Datasets:

feedback-to-code

/

mozzarella

like 1

Follow

Feedback-2-Code 7

[ "keyword_pr_to_issue" ]

[ "Do you want to make a PR?" ]

[]

"2022-06-14T12:39:00Z"

[]

[ "handler/src/main/java/io/netty/handler/flow/FlowControlHandler.java" ]

[ "handler/src/main/java/io/netty/handler/flow/FlowControlHandler.java" ]

[]

"2022-06-06T17:15:26Z"

[ "keyword_pr_to_issue" ]

[ "These are used by the native code loading. I think it'd be better to have a system property where the classifier set can be directly specified, and then avoid reading the files if that property is present. Then it could also serve a dual purpose of overriding the files in case they produce a wrong classifier set for some reason." ]

[]

"2022-06-17T15:31:45Z"

[]

[ "common/src/main/java/io/netty/util/internal/PlatformDependent.java" ]

[ "common/src/main/java/io/netty/util/internal/PlatformDependent.java" ]

[ "common/src/test/java/io/netty/util/internal/OsClassifiersTest.java" ]

"2022-06-16T21:52:24Z"

[ "connected" ]

[ "@gspadotto Are you able to make a build with this and give it a try on your system? https://github.com/netty/netty/pull/12498", "> @gspadotto Are you able to make a build with this and give it a try on your system? #12498\r\n\r\nHi @chrisvest \r\nI'll give it a try even though it's a bit complex for me. \r\nI'll explain what the set-up is to reproduce the issue and then I'll ask you some guidance on the \"correct\" way for me to test your patch. \r\n\r\nI have an instance of [this](https://github.com/wso2/micro-integrator/releases/tag/v4.1.0), running inside a [Docker container](https://github.com/wso2/docker-ei/blob/4.1.x/dockerfiles/alpine/micro-integrator/Dockerfile), where [this row](https://github.com/wso2/docker-ei/blob/4.1.x/dockerfiles/alpine/micro-integrator/Dockerfile#L20) has been changed to \"FROM arm64v8/openjdk:11-jdk-slim\".\r\n\r\nThese are the runtime dependencies that refer to Netty: \r\n\r\n```\r\nwso2/components/plugins/io.netty.buffer_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.codec-http2_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.codec-http_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.codec-socks_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.codec_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.common_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.handler-proxy_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.handler_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.resolver_4.1.73.Final.jar\r\nwso2/components/plugins/io.netty.tcnative-boringssl-static_2.0.47.Final.jar\r\nwso2/components/plugins/io.netty.tcnative-classes_2.0.47.Final.jar\r\nwso2/components/plugins/io.netty.tcnative_2.0.47.Final.jar\r\nwso2/components/plugins/io.netty.transport_4.1.73.Final.jar\r\n```\r\n\r\nMy plan for testing is to overwrite only the io.netty.tcnative* jars with the patched ones and see if the OSGi bundles are activated without issues. \r\n\r\nMy questions for you are: \r\n1) Do you think it's ok to \"mix & match\" different versions of netty dependencies?\r\n2) Which components am I supposed to build and embed/overwrite?\r\n3) Can I build netty components for an architecture (arm64) that is different than the one the host that builds the code is running (i.e. \"cross-compiling\")?\r\n\r\nAs a side note, simply editing the manifest file inside the io.netty.tcnative-boringssl-static_2.0.47.Final.jar to read **processor=aarch64** rather than **processor=aarch_64** already fixes the issue: I don't know if this can be considered as a \"successful test\" or not. \r\n\r\nPlease let me know.\r\n" ]

[]

"2022-06-21T23:31:52Z"

[]

[ "resolver-dns-native-macos/pom.xml", "transport-native-epoll/pom.xml", "transport-native-kqueue/pom.xml" ]

[ "resolver-dns-native-macos/pom.xml", "transport-native-epoll/pom.xml", "transport-native-kqueue/pom.xml" ]

[]

"2022-06-21T14:45:42Z"

[ "keyword_pr_to_issue" ]

[]

[ "actually I think this will not work in netty 4.1 due our java version requirement and so you need to handle this kind of stuff differently.", "The rethrowing Throwable? Then we can use `PlatformDependent.throwException` instead.", "Yep... Only works on java8+" ]

"2022-06-22T21:07:51Z"

[]

[ "codec/src/main/java/io/netty/handler/codec/MessageToMessageEncoder.java" ]

[ "codec/src/main/java/io/netty/handler/codec/MessageToMessageEncoder.java" ]

[]

"2022-06-22T03:47:28Z"

[ "keyword_pr_to_issue" ]

[ "A `ByteBuf` was unexpectedly released while being worked on by a `ByteToMessageDecoder`.\r\nWhatever class extends the `ByteToMessageDecoder` probably called `release()` on the `in` buffer argument, which it shouldn't do.\r\nThe input buffer is managed by the `ByteToMessageDecoder`, so sub-classes should just advance the `readerIndex`.", "Hi @chrisvest ,\r\nWe have mix of GET & POST in the traffic. We are performing load testing on our application(1500 TPS. for 42 hours)\r\nThings work fine for 42 hours(All requests are processed). Suddenly the application starts dropping the request with following exception\r\n**[8c5489df, L:/10.233.114.202:8443 - R:/10.233.90.102:56316] EXCEPTION: io.netty.util.IllegalReferenceCountException: refCnt: 0, decrement: 1**\r\nAttaching the stacktrace for reference.\r\nNeed your assistance here to further understand this [behavior]\r\n[StackTrace.txt](https://github.com/netty/netty/files/9242763/StackTrace.txt)\r\n\r\n", "@joyfulnikhil Looks like the same thing. The bug is likely in the `ByteToMessageDecoder` subclass. I've put up a PR to make the error message more helpful." ]

[ "Can a check for **cumulation.refCount() >0** be placed here? This will ensure that only those bytebuf are released whose refCount>0 and if bytebuf already has a refcount of 0, it is already deallocated or can be returned to the object pool. This will prevent IllegalReferenceCount from occurring. \r\nNot sure about the implication on memory (Whether this check will cause a memory leak ?)" ]

"2022-08-02T22:30:49Z"

[]

[ "codec/src/main/java/io/netty/handler/codec/ByteToMessageDecoder.java" ]

[ "codec/src/main/java/io/netty/handler/codec/ByteToMessageDecoder.java" ]

[]

"2022-05-05T05:29:24Z"

[ "keyword_pr_to_issue", "connected" ]

[ "@chrisvest PTAL", "> Caused by: java.lang.IllegalStateException: This allocator has been **closed**.\r\n> \tat io.netty5.buffer.api.internal.Statics.allocatorClosedException(Statics.java:268)\r\n> \tat io.netty5.buffer.api.pool.PooledBufferAllocator.allocate(PooledBufferAllocator.java:316)\r\n> \tat io.netty5.buffer.api.DefaultBufferAllocators$**UncloseableBufferAllocator**.allocate(DefaultBufferAllocators.java:129)\r\n\r\nQuite the mystery, here… 🤔 \r\n\r\nI'll take a look." ]

[]

"2022-08-08T21:55:31Z"

[]

[ "buffer/src/main/java/io/netty5/buffer/api/DefaultBufferAllocators.java" ]

[ "buffer/src/main/java/io/netty5/buffer/api/DefaultBufferAllocators.java" ]

[]

"2022-08-06T17:58:23Z"

[ "keyword_pr_to_issue" ]

