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`6001499`](netty/netty@6001499) Eliminate redundant bounds checks in CompositeByteBuf accessors ([#16525](https://redirect.github.com/netty/netty/issues/16525)) * [`a7fbb6f`](netty/netty@a7fbb6f) JdkZlibDecoder: accumulate decompressed output before firing channelRead ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`7937553`](netty/netty@7937553) Enforce io.netty.maxDirectMemory accounting on all Java versions ([#16489](https://redirect.github.com/netty/netty/issues/16489)) * [`893ea2e`](netty/netty@893ea2e) Allocate less in QueryStringDecoder.addParam for typical use case ([#16527](https://redirect.github.com/netty/netty/issues/16527)) * Additional commits viewable in [compare view](netty/netty@netty-4.2.10.Final...netty-4.2.12.Final) Updates `io.netty:netty-handler` from 4.2.10.Final to 4.2.12.Final Release notes *Sourced from [io.netty:netty-handler's releases](https://github.com/netty/netty/releases).* > netty-4.2.12.Final > ------------------ > > What's Changed > -------------- > > * Revert "Eliminate redundant bounds checks in CompositeByteBuf accessors" by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16550](https://redirect.github.com/netty/netty/pull/16550) > > **Full Changelog**: <netty/netty@netty-4.2.11.Final...netty-4.2.12.Final> > > netty-4.2.11.Final > ------------------ > > Security > -------- > > * CVE-2026-33871, [HTTP/2 CONTINUATION Frame Flood Denial of Service](GHSA-w9fj-cfpg-grvv) > * CVE-2026-33870, [HTTP Request Smuggling via Chunked Extension Quoted-String Parsing](GHSA-pwqr-wmgm-9rr8) > > What's Changed > -------------- > > * Update to latest JDK 26 EA release by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16230](https://redirect.github.com/netty/netty/pull/16230) > * HTTP3: Allow to support non-standard HTTP3 settings by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16171](https://redirect.github.com/netty/netty/pull/16171) > * Fix Incorrect nanos-to-millis conversion in epoll\_wait EINTR retry loop by [`@adwsingh`](https://github.com/adwsingh) in [netty/netty#16245](https://redirect.github.com/netty/netty/pull/16245) > * Allocate one large segment and slice for each MsgHdrMemory by [`@dreamlike-ocean`](https://github.com/dreamlike-ocean) in [netty/netty#16234](https://redirect.github.com/netty/netty/pull/16234) > * Make RefCntOpenSslContext.deallocate more robust by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16253](https://redirect.github.com/netty/netty/pull/16253) > * Epoll: Fix excessive CPU usage when Channel is only registered but no… by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16250](https://redirect.github.com/netty/netty/pull/16250) > * Update to gcc for arm 10.3-2021.07 by [`@m1ngyuan`](https://github.com/m1ngyuan) in [netty/netty#16255](https://redirect.github.com/netty/netty/pull/16255) > * Add acmeIdentifier extension support to pkitesting by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16256](https://redirect.github.com/netty/netty/pull/16256) > * Update JDK versions to latest patch releases by [`@m1ngyuan`](https://github.com/m1ngyuan) in [netty/netty#16254](https://redirect.github.com/netty/netty/pull/16254) > * Avoid allocation in HttpObjectEncoder.addEncodedLengthHex method by [`@doom369`](https://github.com/doom369) in [netty/netty#16241](https://redirect.github.com/netty/netty/pull/16241) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16269](https://redirect.github.com/netty/netty/pull/16269) > * Revert "Automatic backporting workflow from 4.1 to 4.2" by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16270](https://redirect.github.com/netty/netty/pull/16270) > * HTTP2: Correctly account for padding when decompress by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16264](https://redirect.github.com/netty/netty/pull/16264) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16271](https://redirect.github.com/netty/netty/pull/16271) > * Automatic backporting workflow from 4.1 to 4.2 by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16273](https://redirect.github.com/netty/netty/pull/16273) > * Backport PRs must be created with personal access tokens by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16276](https://redirect.github.com/netty/netty/pull/16276) > * Expose QuicSslContextBuilder::sni by [`@ZeroErrors`](https://github.com/ZeroErrors) in [netty/netty#16178](https://redirect.github.com/netty/netty/pull/16178) > * Add more porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16275](https://redirect.github.com/netty/netty/pull/16275) > * Add more porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16283](https://redirect.github.com/netty/netty/pull/16283) > * Remove the unpooled allocator from