[ "Never saw this big stack trace. ;/", "make sure you have:\r\n\r\n```\r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <version>XXX</version>\r\n <classifier>osx-x86_64</classifier>\r\n </dependency>\r\n```\r\n\r\nin your class path on OS X", "faced with the same problem in Redisson yesterday,\r\nthanx @vietj, your answer was so in time", "> make sure you have:\r\n> \r\n> ```\r\n> <dependency>\r\n> <groupId>io.netty</groupId>\r\n> <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n> <version>XXX</version>\r\n> <classifier>osx-x86_64</classifier>\r\n> </dependency>\r\n> ```\r\n> \r\n> in your class path on OS X\r\n\r\nDoesn't fix the issue anymore.", "For Gradle (with Kotlin syntax) it is:\r\n```\r\nruntimeOnly(\"io.netty:netty-resolver-dns-native-macos:XXX:osx-x86_64\")\r\n```\r\n\r\nWhere `XXX` is version of [netty-resolver-dns-native-macos](https://mvnrepository.com/artifact/io.netty/netty-resolver-dns-native-macos)", "I still get this issue on an Apple M1. I guess there needs to be a DnsServer provider that supports the M1?", "@philsmart we currently not support M1.", "OK, thanks for the update.", "@normanmaurer do you have any workaround or any idea when M1 will be supported?", "@normanmaurer still no update 3 months later ?", "Sorry missed this one. We added supported in the last release (4.1.68.Final).", "> Sorry missed this one. We added supported in the last release (4.1.68.Final).\r\n\r\nI still have the same issue.\r\n\r\n`2021-09-19 12:11:56.115 ERROR 4560 --- [undedElastic-11] i.n.r.d.DnsServerAddressStreamProviders : Unable to load io.netty.resolver.dns.macos.MacOSDnsServerAddressStreamProvider, fallback to system defaults. This may result in incorrect DNS resolutions on MacOS.\r\n\r\njava.lang.reflect.InvocationTargetException: null\r\n\tat java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) ~[na:na]\r\n\tat java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:78) ~[na:na]\r\n\tat java.base/jdk.internal.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45) ~[na:na]\r\n\tat java.base/java.lang.reflect.Constructor.newInstanceWithCaller(Constructor.java:499) ~[na:na]\r\n\tat java.base/java.lang.reflect.Constructor.newInstance(Constructor.java:480) ~[na:na]\r\n\tat io.netty.resolver.dns.DnsServerAddressStreamProviders.<clinit>(DnsServerAddressStreamProviders.java:64) ~[netty-resolver-dns-4.1.68.Final.jar:4.1.68.Final]\r\n\tat io.netty.resolver.dns.DnsNameResolverBuilder.<init>(DnsNameResolverBuilder.java:60) ~[netty-resolver-dns-4.1.68.Final.jar:4.1.68.Final]\r\n\tat reactor.netty.transport.NameResolverProvider.newNameResolverGroup(NameResolverProvider.java:432) ~[reactor-netty-core-1.0.11.jar:1.0.11]\r\n\tat reactor.netty.tcp.TcpResources.getOrCreateDefaultResolver(TcpResources.java:305) ~[reactor-netty-core-1.0.11.jar:1.0.11]\r\n\tat reactor.netty.http.HttpResources.getOrCreateDefaultResolver(HttpResources.java:139) ~[reactor-netty-http-1.0.11.jar:1.0.11]\r\n\tat reactor.netty.http.client.HttpClientConfig.defaultAddressResolverGroup(HttpClientConfig.java:382) ~[reactor-netty-http-1.0.11.jar:1.0.11]\r\n\tat reactor.netty.transport.ClientTransportConfig.resolverInternal(ClientTransportConfig.java:223) ~[reactor-netty-core-1.0.11.jar:1.0.11]\r\n\tat reactor.netty.http.client.HttpClientConfig.resolverInternal(HttpClientConfig.java:436) ~[reactor-netty-http-1.0.11.jar:1.0.11]\r\n\tat reactor.netty.http.client.HttpClientConnect$MonoHttpConnect.lambda$subscribe$0(HttpClientConnect.java:264) ~[reactor-netty-http-1.0.11.jar:1.0.11]\r\n\tat reactor.core.publisher.MonoCreate.subscribe(MonoCreate.java:57) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxRetryWhen.subscribe(FluxRetryWhen.java:77) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoRetryWhen.subscribeOrReturn(MonoRetryWhen.java:46) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.InternalMonoOperator.subscribe(InternalMonoOperator.java:57) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.netty.http.client.HttpClientConnect$MonoHttpConnect.subscribe(HttpClientConnect.java:271) ~[reactor-netty-http-1.0.11.jar:1.0.11]\r\n\tat reactor.core.publisher.Flux.subscribe(Flux.java:8468) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen$ThenIgnoreMain.subscribeNext(MonoIgnoreThen.java:255) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen.subscribe(MonoIgnoreThen.java:51) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoDefer.subscribe(MonoDefer.java:52) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoDefer.subscribe(MonoDefer.java:52) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen$ThenIgnoreMain.subscribeNext(MonoIgnoreThen.java:236) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen$ThenIgnoreMain.onComplete(MonoIgnoreThen.java:203) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxPeek$PeekSubscriber.onComplete(FluxPeek.java:260) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxMap$MapSubscriber.onComplete(FluxMap.java:142) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoNext$NextSubscriber.onComplete(MonoNext.java:102) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoNext$NextSubscriber.onNext(MonoNext.java:83) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxDematerialize$DematerializeSubscriber.onNext(FluxDematerialize.java:98) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxDematerialize$DematerializeSubscriber.onNext(FluxDematerialize.java:44) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxFlattenIterable$FlattenIterableSubscriber.drainAsync(FluxFlattenIterable.java:421) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxFlattenIterable$FlattenIterableSubscriber.drain(FluxFlattenIterable.java:686) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxFlattenIterable$FlattenIterableSubscriber.onNext(FluxFlattenIterable.java:250) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxSwitchIfEmpty$SwitchIfEmptySubscriber.onNext(FluxSwitchIfEmpty.java:74) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.Operators$MonoSubscriber.complete(Operators.java:1816) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoFlatMap$FlatMapInner.onNext(MonoFlatMap.java:249) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen$ThenIgnoreMain.complete(MonoIgnoreThen.java:284) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen$ThenIgnoreMain.onNext(MonoIgnoreThen.java:187) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen$ThenIgnoreMain.subscribeNext(MonoIgnoreThen.java:232) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen$ThenIgnoreMain.onComplete(MonoIgnoreThen.java:203) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreElements$IgnoreElementsSubscriber.onComplete(MonoIgnoreElements.java:89) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxPeek$PeekSubscriber.onComplete(FluxPeek.java:260) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxDematerialize$DematerializeSubscriber.onComplete(FluxDematerialize.java:121) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxDematerialize$DematerializeSubscriber.onNext(FluxDematerialize.java:91) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxDematerialize$DematerializeSubscriber.onNext(FluxDematerialize.java:44) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxIterable$IterableSubscription.fastPath(FluxIterable.java:340) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxIterable$IterableSubscription.request(FluxIterable.java:227) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxDematerialize$DematerializeSubscriber.request(FluxDematerialize.java:127) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxPeek$PeekSubscriber.request(FluxPeek.java:138) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreElements$IgnoreElementsSubscriber.onSubscribe(MonoIgnoreElements.java:72) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxPeek$PeekSubscriber.onSubscribe(FluxPeek.java:171) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxDematerialize$DematerializeSubscriber.onSubscribe(FluxDematerialize.java:77) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxIterable.subscribe(FluxIterable.java:165) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxIterable.subscribe(FluxIterable.java:87) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.Mono.subscribe(Mono.java:4361) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen$ThenIgnoreMain.subscribeNext(MonoIgnoreThen.java:255) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoIgnoreThen.subscribe(MonoIgnoreThen.java:51) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoFlatMap$FlatMapMain.onNext(MonoFlatMap.java:157) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.Operators$MonoSubscriber.complete(Operators.java:1816) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoCollectList$MonoCollectListSubscriber.onComplete(MonoCollectList.java:128) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.DrainUtils.postCompleteDrain(DrainUtils.java:132) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.DrainUtils.postComplete(DrainUtils.java:187) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxMaterialize$MaterializeSubscriber.onComplete(FluxMaterialize.java:141) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxTake$TakeSubscriber.onComplete(FluxTake.java:153) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxTake$TakeSubscriber.onNext(FluxTake.java:133) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxOnErrorResume$ResumeSubscriber.onNext(FluxOnErrorResume.java:79) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.SerializedSubscriber.onNext(SerializedSubscriber.java:99) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxTimeout$TimeoutMainSubscriber.onNext(FluxTimeout.java:180) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.Operators$MonoSubscriber.complete(Operators.java:1816) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.MonoCollectList$MonoCollectListSubscriber.onComplete(MonoCollectList.java:128) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat org.springframework.cloud.commons.publisher.FluxFirstNonEmptyEmitting$FirstNonEmptyEmittingSubscriber.onComplete(FluxFirstNonEmptyEmitting.java:325) ~[spring-cloud-commons-3.1.0-SNAPSHOT.jar:3.1.0-SNAPSHOT]\r\n\tat reactor.core.publisher.FluxSubscribeOn$SubscribeOnSubscriber.onComplete(FluxSubscribeOn.java:166) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.Operators$MultiSubscriptionSubscriber.onComplete(Operators.java:2058) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxIterable$IterableSubscription.fastPath(FluxIterable.java:362) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxIterable$IterableSubscription.request(FluxIterable.java:227) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.Operators$MultiSubscriptionSubscriber.set(Operators.java:2194) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxOnErrorResume$ResumeSubscriber.onSubscribe(FluxOnErrorResume.java:74) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxIterable.subscribe(FluxIterable.java:165) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxIterable.subscribe(FluxIterable.java:87) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxDefer.subscribe(FluxDefer.java:54) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.InternalFluxOperator.subscribe(InternalFluxOperator.java:62) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.publisher.FluxSubscribeOn$SubscribeOnSubscriber.run(FluxSubscribeOn.java:194) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.scheduler.WorkerTask.call(WorkerTask.java:84) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat reactor.core.scheduler.WorkerTask.call(WorkerTask.java:37) ~[reactor-core-3.4.10.jar:3.4.10]\r\n\tat java.base/java.util.concurrent.FutureTask.run(FutureTask.java:264) ~[na:na]\r\n\tat java.base/java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:304) ~[na:na]\r\n\tat java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1130) ~[na:na]\r\n\tat java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:630) ~[na:na]\r\n\tat java.base/java.lang.Thread.run(Thread.java:831) ~[na:na]\r\nCaused by: java.lang.UnsatisfiedLinkError: 'io.netty.resolver.dns.macos.DnsResolver[] io.netty.resolver.dns.macos.MacOSDnsServerAddressStreamProvider.resolvers()'\r\n\tat io.netty.resolver.dns.macos.MacOSDnsServerAddressStreamProvider.resolvers(Native Method) ~[netty-resolver-dns-native-macos-4.1.68.Final.jar:4.1.68.Final]\r\n\tat io.netty.resolver.dns.macos.MacOSDnsServerAddressStreamProvider.retrieveCurrentMappings(MacOSDnsServerAddressStreamProvider.java:127) ~[netty-resolver-dns-native-macos-4.1.68.Final.jar:4.1.68.Final]\r\n\tat io.netty.resolver.dns.macos.MacOSDnsServerAddressStreamProvider.<init>(MacOSDnsServerAddressStreamProvider.java:123) ~[netty-resolver-dns-native-macos-4.1.68.Final.jar:4.1.68.Final]\r\n\t... 91 common frames omitted\r\n\r\n`", "> @normanmaurer do you have any workaround or any idea when M1 will be supported?\r\n\r\nI use implementation(group = \"io.netty\",name = \"netty-resolver-dns-native-macos\", version = \"4.1.70.Final\", classifier = \"osx-aarch_64\")", "Same problem with M1 macbook. Tried this but didn't help: \r\n\r\n```\r\n<dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <version>4.1.72.Final</version>\r\n <classifier>osx-aarch_64</classifier>\r\n</dependency>\r\n```", "> Same problem with M1 macbook. Tried this but didn't help:\r\n> \r\n> ```\r\n> <dependency>\r\n> <groupId>io.netty</groupId>\r\n> <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n> <version>4.1.72.Final</version>\r\n> <classifier>osx-aarch_64</classifier>\r\n> </dependency>\r\n> ```\r\n\r\nworks fine for me on spring boot 2.5.3", "You could also use the following code to fix an error\r\n\r\n```\r\n<dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-all</artifactId>\r\n</dependency>\r\n```\r\n\r\nhttps://stackoverflow.com/a/65976398/3001953", "I have this problem on M1 MacOS with latest Spring Boot", "I have this problem on M1 MacOS with Spring boot 2.5.5", "use below dependency for M1 based mac, it works for me.\r\nimplementation 'io.netty:netty-resolver-dns-native-macos:4.1.72.Final:osx-aarch_64'\r\n", "> use below dependency for M1 based mac, it works for me.\r\n> implementation 'io.netty:netty-resolver-dns-native-macos:4.1.72.Final:osx-aarch_64'\r\n\r\nI test on M1 Mac and deploy to a Docker image that runs on x86. Sorry for my stupid question: How can I include above dependency during development and exclude it from my production Docker image?", "@ksilz there is no need to exclude as it will just not use it on linux if you use linux. If you run on Mac x86 on prod you will need to use the correct classifier when deploying. ", "Adding this resolved my issue with M1. This <classifier>osx-x86_64</classifier> did not work until i switched to below.\r\n```\r\n\r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <version>4.1.73.Final</version>\r\n <classifier>osx-aarch_64</classifier>\r\n </dependency>\r\n```", "> use below dependency for M1 based mac, it works for me. implementation 'io.netty:netty-resolver-dns-native-macos:4.1.72.Final:osx-aarch_64'\r\n\r\nIt works for me\r\n\r\n\r\n<dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <version>4.1.73.Final</version>\r\n <classifier>osx-aarch_64</classifier>\r\n</dependency>", "io.netty netty-resolver-dns-native-macos 4.1.73.Final osx-aarch_64 doesnt work on my M1 pro machine with Spring Boot 2.6.1. Are there any updates planned? \r\n\r\n**Update:** io.netty:netty-all:4.1.73.Final works on my M1 pro with Spring Boot 2.6.1", "I'm using OSX on Intel and I'm having the same problem when I try to run Spring Boot above 2.6.1. Spring Boot 2.6.1 works though. Adding the above mentioned package did not solve the issue for me.", "M1 mac os:\r\n1. exclude netty-resolver-dns-native-macos\r\n2. add classifier of osx-aarch_64.\r\n `<dependency>\r\n <groupId>io.projectreactor.netty</groupId>\r\n <artifactId>reactor-netty-core</artifactId>\r\n <exclusions>\r\n <exclusion>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n </exclusion>\r\n </exclusions>\r\n </dependency>\r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <version>4.1.72.Final</version>\r\n <classifier>osx-aarch_64</classifier>\r\n </dependency>`", "I am using on a MacBook Pro (x86_64) with Spring Boot 2.5.9, and spring-boot-starter-webflux 2.5.9 already pull in the native macos jar, but my running system cannot load it. I tried excluding the one from webflux and adding it directly, but it still doesn't work. Are there any other suggestions?", "To ensure a more timely response, please reply to this email with your POM file.I will resolve this question for you soon.发自我的iPhone------------------ Original ------------------From: Bryan Pugh ***@***.***>Date: Fri, Feb 18, 2022 8:26 AMTo: netty/netty ***@***.***>Cc: wanglianglike ***@***.***>, Comment ***@***.***>Subject: Re: [netty/netty] Unable to loadio.netty.resolver.dns.macos.MacOSDnsServerAddressStreamProvider (#11020)\nI am using on a MacBook Pro (x86_64) with Spring Book 2.5.9, and spring-boot-starter-webflux 2.5.9 already pull in the native macos jar, but my running system cannot load it. I tried excluding the one from webflux and adding it directly, but it still doesn't work. Are there any other suggestions?\n\n—Reply to this email directly, view it on GitHub, or unsubscribe.Triage notifications on the go with GitHub Mobile for iOS or Android.\nYou are receiving this because you commented.Message ID: ***@***.***>\n[\n{\n***@***.***\": \"http://schema.org\",\n***@***.***\": \"EmailMessage\",\n\"potentialAction\": {\n***@***.***\": \"ViewAction\",\n\"target\": \"https://github.com/netty/netty/issues/11020#issuecomment-1043672377\",\n\"url\": \"https://github.com/netty/netty/issues/11020#issuecomment-1043672377\",\n\"name\": \"View Issue\"\n},\n\"description\": \"View this Issue on GitHub\",\n\"publisher\": {\n***@***.***\": \"Organization\",\n\"name\": \"GitHub\",\n\"url\": \"https://github.com\"\n}\n}\n]", "I fixed the problem by upgrading from JDK 11.0.4 to 11.0.11.", "> I fixed the problem by upgrading from JDK 11.0.4 to 11.0.11.\r\n\r\nI can reproduce this problem on openjdk 17.0.2.", "> > I fixed the problem by upgrading from JDK 11.0.4 to 11.0.11.\r\n> \r\n> I can reproduce this problem on openjdk 17.0.2.\r\n\r\nI am using openjdk version \"17.0.2\", spring boot '3.0.0-SNAPSHOT' and still getting this exception on M1. Also tried with above fix but not working for me.", "> \r\n\r\nthis work for me, thank you", "> You could also use the following code to fix an error\r\n> \r\n> ```\r\n> <dependency>\r\n> <groupId>io.netty</groupId>\r\n> <artifactId>netty-all</artifactId>\r\n> </dependency>\r\n> ```\r\n> \r\n> https://stackoverflow.com/a/65976398/3001953\r\nThis fixed the issue on my MacBook Pro M1\r\n", "> Same problem with M1 macbook. Tried this but didn't help:\r\n> \r\n> ```\r\n> <dependency>\r\n> <groupId>io.netty</groupId>\r\n> <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n> <version>4.1.72.Final</version>\r\n> <classifier>osx-aarch_64</classifier>\r\n> </dependency>\r\n> ```\r\n\r\nthis worked for me", "The dependency has only runtime scope:\r\n\r\n`runtimeOnly \"io.netty:netty-resolver-dns-native-macos:4.1.76.Final:osx-aarch_64\"`\r\n\r\nHow can we make this dependency platform agnostic?\r\nWe don't really want to have a platform dependency in our java builds.", "I added `runtimeOnly(\"io.netty:netty-resolver-dns-native-macos:4.1.77.Final:osx-aarch_64\")` to my dependencies and still facing the issue. I played a little bit around with different versions of both netty and SpringBoot but nothing helps so far. I have an M1 chip on a MB Pro 2020. Any ideas on that?", "I have this problem on M1 MacOS with Spring boot 2.6.7", "Even when adding the correct version maven or gradle in the end might resolve to another version.\r\nTo validate the version you need to look into the resulting dependency tree.\r\n\r\n```\r\nmvn dependency:tree\r\ngradle dependencies\r\n```\r\n", "> I have this problem on M1 MacOS with Spring boot 2.6.7\r\n\r\nSame issue with spring boot 2.7.0 too", "I got it to work on my M1 Pro but none of the solutions above worked for me. The switch to Spring Boot 2.6.7 triggered the DNS resolver issue. I've had to override the dependency with the version below to get it to work locally (Spring Boot 2.6.7 provides 1.0.18):\r\n` <dependency>\r\n <groupId>io.projectreactor.netty</groupId>\r\n <artifactId>reactor-netty</artifactId>\r\n <version>0.9.20.RELEASE</version>\r\n </dependency>`\r\n\r\nGoing for an older version doesn't seem ideal though...", "workaround\r\n\r\n```kotlin\r\n// build.gradle.ktx\r\n\r\ndependencies {\r\n \r\n // ...\r\n\r\n // mac silicon only\r\n // https://github.com/apache/commons-lang/blob/master/src/main/java/org/apache/commons/lang3/SystemUtils.java#L1173\r\n val isMacOS: Boolean = System.getProperty(\"os.name\").startsWith(\"Mac OS X\")\r\n val architecture = System.getProperty(\"os.arch\").toLowerCase()\r\n if (isMacOS && architecture == \"aarch64\") {\r\n developmentOnly(\"io.netty:netty-resolver-dns-native-macos:4.1.68.Final:osx-aarch_64\")\r\n }\r\n\r\n // ...\r\n\r\n}\r\n```", "this works for me\r\n`implementation 'io.netty:netty-all'`\r\n`implementation(group: \"io.netty\", name: \"netty-resolver-dns-native-macos\", version: \"4.1.72.Final\", classifier: \"osx-aarch_64\")`", "It looks like this is an issue in [io.projectreactor.netty:reactor-netty-http:1.0.22](https://search.maven.org/artifact/io.projectreactor.netty/reactor-netty-http/1.0.22/jar), because that has a hardcoded dependency on `io.netty:netty-resolver-dns-native-macos` with classifier `osx-x86_64`.", "For Maven the workaround looks like following:\r\n`\r\n<dependency>\r\n <groupId>org.springframework.boot</groupId>\r\n <artifactId>spring-boot-starter-webflux</artifactId>\r\n <exclusions>\r\n <exclusion>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n </exclusion>\r\n </exclusions>\r\n </dependency>\r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <version>4.1.73.Final</version>\r\n <classifier>osx-aarch_64</classifier>\r\n </dependency>\r\n`\r\nFirst exclude the referenced library. And then add it explicit with the classifier. Worked on M2 chip MacBook Pro 2022.", "If you don't want to include native support properly (arm & intel) you should scope the dependency not as \"implementation\" but rather as \"compileOnly\" - (talking gradle slang :-) and document which dependencies can be included optionally.", "In Gradle, you can use [osdetector-gradle-plugin](https://github.com/google/osdetector-gradle-plugin) to fetch dependencies for specific platforms:\r\n\r\n```kotlin\r\n// build.gradle.kts\r\n\r\nplugins {\r\n id(\"com.google.osdetector\") version \"1.7.1\"\r\n}\r\n\r\ndependencies {\r\n if (osdetector.classifier == \"osx-aarch_64\") {\r\n runtimeOnly(\"io.netty:netty-resolver-dns-native-macos:4.1.77.Final:${osdetector.classifier}\")\r\n }\r\n}\r\n```", "> I still get this issue on an Apple M1. I guess there needs to be a DnsServer provider that supports the M1?\r\nme too M1 , i use this pom solve this problem. \r\n\r\nartifactId like this \r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-all</artifactId>\r\n <version>4.1.84.Final</version>\r\n </dependency>", "> <dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-webflux</artifactId> <exclusions> <exclusion> <groupId>io.netty</groupId> <artifactId>netty-resolver-dns-native-macos</artifactId> </exclusion> </exclusions> </dependency> <dependency> <groupId>io.netty</groupId> <artifactId>netty-resolver-dns-native-macos</artifactId> <version>4.1.73.Final</version> <classifier>osx-aarch_64</classifier> </dependency>\r\n\r\nThis has worked for me - M1 Pro 2022", "> In Gradle, you can use [osdetector-gradle-plugin](https://github.com/google/osdetector-gradle-plugin) to fetch dependencies for specific platforms:\r\n> \r\n> ```kotlin\r\n> // build.gradle.kts\r\n> \r\n> plugins {\r\n> id(\"com.google.osdetector\") version \"1.7.1\"\r\n> }\r\n> \r\n> dependencies {\r\n> if (osdetector.classifier == \"osx-aarch_64\") {\r\n> runtimeOnly(\"io.netty:netty-resolver-dns-native-macos:4.1.77.Final:${osdetector.classifier}\")\r\n> }\r\n> }\r\n> ```\r\n\r\nNote that if you deploy on a different OS/arch than the one you build on, this doesn't really solve the issue. This resolver is a runtime component, so the OS/arch of the build machine is not necessarily the same.\r\n\r\nHowever, I don't believe it's a problem to just include resolvers for multiple architectures in your application. The one to use will be resolved at runtime.", "**PROBLEM with null**\r\n\r\nWe have issues with Webflux and get null as a response (both in the console in our react native project and in Postman). \r\n\r\nWe have macos M1 and run the application with Spring boot version 3.0.0. We have added below mentioned dependency and it removed the bug with failed build. \r\n<dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <version>4.1.86.Final</version>\r\n <classifier>osx-aarch_64</classifier>\r\n</dependency>\r\n\r\nWe have used Webflux before and it worked like a charm with above mentioned dependency, but that project was run with Spring boot version 2.7.5. \r\n\r\nI have no idea if this matters at all, but I should also mention that we use Jakarta to serialize our entity objects. \r\n\r\nOur friend that uses PC has no problems with the serialization.\r\n\r\nAny ideas on what we might have done wrong? We get no error messages so, unfortunately, there is nothing to show.\r\nThanks in advance for any suggestions on how to move forward!", "I have this error today without knowing why, I tried every step here and nothing worked.", "I had trouble with an API key to google, which was an underlying problem. And when that was fixed, I used this and made it work: \r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <version>4.1.76.Final</version>\r\n <classifier>osx-aarch_64</classifier>\r\n </dependency>\r\n\r\nHope it works out for you AlexLombry!", "For maven and a project where this is only needed for local development on machines running mac M1 you can use a profile activated based on the OS:\r\n```\r\n <profile>\r\n <id>macos-m1</id>\r\n <activation>\r\n <os>\r\n <family>mac</family>\r\n <arch>aarch64</arch>\r\n </os>\r\n </activation>\r\n <dependencies>\r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <classifier>osx-aarch_64</classifier>\r\n </dependency>\r\n </dependencies>\r\n </profile>\r\n```", "mac M1 with build.gradle.kts\r\n\r\n```kts\r\n testImplementation(\r\n group = \"io.netty\",\r\n name = \"netty-resolver-dns-native-macos\",\r\n classifier = \"osx-aarch_64\"\r\n )\r\n```", "> mac M1 with build.gradle.kts\r\n> \r\n> ```kotlin\r\n> testImplementation(\r\n> group = \"io.netty\",\r\n> name = \"netty-resolver-dns-native-macos\",\r\n> classifier = \"osx-aarch_64\"\r\n> )\r\n> ```\r\n\r\nThanks! This worked for me. I also locked mine to Mac M1 machines only since we use different kinds of machines here at work:\r\n```groovy\r\n\tdef isMacOS = System.getProperty('os.name').startsWith('Mac OS X')\r\n\tdef architecture = System.getProperty('os.arch').toLowerCase()\r\n\tif (isMacOS && architecture == 'aarch64') {\r\n\t\ttestImplementation('io.netty:netty-resolver-dns-native-macos:4.1.90.Final:osx-aarch_64')\r\n\t}\r\n```", "> For maven and a project where this is only needed for local development on machines running mac M1 you can use a profile activated based on the OS:\r\n> \r\n> ```\r\n> <profile>\r\n> <id>macos-m1</id>\r\n> <activation>\r\n> <os>\r\n> <family>mac</family>\r\n> <arch>aarch64</arch>\r\n> </os>\r\n> </activation>\r\n> <dependencies>\r\n> <dependency>\r\n> <groupId>io.netty</groupId>\r\n> <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n> <classifier>osx-aarch_64</classifier>\r\n> </dependency>\r\n> </dependencies>\r\n> </profile>\r\n> ```\r\n\r\n~~FYI: Maven doesn't allow to declare dependencies in this place. Unfortunately the single solution is to declare this dependency with classifier `osx-aarch_64` per project or in some parent POM.~~\n\nWrong, see below.", "According to https://maven.apache.org/guides/introduction/introduction-to-profiles.html#Profiles_in_POMs Maven profiles can modify `<dependencies>`.", "> According to https://maven.apache.org/guides/introduction/introduction-to-profiles.html#Profiles_in_POMs Maven profiles can modify `<dependencies>`.\n\nHmmm, you're right. Seems that it works **only** when used in some of the project's POMs, but doesn't work outside the project e.g. in the user's `settings.xml` 🤔", "Yeah, this method worked for me when used in the project's POM.", "Fellows,\r\n\r\nI've stumbled upon this in my project recently. My two cents: while I do develop things on my M1 Mac, and this is an annoying issue, I never intend to run a production environment on my laptop 😂. With that been said, how do I make sure that the problem won't bite me when I deploy my code to a cloud provider? \r\n\r\nAny reflection on the 👆🏻 is greatly appreciated! \r\n\r\n\r\n ", "@singulart As far as I can see the code will only be loaded on an ARM Mac and it's also a pretty small library, so it should be harmless to add the dependency by default. Or, assuming your CI hardware is not ARM Mac, you can add the dependency only when building on ARM Macs using an auto-activating Maven profile like this:\r\n\r\n```\r\n<profile>\r\n <id>netty-resolver-dns-native-macos-aarch64</id>\r\n <activation>\r\n <os>\r\n <family>mac</family>\r\n <arch>aarch64</arch>\r\n </os>\r\n </activation>\r\n <dependencies>\r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <classifier>osx-aarch_64</classifier>\r\n </dependency>\r\n </dependencies>\r\n</profile>\r\n```", " implementation('io.netty:netty-resolver-dns-native-macos')\r\nThis worked for me today", "> io.netty netty-resolver-dns-native-macos 4.1.73.Final osx-aarch_64 doesnt work on my M1 pro machine with Spring Boot 2.6.1. Are there any updates planned?\r\n> \r\n> **Update:** io.netty:netty-all:4.1.73.Final works on my M1 pro with Spring Boot 2.6.1\r\n\r\nThis worked for me... thanks!", "\r\n ERROR 2481 --- [API-GATEWAY] [ender@6e0b642c}] [ ] i.n.r.d.DnsServerAddressStreamProviders : Unable to load io.netty.resolver.dns.macos.MacOSDnsServerAddressStreamProvider, fallback to system defaults. This may result in incorrect DNS resolutions on MacOS. Check whether you have a dependency on 'io.netty:netty-resolver-dns-native-macos'. Use DEBUG level to see the full stack: java.lang.UnsatisfiedLinkError: failed to load the required native library \r\n\r\nGetting the above error when i am using Spring Reactive Web dependancy on my spring boot microservice to create API-GATEWAY. I am using it on Macbook air M1 laptop. tried adding the following dependancy and other possible solutions like creating profile but it is not fixing my issue. \r\n<dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <classifier>osx-aarch_64</classifier>\r\n</dependency> \r\n\r\nPlease suggest me , how i can resolve this issue. ", "# following setup works. \r\n\r\nApple M1/Sonama 14.4.1 \r\nJava : 17-Temurin \r\nSpring Boot: 3.2.4 \r\nAzure, Postgresql \r\n\r\nworks fine. latest versions are not working. fyi. \r\n\r\n```xml\r\n <dependency>\r\n <groupId>io.projectreactor.netty</groupId>\r\n <artifactId>reactor-netty-http</artifactId>\r\n <version>1.0.39</version>\r\n </dependency>\r\n <dependency>\r\n <groupId>io.projectreactor.netty</groupId>\r\n <artifactId>reactor-netty-core</artifactId>\r\n <version>1.0.39</version>\r\n </dependency>\r\n <dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <classifier>osx-aarch_64</classifier>\r\n </dependency>\r\n```", "~~Mac M3 this is work for me~~\r\n\r\n```\r\n<dependency>\r\n <groupId>io.netty</groupId>\r\n <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n <classifier>${os.detected.classifier}</classifier>\r\n </dependency>\r\n\r\n<build>\r\n <extensions>\r\n <extension>\r\n <groupId>kr.motd.maven</groupId>\r\n <artifactId>os-maven-plugin</artifactId>\r\n <version>1.7.1</version>\r\n </extension>\r\n </extensions>\r\n </build>\r\n```\r\n\r\n:cry: It look like os.detected.classifier can not used in dependency tag.... \r\n\r\ntry https://github.com/netty/netty/issues/11020#issuecomment-1570501849\r\n\r\n", "> Mac M3 this is work for me\r\n> \r\n> ```\r\n> <dependency>\r\n> <groupId>io.netty</groupId>\r\n> <artifactId>netty-resolver-dns-native-macos</artifactId>\r\n> <classifier>${os.detected.classifier}</classifier>\r\n> </dependency>\r\n> </dependencies>\r\n> \r\n> <build>\r\n> <extensions>\r\n> <extension>\r\n> <groupId>kr.motd.maven</groupId>\r\n> <artifactId>os-maven-plugin</artifactId>\r\n> <version>1.7.1</version>\r\n> </extension>\r\n> </extensions>\r\n> </build>\r\n> ```\r\n\r\n@shalk You're making the `classifier` dynamic based on OS but `artifactId` contains a static `macos` part 😬 \r\n\r\nMoreover you're adding an additional Maven extension which currently has \"just\" 293 starts...\r\n\r\n@breun 's solution with a Maven profile and `activation` block is more robust and thought out, at least in my opinion.", "> Moreover you're adding an additional Maven plugin which currently has \"just\" 293 starts...\r\n\r\nThe extension is maintained by @trustin, and we use it in Netty." ]