test permutations by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16282](https://redirect.github.com/netty/netty/pull/16282) > * Some polishing of the porting workflows by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16288](https://redirect.github.com/netty/netty/pull/16288) > * Allow to set destination connection id when creating a client side QuicheChannel by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16286](https://redirect.github.com/netty/netty/pull/16286) > * Update to latest JDK26 EA build by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16295](https://redirect.github.com/netty/netty/pull/16295) > * Add javadoc to clarify responsibility of the user when generating the remote connection id by [`@normanmaurer`](https://github.com/normanmaurer) in [netty/netty#16293](https://redirect.github.com/netty/netty/pull/16293) > * Make the build run faster by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16290](https://redirect.github.com/netty/netty/pull/16290) > * Fix IDE warnings in SslHandler by [`@doom369`](https://github.com/doom369) in [netty/netty#16237](https://redirect.github.com/netty/netty/pull/16237) > * Decrease Long allocations and map.put calls in ReferenceCountedOpenSllEngine in handshake() method by [`@doom369`](https://github.com/doom369) in [netty/netty#16242](https://redirect.github.com/netty/netty/pull/16242) > * Support boringssl SSLCredential API by [`@jmcrawford45`](https://github.com/jmcrawford45) in [netty/netty#15919](https://redirect.github.com/netty/netty/pull/15919) > * Fix high-order bit aliasing in HttpUtil.validateToken by [`@furkanvarol`](https://github.com/furkanvarol) in [netty/netty#16279](https://redirect.github.com/netty/netty/pull/16279) > * Improve multi-byte access performance when UNALIGNED availability is unknown by [`@Songdoeon`](https://github.com/Songdoeon) in [netty/netty#16207](https://redirect.github.com/netty/netty/pull/16207) > * Avoid unnecessary SSL.getVersion() call and string allocation in ReferenceCountedOpenSslEngine by [`@doom369`](https://github.com/doom369) in [netty/netty#16278](https://redirect.github.com/netty/netty/pull/16278) > * Support more branch freedom for auto-porting by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16300](https://redirect.github.com/netty/netty/pull/16300) > * fix: the precedence of + is higher than >> by [`@cuiweixie`](https://github.com/cuiweixie) in [netty/netty#16312](https://redirect.github.com/netty/netty/pull/16312) > * AdaptiveByteBufAllocator: make sure byteBuf.capacity() not greater than byteBuf.maxCapacity() by [`@laosijikaichele`](https://github.com/laosijikaichele) in [netty/netty#16309](https://redirect.github.com/netty/netty/pull/16309) > * Fix flaky PooledByteBufAllocatorTest by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16313](https://redirect.github.com/netty/netty/pull/16313) > * Fix pooled arena accounting tests by [`@chrisvest`](https://github.com/chrisvest) in [netty/netty#16321](https://redirect.github.com/netty/netty/pull/16321) ... (truncated) Commits * [`67ce541`](netty/netty@67ce541) [maven-release-plugin] prepare release netty-4.2.12.Final * [`7074624`](netty/netty@7074624) Revert "Eliminate redundant bounds checks in CompositeByteBuf accessors" ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`c3b0a43`](netty/netty@c3b0a43) [maven-release-plugin] prepare for next development iteration * [`c94a818`](netty/netty@c94a818) [maven-release-plugin] prepare release netty-4.2.11.Final * [`3b76df1`](netty/netty@3b76df1) Merge commit from fork * [`aae944a`](netty/netty@aae944a) Auto-port 4.2: Limit the number of Continuation frames per HTTP2 Headers ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`6001499`](netty/netty@6001499) Eliminate redundant bounds checks in CompositeByteBuf accessors ([#16525](https://redirect.github.com/netty/netty/issues/16525)) * [`a7fbb6f`](netty/netty@a7fbb6f) JdkZlibDecoder: accumulate decompressed output before firing channelRead ([#16](https://redirect.github.com/netty/netty/issues/16)... * [`7937553`](netty/netty@7937553) Enforce io.netty.maxDirectMemory accounting on all Java versions ([#16489](https://redirect.github.com/netty/netty/issues/16489)) * [`893ea2e`](netty/netty@893ea2e) Allocate less in QueryStringDecoder.addParam for typical use case ([#16527](https://redirect.github.com/netty/netty/issues/16527)) * Additional commits viewable in [compare view](netty/netty@netty-4.2.10.Final...netty-4.2.12.Final) Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`. [//]: # (dependabot-automerge-start) [//]: # (dependabot-automerge-end) --- Dependabot commands and options You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot show ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