[ "```suggestion\r\n + \"incorrect DNS resolutions on MacOS. Check whether you have a dependency on \"\r\n```", "```suggestion\r\n + \"incorrect DNS resolutions on MacOS. Check whether you have a dependency on \"\r\n```", "thanks", "thanks" ]

"2022-08-24T07:47:12Z"

[]

[ "resolver-dns/src/main/java/io/netty/resolver/dns/DnsServerAddressStreamProviders.java" ]

[ "resolver-dns/src/main/java/io/netty/resolver/dns/DnsServerAddressStreamProviders.java" ]

[]

"2021-02-12T04:45:51Z"

[ "keyword_pr_to_issue" ]

[ "Thanks for the report. This should be fixed by https://github.com/netty/netty/pull/12746", "Thank you for the quick fix!\r\n" ]

[]

"2022-08-26T23:58:59Z"

[]

[ "handler/src/main/java/io/netty/handler/ssl/SslContext.java" ]

[ "handler/src/main/java/io/netty/handler/ssl/SslContext.java" ]

[]

"2022-08-26T23:07:17Z"

[ "keyword_pr_to_issue" ]

[ "Thanks for the report." ]

[]

"2022-09-09T18:39:30Z"

[]

[ "buffer/src/main/java/io/netty/buffer/WrappedCompositeByteBuf.java" ]

[ "buffer/src/main/java/io/netty/buffer/WrappedCompositeByteBuf.java" ]

[ "buffer/src/test/java/io/netty/buffer/AbstractCompositeByteBufTest.java" ]

"2022-09-09T13:06:22Z"

[ "keyword_pr_to_issue" ]

[ "Just verified that the validation is fine for Safari but fails with Firefox." ]

[]

"2022-09-13T01:27:13Z"

[]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/HpackDecoder.java" ]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/HpackDecoder.java" ]

[ "codec-http2/src/test/java/io/netty/handler/codec/http2/DefaultHttp2HeadersDecoderTest.java" ]

"2022-09-13T00:57:02Z"

[ "keyword_issue_to_pr" ]

[ "\r\nMissing: header-> origin", "Please explain in more detail.", "When the websocketclient connection request header is missing the origin, a null pointer problem will occur", "@amizurov What's your take? Do we need validation to check for the origin header somewhere?", "Now he throws a null pointer. I think we can give some warning logs", "Hi @heixiaoma , @chrisvest. Let me take a look.", "@heixiaoma Could you please clarify, did you set `webSocketURL` -> \"/ws\" via constructor and `customHeaders` which contains `Host` but does not contain `Origin` header ?", "> @heixiaoma 您能否澄清一下，您是否`webSocketURL`通过构造函数设置了 -> \"/ws\" 并且`customHeaders`包含`Host`但不包含`Origin`header ？\r\n\r\nyes", "So the main problem here, that if we pass the `webSocketURL` without host (e.g. '/ws') we faced with NPE even when we building the `Host` header value. But if we pass the `Host` header value via `customHeaders` everything should works fine, but now we got the NPE on the building `Origin` header value (see https://github.com/netty/netty/issues/9673). \r\n\r\nI created the PR which should fix the `Origin` header behaviour (https://github.com/netty/netty/pull/12941). Also i think we can add additional check that if `webSocketURL` does not contain host and `Host` header does not present in custom headers we throw an invalid arg exception. @chrisvest WDYT ?\r\n\r\n", "@amizurov Sounds good. Do you want to add that check to https://github.com/netty/netty/pull/12941, or in a different PR?", "> @amizurov Sounds good. Do you want to add that check to https://github.com/netty/netty/pull/12941, or in a different PR?\n\nAdded to https://github.com/netty/netty/pull/12941", "I think this was fixed by #12941" ]

[ "This is an odd usage of assertThrows:\r\n\r\n```suggestion\r\n assertInstanceOf(IllegalArgumentException.class, handshakeFuture.cause());\r\n```", "This is an odd usage of assertThrows:\r\n\r\n```suggestion\r\n assertInstanceOf(IllegalArgumentException.class, handshakeFuture.cause());\r\n```\r\n```", "Done", "Done" ]

"2022-10-31T11:23:19Z"

[]

[ "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker00.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker07.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker08.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker13.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshakerFactory.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientProtocolConfig.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientProtocolHandler.java" ]

[ "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker00.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker07.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker08.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker13.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshakerFactory.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientProtocolConfig.java", "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketClientProtocolHandler.java" ]

[ "codec-http/src/test/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker00Test.java", "codec-http/src/test/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker07Test.java", "codec-http/src/test/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker08Test.java", "codec-http/src/test/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshaker13Test.java", "codec-http/src/test/java/io/netty/handler/codec/http/websocketx/WebSocketClientHandshakerTest.java" ]

"2022-10-26T06:16:21Z"

[ "keyword_pr_to_issue" ]

[]

[]

"2022-11-08T22:14:45Z"

[]

[ "common/src/main/java/io/netty/util/concurrent/FastThreadLocalThread.java", "common/src/main/java/io/netty/util/internal/Hidden.java" ]

[ "common/src/main/java/io/netty/util/concurrent/FastThreadLocalThread.java", "common/src/main/java/io/netty/util/internal/Hidden.java" ]

[ "transport-blockhound-tests/src/test/java/io/netty/util/internal/NettyBlockHoundIntegrationTest.java" ]

"2022-11-02T14:07:07Z"

[ "keyword_pr_to_issue" ]

[ "I can contribute a fix for this." ]

[ "```suggestion\r\n // Check for dynamic table size updates, which must occur at the beginning:\r\n // https://www.rfc-editor.org/rfc/rfc7541.html#section-4.2\r\n decodeDynamicTableSizeUpdates(in);\r\n```", "```suggestion\r\n // See https://www.rfc-editor.org/rfc/rfc7541.html#section-4.2\r\n throw connectionError(COMPRESSION_ERROR, \"Dynamic table size update must happen \" +\r\n```", "can I commit this directly from github interface ?", "Yep... I just pressed the button for you :)", "thanks :-D" ]

"2022-11-11T09:32:53Z"

[]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/HpackDecoder.java" ]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/HpackDecoder.java" ]

[ "codec-http2/src/test/java/io/netty/handler/codec/http2/HpackDecoderTest.java" ]

"2022-11-09T15:22:23Z"

[ "keyword_pr_to_issue" ]

[ "@normanmaurer PTAL" ]

[ "I think this is not correct... this should still be optional otherwise we have it as non optional dependency for handler as well. \r\n\r\nSame its true for the other change in here. " ]

"2022-11-16T12:23:20Z"

[]

[ "all/pom.xml", "bom/pom.xml", "handler-ssl-ocsp/pom.xml" ]

[ "all/pom.xml", "bom/pom.xml", "handler-ssl-ocsp/pom.xml" ]

[]

"2022-11-16T12:08:43Z"

[ "keyword_pr_to_issue" ]

[ "@chrisvest I investigated the issue and didn't find a behavior change in HashedWheelTimer. I used HashedWheelTimerTest and made several variations of testExecutionOnTime. \r\n\r\nThere might be some subtle change that causes the issue, and we'll just have to deal with that in Pulsar.", "@lhotari After reading your comment, I reviewed my code and realised I made a mistake in #12888.\r\nMy optimisation based on the assumption that the tasks in a bucket are originally in the order of execution time, which are in fact not.\r\nSo breaking the loop at `execRound > currRound` may cause latter tasks not respond in time.", "Sorry for making that trouble. I already made a PR #13021 to revert it. @lhotari @chrisvest ", "> @lhotari After reading your comment, I reviewed my code and realised I made a mistake in #12888.\n> My optimisation based on the assumption that the tasks in a bucket are originally in the order of execution time, which are in fact not.\n> So breaking the loop at `execRound > currRound` may cause latter tasks not respond in time.\n\n@needmorecode Thanks for the quick confirmation and investigation. Your explanation makes sense. I missed that case when I was trying to add a unit test that would prove an issue." ]

[]

"2022-11-27T00:16:58Z"

[]

[ "common/src/main/java/io/netty/util/HashedWheelTimer.java" ]

[ "common/src/main/java/io/netty/util/HashedWheelTimer.java" ]

[]

"2022-11-25T19:20:42Z"

[ "keyword_pr_to_issue" ]

[ "Related to https://github.com/spring-cloud/spring-cloud-gateway/issues/2493" ]

[ "I think we still need to ensure we only create a `SSLHandshakeException` if we are really still handshaking.\r\n```suggestion\r\n final Exception exception;\r\n if (handshakePromise.isDone()) {\r\n exception = new ClosedChannelException();\r\n } else {\r\n // Closed before the handshake was done.\r\n exception = new SSLHandshakeException(\"Connection closed during handshake\");\r\n exception.initCause(new ClosedChannelException()); \r\n }\r\n```", "There is a possibility that the channel will become inactive even before it was able to read anything and start the handshake. Consider `!isStateSet(STATE_HANDSHAKE_STARTED) || handshakePromise.isDone()` ", "It doesn't provide much value and produces too much logging output if we generate stack-trace twice from the same method. Consider making either `ClosedChannelException` or `SSLHandshakeException` stack-less in this case.", "In the previous round of feedback the logic was opposite. The above condition was for `ClosedChannelException`. Since now we changed the logic, we have to flip this condition as well. Consider: `isStateSet(STATE_HANDSHAKE_STARTED) && !handshakePromise.isDone()`.", "\"before\" doesn't help to understand if it was \"before handshake started\" or \"after it started but before it complete\". Consider clarifying that the handshake started, but didn't complete", "Since it's a subtype of `SSLHandshakeException` with the same meaning, consider making it pkg-private.", "Yeah, sorry. My bad. ;/", "```suggestion\r\n new StacklessSSLHandshakeException(\"Connection closed while SSL/TLS handshake was in progress\"));\r\n```" ]

"2023-01-05T19:39:01Z"

[]

[ "handler/src/main/java/io/netty/handler/ssl/SslHandler.java" ]

[ "handler/src/main/java/io/netty/handler/ssl/SslHandler.java", "handler/src/main/java/io/netty/handler/ssl/StacklessSSLHandshakeException.java" ]

[ "handler/src/test/java/io/netty/handler/ssl/SslHandlerTest.java" ]

"2022-01-13T13:06:59Z"

[ "keyword_pr_to_issue" ]