Use VarHandle for multi-byte access when UNALIGNED availability is un…

b78209c

…known

Songdoeon changed the title ~~Use VarHandle for multi-byte access when UNALIGNED availability is un…~~ Improve multi-byte access performance when UNALIGNED availability is unknown Jan 31, 2026

Songdoeon added 3 commits February 1, 2026 15:17

Use VarHandle for multi-byte access when UNALIGNED availability is un…

237d06a

…known

Merge branch 'Poor_access_performance_UNALIGNED' of https://github.co…

c7ac92a

…m/Songdoeon/netty into Poor_access_performance_UNALIGNED

Add VarHandle paths

5f1833b

normanmaurer reviewed Feb 1, 2026

View reviewed changes

normanmaurer requested changes Feb 1, 2026

View reviewed changes

microbench/src/main/java/io/netty/buffer/UnalignedAccessUnavailableBenchmark.java Outdated Show resolved Hide resolved

microbench/src/main/java/io/netty/microbench/search/UnalignedIndexOfBenchmark.java Outdated Show resolved Hide resolved

Extract USE_VAR_HANDLE boolean for VarHandle access path

430fb12

franz1981 requested changes Feb 2, 2026

View reviewed changes

Songdoeon and others added 4 commits February 3, 2026 09:38

Merge branch '4.2' into Poor_access_performance_UNALIGNED

0eaaba6

Update microbench/src/main/java/io/netty/microbench/search/UnalignedI…

aba8750

…ndexOfBenchmark.java

Update microbench/src/main/java/io/netty/buffer/UnalignedAccessUnavai…

1b025a5

…lableBenchmark.java

Merge branch '4.2' into Poor_access_performance_UNALIGNED

efccf87

franz1981 approved these changes Feb 4, 2026

View reviewed changes

Songdoeon requested a review from normanmaurer February 5, 2026 00:03

normanmaurer requested changes Feb 5, 2026

View reviewed changes

buffer/src/main/java/io/netty/buffer/ByteBufUtil.java Outdated Show resolved Hide resolved

buffer/src/main/java/io/netty/buffer/UnsafeByteBufUtil.java Outdated Show resolved Hide resolved

franz1981 mentioned this pull request Feb 5, 2026

Native: Enable Netty JCTools' MpscUnboundedArrayQueue quarkusio/quarkus#52418

Merged

Songdoeon and others added 3 commits February 6, 2026 09:27

simplify unaligned checks

1cd2b93

Merge branch '4.2' into Poor_access_performance_UNALIGNED

2c4fc37

Merge branch '4.2' into Poor_access_performance_UNALIGNED

b8114e5

normanmaurer added this to the 4.2.11.Final milestone Feb 9, 2026

chrisvest reviewed Feb 12, 2026

View reviewed changes

buffer/src/main/java/io/netty/buffer/UnsafeByteBufUtil.java Outdated Show resolved Hide resolved

common/src/main/java/io/netty/util/internal/PlatformDependent.java Outdated Show resolved Hide resolved

Songdoeon added 2 commits February 14, 2026 17:02

Simplify and centralize multi-byte access decision logic

a9bbb67

fix javadoc

8ef7c12

chrisvest approved these changes Feb 16, 2026

View reviewed changes

chrisvest added the needs-cherry-pick-5.0 This PR should be cherry-picked to 5.0 once merged. label Feb 16, 2026

Merge branch '4.2' into Poor_access_performance_UNALIGNED

b2c7ea9

normanmaurer merged commit 0ee9723 into netty:4.2 Feb 19, 2026
19 checks passed

Songdoeon deleted the Poor_access_performance_UNALIGNED branch February 24, 2026 04:36

Uh oh!

Conversation

Songdoeon commented Jan 31, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Summary

Uh oh!

franz1981 commented Jan 31, 2026

Uh oh!

franz1981 commented Jan 31, 2026

Uh oh!

Songdoeon commented Feb 1, 2026

Uh oh!

normanmaurer Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

Songdoeon Feb 2, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

franz1981 Feb 2, 2026

Choose a reason for hiding this comment

Uh oh!

Songdoeon Feb 3, 2026

Choose a reason for hiding this comment

Uh oh!

franz1981 left a comment

Choose a reason for hiding this comment

Uh oh!

normanmaurer left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

normanmaurer commented Feb 10, 2026

Uh oh!

chrisvest left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Songdoeon commented Feb 14, 2026

Uh oh!

chrisvest commented Feb 16, 2026

Uh oh!

Uh oh!

normanmaurer commented Feb 19, 2026

Uh oh!

netty-project-bot commented Feb 19, 2026

Uh oh!

chrisvest commented Feb 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Songdoeon commented Jan 31, 2026 •

edited

Loading