[ "Can you add instructions for building and running your repro?", "Sorry yes, updated comment.", "@perezd How do you make the repro use a Netty version that's built and installed into the local `~/.m2/` repository? I'm trying to make it use a local `4.1.53.Final-SNAPSHOT` version, but just changing the `NETTY_VERSION` constant in the `WORKSPACE` file is not enough.", "This would work, if you add the jars to the BUILD file:\r\nhttps://docs.bazel.build/versions/master/be/java.html#java_import\r\n\r\nThen just changes `deps` to point to that new target and it should work.", "I think this was fixed ?", "/cc @chrisvest ", "I get the following error, which I don't understand, when I run the reproducer build with a 4.1.54.Final-SNAPSHOT version of Netty:\r\n\r\n```\r\n[chris@localhost netty-epoll-native-repro]$ bazel run :main-native --sandbox_debug\r\nINFO: Analyzed target //:main-native (1 packages loaded, 11 targets configured).\r\nINFO: Found 1 target...\r\nINFO: From Extracting interface //third_party:netty:\r\n1603979213.739407504: src/main/tools/linux-sandbox.cc:152: calling pipe(2)...\r\n1603979213.739430730: src/main/tools/linux-sandbox.cc:171: calling clone(2)...\r\n1603979213.739633277: src/main/tools/linux-sandbox.cc:180: linux-sandbox-pid1 has PID 30613\r\n1603979213.739668113: src/main/tools/linux-sandbox-pid1.cc:434: Pid1Main started\r\n1603979213.739768241: src/main/tools/linux-sandbox.cc:197: done manipulating pipes\r\n1603979213.739791637: src/main/tools/linux-sandbox-pid1.cc:176: working dir: /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/48/execroot/grl\r\n1603979213.739807831: src/main/tools/linux-sandbox-pid1.cc:208: writable: /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/48/execroot/grl\r\n1603979213.739815223: src/main/tools/linux-sandbox-pid1.cc:208: writable: /tmp\r\n1603979213.739820767: src/main/tools/linux-sandbox-pid1.cc:208: writable: /dev/shm\r\n1603979213.739867607: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /\r\n1603979213.739877071: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys\r\n1603979213.739882570: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/kernel/security\r\n1603979213.739889160: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/cgroup\r\n1603979213.739894040: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/pstore\r\n1603979213.739899400: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/bpf\r\n1603979213.739904071: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/kernel/config\r\n1603979213.739908633: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/selinux\r\n1603979213.739913332: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/kernel/tracing\r\n1603979213.739917879: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/kernel/debug\r\n1603979213.739922570: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/fuse/connections\r\n1603979213.739928426: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /dev\r\n1603979213.739932441: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /dev/shm\r\n1603979213.739954786: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /dev/pts\r\n1603979213.739960071: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /dev/mqueue\r\n1603979213.739963898: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /dev/hugepages\r\n1603979213.739967826: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /run\r\n1603979213.739972297: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /run/vmblock-fuse\r\n1603979213.739976810: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /run/user/1000\r\n1603979213.739981507: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /run/user/1000/gvfs\r\n1603979213.739985922: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /proc\r\n1603979213.739990035: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /proc/sys/fs/binfmt_misc\r\n1603979213.739996764: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /tmp\r\n1603979213.740000021: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /boot\r\n1603979213.740003624: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /var/lib/nfs/rpc_pipefs\r\n1603979213.740010159: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /mnt/hgfs\r\n1603979213.740017960: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/48/execroot/grl\r\n1603979213.740022873: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/48/execroot/grl\r\n1603979213.740026346: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /tmp\r\n1603979213.740029038: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /dev/shm\r\n1603979213.740089170: src/main/tools/linux-sandbox-pid1.cc:363: calling fork...\r\n1603979213.740196745: src/main/tools/linux-sandbox-pid1.cc:393: child started with PID 2\r\n1603979213.757556894: src/main/tools/linux-sandbox-pid1.cc:410: wait returned pid=2, status=0x00\r\n1603979213.757569016: src/main/tools/linux-sandbox-pid1.cc:428: child exited normally with code 0\r\n1603979213.757889995: src/main/tools/linux-sandbox.cc:233: child exited normally with code 0\r\nERROR: /home/chris/netty-epoll-native-repro/BUILD:19:13: Action main-native-bin failed (Exit 1): linux-sandbox failed: error executing command \r\n (cd /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/50/execroot/grl && \\\r\n exec env - \\\r\n PATH=/usr/bin:/bin \\\r\n TMPDIR=/tmp \\\r\n /home/chris/.cache/bazel/_bazel_chris/install/792f43108e6f3e1e71623e5a11a5a382/linux-sandbox -t 15 -w /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/50/execroot/grl -w /tmp -w /dev/shm -D -- external/graal/bin/native-image --no-server --no-fallback -cp bazel-out/k8-fastbuild/bin/libmain.jar:third_party/netty/netty-common-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-buffer-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-codec-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-codec-http-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-transport-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-transport-native-unix-common-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-transport-native-epoll-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-handler-4.1.54.Final-SNAPSHOT.jar '-H:Class=Main' '-H:Name=bazel-out/k8-fastbuild/bin/main-native-bin' '-H:CCompilerPath=/usr/bin/gcc' -H:+ReportExceptionStackTraces -H:+TraceClassInitialization --static) linux-sandbox failed: error executing command \r\n (cd /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/50/execroot/grl && \\\r\n exec env - \\\r\n PATH=/usr/bin:/bin \\\r\n TMPDIR=/tmp \\\r\n /home/chris/.cache/bazel/_bazel_chris/install/792f43108e6f3e1e71623e5a11a5a382/linux-sandbox -t 15 -w /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/50/execroot/grl -w /tmp -w /dev/shm -D -- external/graal/bin/native-image --no-server --no-fallback -cp bazel-out/k8-fastbuild/bin/libmain.jar:third_party/netty/netty-common-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-buffer-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-codec-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-codec-http-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-transport-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-transport-native-unix-common-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-transport-native-epoll-4.1.54.Final-SNAPSHOT.jar:third_party/netty/netty-handler-4.1.54.Final-SNAPSHOT.jar '-H:Class=Main' '-H:Name=bazel-out/k8-fastbuild/bin/main-native-bin' '-H:CCompilerPath=/usr/bin/gcc' -H:+ReportExceptionStackTraces -H:+TraceClassInitialization --static)\r\n1603979213.837758895: src/main/tools/linux-sandbox.cc:152: calling pipe(2)...\r\n1603979213.837780965: src/main/tools/linux-sandbox.cc:171: calling clone(2)...\r\n1603979213.837993852: src/main/tools/linux-sandbox.cc:180: linux-sandbox-pid1 has PID 30618\r\n1603979213.838038905: src/main/tools/linux-sandbox-pid1.cc:434: Pid1Main started\r\n1603979213.838124208: src/main/tools/linux-sandbox.cc:197: done manipulating pipes\r\n1603979213.838170860: src/main/tools/linux-sandbox-pid1.cc:176: working dir: /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/50/execroot/grl\r\n1603979213.838188901: src/main/tools/linux-sandbox-pid1.cc:208: writable: /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/50/execroot/grl\r\n1603979213.838195746: src/main/tools/linux-sandbox-pid1.cc:208: writable: /tmp\r\n1603979213.838201268: src/main/tools/linux-sandbox-pid1.cc:208: writable: /dev/shm\r\n1603979213.838264214: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /\r\n1603979213.838274941: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys\r\n1603979213.838280548: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/kernel/security\r\n1603979213.838287149: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/cgroup\r\n1603979213.838292243: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/pstore\r\n1603979213.838297864: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/bpf\r\n1603979213.838302218: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/kernel/config\r\n1603979213.838306876: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/selinux\r\n1603979213.838311431: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/kernel/tracing\r\n1603979213.838315842: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/kernel/debug\r\n1603979213.838320456: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /sys/fs/fuse/connections\r\n1603979213.838326254: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /dev\r\n1603979213.838330415: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /dev/shm\r\n1603979213.838353771: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /dev/pts\r\n1603979213.838358470: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /dev/mqueue\r\n1603979213.838362393: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /dev/hugepages\r\n1603979213.838366317: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /run\r\n1603979213.838370249: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /run/vmblock-fuse\r\n1603979213.838374337: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /run/user/1000\r\n1603979213.838378989: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /run/user/1000/gvfs\r\n1603979213.838383515: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /proc\r\n1603979213.838387608: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /proc/sys/fs/binfmt_misc\r\n1603979213.838394500: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /tmp\r\n1603979213.838397649: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /boot\r\n1603979213.838401352: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /var/lib/nfs/rpc_pipefs\r\n1603979213.838407632: src/main/tools/linux-sandbox-pid1.cc:278: remount ro: /mnt/hgfs\r\n1603979213.838414908: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/50/execroot/grl\r\n1603979213.838419872: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /home/chris/.cache/bazel/_bazel_chris/79c5fdc3f25858db1b7fbca7ad47271b/sandbox/linux-sandbox/50/execroot/grl\r\n1603979213.838423186: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /tmp\r\n1603979213.838425830: src/main/tools/linux-sandbox-pid1.cc:278: remount rw: /dev/shm\r\n1603979213.838471119: src/main/tools/linux-sandbox-pid1.cc:363: calling fork...\r\n1603979213.838560767: src/main/tools/linux-sandbox-pid1.cc:393: child started with PID 2\r\nFatal error: com.oracle.graal.pointsto.util.AnalysisError$ParsingError: Error encountered while parsing com.oracle.svm.reflect.ClassLoader_defineClass_c6c343b4d6dc22ca64eb2d8503b13ac9c340dcb3.invoke(java.lang.Object, java.lang.Object[]) \r\nParsing context:\r\n\tparsing java.lang.reflect.Method.invoke(Method.java:566)\r\n\tparsing io.netty.util.internal.CleanerJava9.freeDirectBuffer(CleanerJava9.java:88)\r\n\tparsing io.netty.util.internal.PlatformDependent.freeDirectBuffer(PlatformDependent.java:483)\r\n\tparsing io.netty.channel.unix.Buffer.free(Buffer.java:33)\r\n\tparsing io.netty.channel.epoll.EpollEventArray.free(EpollEventArray.java:92)\r\n\tparsing io.netty.channel.epoll.EpollEventLoop.cleanup(EpollEventLoop.java:552)\r\n\tparsing io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:1036)\r\n\tparsing com.oracle.svm.core.jdk.RuntimeSupport.executeHooks(RuntimeSupport.java:144)\r\n\tparsing com.oracle.svm.core.jdk.RuntimeSupport.executeTearDownHooks(RuntimeSupport.java:121)\r\n\tparsing com.oracle.svm.core.graal.snippets.CEntryPointSnippets.tearDownIsolate(CEntryPointSnippets.java:361)\r\n\r\n\tat com.oracle.graal.pointsto.util.AnalysisError.parsingError(AnalysisError.java:138)\r\n\tat com.oracle.graal.pointsto.flow.MethodTypeFlow.doParse(MethodTypeFlow.java:327)\r\n\tat com.oracle.graal.pointsto.flow.MethodTypeFlow.ensureParsed(MethodTypeFlow.java:300)\r\n\tat com.oracle.graal.pointsto.flow.MethodTypeFlow.addContext(MethodTypeFlow.java:107)\r\n\tat com.oracle.graal.pointsto.DefaultAnalysisPolicy$DefaultVirtualInvokeTypeFlow.onObservedUpdate(DefaultAnalysisPolicy.java:191)\r\n\tat com.oracle.graal.pointsto.flow.TypeFlow.notifyObservers(TypeFlow.java:343)\r\n\tat com.oracle.graal.pointsto.flow.TypeFlow.update(TypeFlow.java:385)\r\n\tat com.oracle.graal.pointsto.BigBang$2.run(BigBang.java:511)\r\n\tat com.oracle.graal.pointsto.util.CompletionExecutor.lambda$execute$0(CompletionExecutor.java:171)\r\n\tat java.base/java.util.concurrent.ForkJoinTask$RunnableExecuteAction.exec(ForkJoinTask.java:1426)\r\n\tat java.base/java.util.concurrent.ForkJoinTask.doExec(ForkJoinTask.java:290)\r\n\tat java.base/java.util.concurrent.ForkJoinPool$WorkQueue.topLevelExec(ForkJoinPool.java:1020)\r\n\tat java.base/java.util.concurrent.ForkJoinPool.scan(ForkJoinPool.java:1656)\r\n\tat java.base/java.util.concurrent.ForkJoinPool.runWorker(ForkJoinPool.java:1594)\r\n\tat java.base/java.util.concurrent.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:177)\r\nCaused by: com.oracle.svm.hosted.substitute.DeletedElementException: Unsupported method java.lang.ClassLoader.defineClass(String, byte[], int, int) is reachable: The declaring class of this element has been substituted, but this element is not present in the substitution class\r\nTo diagnose the issue, you can add the option --report-unsupported-elements-at-runtime. The unsupported element is then reported at run time when it is accessed the first time.\r\n\tat com.oracle.svm.hosted.substitute.AnnotationSubstitutionProcessor.lookup(AnnotationSubstitutionProcessor.java:183)\r\n\tat com.oracle.graal.pointsto.infrastructure.SubstitutionProcessor$ChainedSubstitutionProcessor.lookup(SubstitutionProcessor.java:128)\r\n\tat com.oracle.graal.pointsto.infrastructure.SubstitutionProcessor$ChainedSubstitutionProcessor.lookup(SubstitutionProcessor.java:128)\r\n\tat com.oracle.graal.pointsto.meta.AnalysisUniverse.lookupAllowUnresolved(AnalysisUniverse.java:397)\r\n\tat com.oracle.graal.pointsto.meta.AnalysisUniverse.lookup(AnalysisUniverse.java:377)\r\n\tat com.oracle.graal.pointsto.meta.AnalysisUniverse.lookup(AnalysisUniverse.java:75)\r\n\tat com.oracle.graal.pointsto.infrastructure.UniverseMetaAccess.lookupJavaMethod(UniverseMetaAccess.java:93)\r\n\tat com.oracle.graal.pointsto.meta.AnalysisMetaAccess.lookupJavaMethod(AnalysisMetaAccess.java:66)\r\n\tat com.oracle.graal.pointsto.meta.AnalysisMetaAccess.lookupJavaMethod(AnalysisMetaAccess.java:39)\r\n\tat com.oracle.svm.reflect.hosted.ReflectionSubstitutionType$ReflectiveInvokeMethod.buildGraph(ReflectionSubstitutionType.java:511)\r\n\tat com.oracle.graal.pointsto.meta.AnalysisMethod.buildGraph(AnalysisMethod.java:319)\r\n\tat com.oracle.graal.pointsto.flow.MethodTypeFlowBuilder.parse(MethodTypeFlowBuilder.java:185)\r\n\tat com.oracle.graal.pointsto.flow.MethodTypeFlowBuilder.apply(MethodTypeFlowBuilder.java:340)\r\n\tat com.oracle.graal.pointsto.flow.MethodTypeFlow.doParse(MethodTypeFlow.java:310)\r\n\t... 13 more\r\nError: Image build request failed with exit status 1\r\n1603979228.378204816: src/main/tools/linux-sandbox-pid1.cc:410: wait returned pid=2, status=0x100\r\n1603979228.378218295: src/main/tools/linux-sandbox-pid1.cc:428: child exited normally with code 1\r\n1603979228.378818653: src/main/tools/linux-sandbox.cc:233: child exited normally with code 1\r\n[bazel-out/k8-fastbuild/bin/main-native-bin:21] classlist: 2,611.26 ms\r\n[bazel-out/k8-fastbuild/bin/main-native-bin:21] (cap): 589.29 ms\r\n[bazel-out/k8-fastbuild/bin/main-native-bin:21] setup: 1,557.43 ms\r\n[bazel-out/k8-fastbuild/bin/main-native-bin:21] analysis: 9,680.26 ms\r\nTarget //:main-native failed to build\r\nUse --verbose_failures to see the command lines of failed build steps.\r\nINFO: Elapsed time: 14.778s, Critical Path: 14.64s\r\nINFO: 6 processes: 4 internal, 1 linux-sandbox, 1 worker.\r\nFAILED: Build did NOT complete successfully\r\nFAILED: Build did NOT complete successfully\r\n[chris@localhost netty-epoll-native-repro]$ \r\n```\r\n\r\nThese are the changes I made to the reproducer repo in order to get it sort-of building with custom jar files:\r\n\r\n```patch\r\ndiff --git a/BUILD b/BUILD\r\nindex 1c8f0c3..c953f9b 100644\r\n--- a/BUILD\r\n+++ b/BUILD\r\n@@ -5,10 +5,11 @@ java_library(\r\n name = \"main\",\r\n srcs = [\"Main.java\"],\r\n deps = [\r\n- \"@maven//:io_netty_netty_codec_http\",\r\n- \"@maven//:io_netty_netty_transport\",\r\n- \"@maven//:io_netty_netty_transport_native_epoll_linux_x86_64\",\r\n- \"@maven//:io_netty_netty_handler\",\r\n+ \t\"//third_party:netty\"\r\n+ #\"@maven//:io_netty_netty_codec_http\",\r\n+ #\"@maven//:io_netty_netty_transport\",\r\n+ #\"@maven//:io_netty_netty_transport_native_epoll_linux_x86_64\",\r\n+ #\"@maven//:io_netty_netty_handler\",\r\n ],\r\n )\r\n \r\ndiff --git a/WORKSPACE b/WORKSPACE\r\nindex 1beffb0..e0cb416 100644\r\n--- a/WORKSPACE\r\n+++ b/WORKSPACE\r\n@@ -4,8 +4,8 @@ load(\"@bazel_tools//tools/build_defs/repo:http.bzl\", \"http_archive\")\r\n \r\n http_archive(\r\n name = \"rules_graal\",\r\n- sha256 = \"c1445dd6263b17b110da69178c238dc0925bbea1bffc93c46f5af4f8027245cc\",\r\n- strip_prefix = \"rules_graal-master\",\r\n+ sha256 = \"f5fc694e0abe24699987aab1c4054536c1792ed8a17123a9d7223511fac2367d\",\r\n+ strip_prefix = \"rules_graal-1b5f6deef8352c4b64541bb0f699fd2af5ae6b0b\",\r\n urls = [\r\n \"https://github.com/andyscott/rules_graal/archive/1b5f6deef8352c4b64541bb0f699fd2af5ae6b0b.zip\",\r\n ],\r\n@@ -72,3 +72,4 @@ maven_install(\r\n load(\"@maven//:defs.bzl\", \"pinned_maven_install\")\r\n \r\n pinned_maven_install()\r\n+\r\ndiff --git a/third_party/BUILD b/third_party/BUILD\r\nnew file mode 100644\r\nindex 0000000..aa16a03\r\n--- /dev/null\r\n+++ b/third_party/BUILD\r\n@@ -0,0 +1,14 @@\r\n+package(default_visibility = [\"//visibility:public\"])\r\n+java_import(\r\n+\tname = \"netty\",\r\n+\tjars = [\r\n+\t \"netty/netty-common-4.1.54.Final-SNAPSHOT.jar\",\r\n+\t \"netty/netty-buffer-4.1.54.Final-SNAPSHOT.jar\",\r\n+\t \"netty/netty-codec-4.1.54.Final-SNAPSHOT.jar\",\r\n+\t \"netty/netty-codec-http-4.1.54.Final-SNAPSHOT.jar\",\r\n+\t \"netty/netty-transport-4.1.54.Final-SNAPSHOT.jar\",\r\n+\t \"netty/netty-transport-native-unix-common-4.1.54.Final-SNAPSHOT.jar\",\r\n+\t \"netty/netty-transport-native-epoll-4.1.54.Final-SNAPSHOT.jar\",\r\n+\t \"netty/netty-handler-4.1.54.Final-SNAPSHOT.jar\",\r\n+\t],\r\n+)\r\n```", "The Epoll transport is currently not supported in Graal because it loads classes from JNI code, and those classes needs to be initialised at runtime.\r\n\r\nYou can work around this by adding something like this to the `native-image` command:\r\n\r\n```\r\n--initialize-at-run-time=io.netty.handler.ssl.util.ThreadLocalInsecureRandom,io.netty.channel.unix,io.netty.channel.epoll,io.netty.channel.kqueue\r\n```", "@perezd I tried making epoll work in Graal native-image in #11163, but it's running into issues I don't fully understand. If you have any ideas feel free to chime in.", "Bit out if my depths here but wondering if it could have to do with JNI config? https://www.graalvm.org/reference-manual/native-image/JNI/\r\n\r\nDoes netty use reflection related to JNI? This could cause an issue potentially?", "@chrisvest So where does this stand now that https://github.com/netty/netty/pull/11163 has been closed?", "@ScottPierce You will have to use NIO transport with native-image for the foreseeable future. ", "What if we don't have the ability to choose? The library we use just pulls this stuff in.", "@ScottPierce we are happy to review a PR but we clearly not have the needed knowledge to fix this ourselves. Sorry ", "I wasn't being snarky. I was more asking if I can exclude dependencies and the pull in the right one and have a reasonable expectation of it still working. Thanks! ", "@ScottPierce I guess that is more the question for the library that pulls in netty :)", "Hello, I try to build a native image of a **Micronaut** service and have lots of similar errors \r\n\r\n> (Error: Classes that should be initialized at run time got initialized during image building:\r\n> io.grpc.netty.shaded.io.netty.buffer.ByteBufUtil the class was requested to be initialized at run time (from jar:file:///home/jacek/.gradle/caches/modules-2/files-2.1/io.grpc/grpc-netty-shaded/1.41.0/43fee785c101b3bfa3f7d916d8b47d5d736cbff2/grpc-netty-shaded-1.41.0.jar!/META-INF/native-image/io.grpc.netty.shaded.io.netty/buffer/native-image.properties with 'io.grpc.netty.shaded.io.netty.buffer.ByteBufUtil'). To see why io.grpc.netty.shaded.io.netty.buffer.ByteBufUtil got initialized use --trace-class-initialization=io.grpc.netty.shaded.io.netty.buffer.ByteBufUtil\r\n> io.grpc.netty.shaded.io.netty.buffer.ByteBufAllocator the class was requested to be initialized at run time (from jar:file:///home/jacek/.gradle/caches/modules-2/files-2.1/io.grpc/grpc-netty-shaded/1.41.0/43fee785c101b3bfa3f7d916d8b47d5d736cbff2/grpc-netty-shaded-1.41.0.jar!/META-INF/native-image/io.grpc.netty.shaded.io.netty/buffer/native-image.properties with 'io.grpc.netty.shaded.io.netty.buffer.ByteBufAllocator'). To see why io.grpc.netty.shaded.io.netty.buffer.ByteBufAllocator got initialized use --trace-class-initialization=io.grpc.netty.shaded.io.netty.buffer.ByteBufAllocator\r\n> .............\r\n\r\nIs there a known solution or a workaround for this?\r\nFrom various suggestions I tried\r\n```\r\n buildArgs.add(\r\n \"--initialize-at-run-time=\" +\r\n \"io.grpc.netty.shaded.io.netty.util.internal.logging.Log4JLogger,\" +\r\n \"io.grpc.netty.shaded.io.netty.handler.ssl.JdkNpnApplicationProtocolNegotiator,\" +\r\n \"io.netty.handler.ssl.util.ThreadLocalInsecureRandom,\" +\r\n \"io.netty.channel.unix,io.netty.channel.epoll,io.netty.channel.kqueue\"\r\n)\r\n```\r\nbut it didn't help.\r\nIs there some radical workaround which maybe sacrifice performance but at least allow to successfully compile?", "I also ran into this again with Netty 4.1.70 and GraalVM 21.3.0 but was able to get it working with:\r\n```\r\n --initialize-at-run-time=io.netty.channel.DefaultFileRegion\r\n --initialize-at-run-time=io.netty.channel.epoll.Native\r\n --initialize-at-run-time=io.netty.channel.epoll.Epoll\r\n --initialize-at-run-time=io.netty.channel.epoll.EpollEventLoop\r\n --initialize-at-run-time=io.netty.channel.epoll.EpollEventArray\r\n --initialize-at-run-time=io.netty.channel.kqueue.KQueue\r\n --initialize-at-run-time=io.netty.channel.kqueue.KQueueEventLoop\r\n --initialize-at-run-time=io.netty.channel.kqueue.KQueueEventArray\r\n --initialize-at-run-time=io.netty.channel.kqueue.Native\r\n --initialize-at-run-time=io.netty.channel.unix.Limits\r\n --initialize-at-run-time=io.netty.channel.unix.Errors\r\n --initialize-at-run-time=io.netty.channel.unix.IovArray\r\n```", "I tinkered a bit with epoll and GraalVM 23.0 - I managed to build the original sample project, but I couldn't figure out how to tweak Bazel to use a newer GraalVM version - I've created the following sample repository that uses Gradle: https://github.com/gradinac/netty-epoll-native-image. This project uses the original reproducer from @perezd, but uses Gradle as the build system\r\n\r\nI've hit two different categories of issues:\r\n 1. Class initialization problems - Netty 4.1 is by default fully initialized at image build time. However, epoll classes require native code and shouldn't be initialized at image build time. On top of that, these classes also use other classes in their static initializer that shouldn't be used during the image build (example: `io.netty.util.AbstractReferenceCounted`). Thank you @jamesward for the above list! :) It works without any modifications, but I've modified it a little bit in the repo. A sidenote: for Gradle and Maven projects, users can also add the [Native Build Tools](https://github.com/graalvm/native-build-tools) to their project and enable the metadata repository. Netty 4.1 metadata in this repository will automatically disable build-time initialization of Netty\r\n 2. Missing metadata - it appears we are missing some metadata for epoll, especially for JNI. Netty also appears to have many ways to fetch the native transport library, out of which one is to fetch it as a Java resource and extract it under `/tmp` at runtime. That approach seems to work fine, so I've also added a `resource-config.json` that includes this library if epoll is used.\r\n\r\nIf it's okay, I'll create a PR that adds the flags and the metadata to Netty 4.1\r\n\r\n" ]

[]

"2023-01-26T18:53:30Z"

[ "graal" ]

[]

[ "transport-classes-epoll/src/main/resources/META-INF/native-image/io.netty/netty-transport-classes-epoll/jni-config.json", "transport-classes-epoll/src/main/resources/META-INF/native-image/io.netty/netty-transport-classes-epoll/native-image.properties", "transport-classes-epoll/src/main/resources/META-INF/native-image/io.netty/netty-transport-classes-epoll/reflect-config.json", "transport-classes-epoll/src/main/resources/META-INF/native-image/io.netty/netty-transport-classes-epoll/resource-config.json" ]

[]

"2020-09-28T00:33:12Z"

[ "keyword_pr_to_issue" ]

[ "When can this be released ?", "I think in the next 2-3 weeks. " ]

[ "API Breakage", "API Breakage", "API Breakage", "API Breakage", "API Breakage", "API Breakage", "API Breakage", "I know there is API breakage, but I was wondering if we can still afford it given that this interface is pretty recent.", "No, we can't break applications.", "Hmm, then I need to figure out another way to add this debug information. BTW the original change of adding the validation itself is a breaking change. ", "No, it isn't. The public API was preserved.", "The public API is preserved but the actual runtime behavior was changed. Prior to the validation change the header values were leniently accepted but now there is a stricter validation, which IMO is a good thing but it broke one of our applications which was adding an empty space in a header.\r\n\r\nNow an error is thrown but it isn't clear which header is the one which has a whitespace.", "I agree the runtime behaviour changed but I still think it was an \"acceptable\" change due the security impact of not doing so", "I didn't look to closely but maybe what could be done is to \"catch and rethrow\" when this happens. The caller would then add the extra details ", "@normanmaurer That was what I initially thought of doing, but it seemed hacky. I have another approach of creating another interface which is also API compatible. I will update the PR, if you like it that's fine. Otherwise will fallback to catch, change message and rethrow.", " ~~@normanmaurer This is what I am thinking of doing,~~\r\n\r\n~~1. Create a new interface `HeaderValueValidator<K, V>`, the benefit of this interface would be if we want to do a contextual validations of headers on the basis of what the header represents it would be possible via this interface. I feel validation of header values without having context of their name works now but seems inflexible to me.~~\r\n~~2. Mark `ValueValidator` as deprecated but still keep it around because we can't break public APIs. During actual validation we check which of `HeaderValueValidator` or `ValueValidator` has been provided and based on that call the appropriate validator. If a user is currently extending `DefaultHeaders` their implementations would still pass `ValueValidator` and continue to function as is, new users would pass a `HeaderValueValidator`.~~\r\n\r\n~~I could have made `HeaderValueValidator` extend `ValueValidator` but that pollutes the new interface with an unnecessary `validate(V value)`.~~\r\n\r\n~~WDYT ?~~\r\n\r\nI ended up catching and throwing, because we do it many places in the codebase.", "Throwing a stack trace is nothing less than making applications vulnerable to Denial-of-Service (DoS). This can be abused easily.\r\n\r\nConsider making it stackless.", "@normanmaurer WDYT?", "I am not inclined to make this exception stackless, at the very most I can maybe remove the cause, but making it stackless severely hampers debugging ability and is not even the standard norm in the Netty codebase.\r\n\r\n> Throwing a stack trace is nothing less than making applications vulnerable to Denial-of-Service (DoS)\r\n\r\nIf someone is trying to DoS an application by sending invalid headers, I don't think generating an exception stacktrace would be the bottleneck. Logging the stack trace might be but there are other mitigations for that in the logger configurations which applications need to use anyways.", "No need to concat the message. It will be part of the wrapped cause anyway. \r\n```suggestion\r\n throw new IllegalArgumentException(\"Validation failed for header '\" + name + \"'\", e);\r\n```", "I agree with @adwsingh here... I think its ok to just wrap.", "I will do profiling in the future for this and see how things behave in stack trace vs stackless." ]

"2023-02-10T09:22:22Z"

[]

[ "codec/src/main/java/io/netty/handler/codec/DefaultHeaders.java" ]

[ "codec/src/main/java/io/netty/handler/codec/DefaultHeaders.java" ]

[ "codec-http/src/test/java/io/netty/handler/codec/http/DefaultHttpHeadersTest.java" ]

"2023-02-10T09:12:50Z"

[ "keyword_issue_to_pr" ]

[ ">Netty version Used : 1.0.24\r\n\r\n\r\n@adhanoti Is this version, Reactor Netty version?\r\n\r\nIf yes then may be we would better discuss this in Reactor Netty issue tracker", "Thanks. Yes it is reactor netty version.\n\nDo i need to create separate issue?\n\nRegards,\nAshish\n\nOn Tue, Feb 21, 2023 at 12:07 PM Violeta Georgieva ***@***.***>\nwrote:\n\n> Netty version Used : 1.0.24\n>\n> @adhanoti <https://github.com/adhanoti> Is this version, Reactor Netty\n> version?\n>\n> If yes then may be we would better discuss this in Reactor Netty issue\n> tracker\n>\n> —\n> Reply to this email directly, view it on GitHub\n> <https://github.com/netty/netty/issues/13235#issuecomment-1437929581>, or\n> unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AK5F3SMW3DUUTUI7RB3HXW3WYRPCLANCNFSM6AAAAAAVCUIW2I>\n> .\n> You are receiving this because you were mentioned.Message ID:\n> ***@***.***>\n>\n", "yes please", "RFC non complaince : RFC 7540 \"Hypertext Transfer Protocol Version 2\n(HTTP/2)\" https://datatracker.ietf.org/doc/html/rfc7540 · Issue #2702 ·\nreactor/reactor-netty (github.com)\n<https://github.com/reactor/reactor-netty/issues/2702>\n\nTicket Created.\n\nLet me know if you need wireshark for same. In ticket i am not able to\nupload wireshark.\n\nRegards,\nAshish\n\nOn Tue, Feb 21, 2023 at 12:18 PM Violeta Georgieva ***@***.***>\nwrote:\n\n> yes please\n>\n> —\n> Reply to this email directly, view it on GitHub\n> <https://github.com/netty/netty/issues/13235#issuecomment-1437937763>, or\n> unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AK5F3SIQG5XSC3UIIN2YKN3WYRQMBANCNFSM6AAAAAAVCUIW2I>\n> .\n> You are receiving this because you were mentioned.Message ID:\n> ***@***.***>\n>\n", "@normanmaurer @chrisvest @idelpivnitskiy This PR #12975 removes `:path` not empty check. IMO we should keep it. Wdyt?", "hm ... it was checking before for duplicate pseudo-header ... ok ignore my comment above", "@adhanoti The issue with `:path` should be address with #13238\r\n\r\nThe issue with `connection-specific header` should be address with #12975", "@normanmaurer @chrisvest @idelpivnitskiy \r\n\r\nThe issue with `Content-Length` that is not matching the actual data is interesting.\r\nNetty correctly identifies that there is a mismatch and throws a stream error.\r\n\r\n`Http2ConnectionHandler` correctly resets the stream with `PROTOCOL_ERROR`\r\nhttps://github.com/netty/netty/blob/59aa6e635b9996cf21cd946e64353270679adc73/codec-http2/src/main/java/io/netty/handler/codec/http2/Http2ConnectionHandler.java#L744\r\n\r\nHowever when `Http2FrameCodec` and `Http2MultiplexHandler`, the flow is the following:\r\n\r\n`Http2FrameCodec` forwards the stream error\r\nhttps://github.com/netty/netty/blob/59aa6e635b9996cf21cd946e64353270679adc73/codec-http2/src/main/java/io/netty/handler/codec/http2/Http2FrameCodec.java#L568\r\n\r\n`Http2MultiplexHandler` notifies the child channel and closes it.\r\nhttps://github.com/netty/netty/blob/59aa6e635b9996cf21cd946e64353270679adc73/codec-http2/src/main/java/io/netty/handler/codec/http2/Http2MultiplexHandler.java#L274-L278\r\n\r\nWhen the stream channel is closed, it is closed always with `CANCEL`\r\nhttps://github.com/netty/netty/blob/59aa6e635b9996cf21cd946e64353270679adc73/codec-http2/src/main/java/io/netty/handler/codec/http2/AbstractHttp2StreamChannel.java#L681\r\n\r\n I don't see a way to propagate the actual error code :(\r\n", "@violetagg let me have a look ", "This should be fixed by https://github.com/netty/netty/pull/13309" ]

[ "These lines should be moved into the `VALUE_VALIDATOR` on line 86.", "The `VALUE_VALIDATOR` receives only the value. How can we match name `:path` with the value?\r\nhttps://github.com/netty/netty/blob/59aa6e635b9996cf21cd946e64353270679adc73/codec-http2/src/main/java/io/netty/handler/codec/http2/DefaultHttp2Headers.java#L86-L95\r\n\r\nAlso `Http2FrameCodecBuilder ` always builds `DefaultHttp2HeadersDecoder` with value validation disabled.\r\nhttps://github.com/netty/netty/blob/59aa6e635b9996cf21cd946e64353270679adc73/codec-http2/src/main/java/io/netty/handler/codec/http2/Http2FrameCodecBuilder.java#L208-L210", "Thanks for finding this missed case @violetagg.\r\nAfter reading the whole [RFC section 8.3](https://datatracker.ietf.org/doc/html/rfc9113#section-8.3), it sounds like any pseudo-header filed must not be empty, not only the `:path`. The only legit way is to omit a pseudo-header. WDYT?\r\n\r\nPerforming the check inside `DefaultHttp2Headers` seem reasonable, we just need to find a good spot where we have access to both name and value. I don't have any better idea except `validateValue` method as it's right now. ", "We need to protect this validation by a configuration flag to make sure users can disable it.\r\n\r\nTaking into account the RFC requirements to always run this validation (at least by default), the fact that `validateValues == false` by default, I would propose to use `validate` flag that controls header name validation. While in practice the check validates a value, I would say that technically it's a \"validation of a preudo-header name\". The purpose of `validateValues` is quite different. It's disabled by default because it will cause significant perf overhead for most users.", "Agree", "Please check the new commit.", "> Taking into account the RFC requirements to always run this validation (at least by default), the fact that validateValues == false by default, I would propose to use validate flag that controls header name validation.\r\n\r\n@idelpivnitskiy Is this \"always\" requirement any different from checking that byte ranges for values are valid, which we skip by default?", "The biggest differences are:\r\n1. the cost associated with a check\r\n2. how likely is it to hit the problem (messing up with pseudo-headers seems more likely)\r\n3. how it will affect the h2-protocol related flow vs user's business logic (having a bad char in some header won't impact h2 codec or h2->h1 transition)\r\n4. is the header name required to perform the check\r\n\r\nThis case seems to me close to `TE` header validation we have in `HpackDecoder`." ]

"2023-02-22T08:54:46Z"

[]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/DefaultHttp2Headers.java" ]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/DefaultHttp2Headers.java" ]

[ "codec-http2/src/test/java/io/netty/handler/codec/http2/HpackDecoderTest.java", "codec-http2/src/test/java/io/netty/handler/codec/http2/HttpConversionUtilTest.java" ]

"2023-02-21T06:32:31Z"

[ "keyword_pr_to_issue" ]

[ "Extend the `HttpContentDecompressor` and override the `newContentDecoder` for Snappy and delegate the rest to `super#newContentDecoder`.", "> Extend the `HttpContentDecompressor` and override the `newContentDecoder` for Snappy and delegate the rest to `super#newContentDecoder`.\r\n\r\nThanks for the guidance @hyperxpro, I assume this means Pr's are welcome ? I'll be working on it then ^^", "They're! ^^" ]

[ "```suggestion\r\n if (SNAPPY.contentEqualsIgnoreCase(contentEncoding)) {\r\n```", "```suggestion\r\n \r\n```", "```suggestion\r\n -22, 104, 101, 108, 108, 111, 44, 32, 119, 111, 114, 108, 100\r\n```", "```suggestion\r\n```", "```suggestion\r\n```", "```suggestion\r\n return new EmbeddedChannel(ctx.channel().id(), ctx.channel().metadata().hasDisconnect(),\r\n```" ]

"2023-03-30T19:36:44Z"

[]

[ "codec-http/src/main/java/io/netty/handler/codec/http/HttpContentDecompressor.java", "codec-http/src/main/java/io/netty/handler/codec/http/HttpHeaderValues.java" ]

[ "codec-http/src/main/java/io/netty/handler/codec/http/HttpContentDecompressor.java", "codec-http/src/main/java/io/netty/handler/codec/http/HttpHeaderValues.java" ]

[ "codec-http/src/test/java/io/netty/handler/codec/http/HttpContentDecoderTest.java" ]

"2023-03-29T13:52:42Z"

[ "keyword_pr_to_issue" ]

[]

[ "should this be `execute(task, false)`?", "check `SingleThreadEventExecutor::lazyExecute` overridden method:\r\n\r\n```java\r\n @Override\r\n public void lazyExecute(Runnable task) {\r\n lazyExecute0(task);\r\n }\r\n\r\n private void lazyExecute0(@Schedule Runnable task) {\r\n execute(ObjectUtil.checkNotNull(task, \"task\"), false);\r\n }\r\n```", "```suggestion\r\n * @deprecated override {@link SingleThreadEventExecutor#wakesUpForTask} to re-create this behaviour\r\n```", "```suggestion\r\n * @deprecated override {@link SingleThreadEventExecutor#wakesUpForTask} to re-create this behaviour\r\n```" ]

"2023-04-17T09:51:25Z"

[ "cleanup" ]

[ "common/src/main/java/io/netty/util/concurrent/AbstractEventExecutor.java", "common/src/main/java/io/netty/util/concurrent/SingleThreadEventExecutor.java", "transport/src/main/java/io/netty/channel/SingleThreadEventLoop.java" ]

[ "common/src/main/java/io/netty/util/concurrent/AbstractEventExecutor.java", "common/src/main/java/io/netty/util/concurrent/SingleThreadEventExecutor.java", "transport/src/main/java/io/netty/channel/SingleThreadEventLoop.java" ]

[ "common/src/test/java/io/netty/util/concurrent/SingleThreadEventExecutorTest.java" ]

"2023-04-17T09:45:03Z"

[ "keyword_issue_to_pr", "keyword_pr_to_issue" ]

[ "This is either because you don't have an `HttpRequestDecoder` or a `HttpServerCodec` in your pipeline, or something else in your pipeline dropped the handshake response in the write path.", "If the pipeline checks fail, how can i release the handshake response.", "> If the pipeline checks fail, how can i release the handshake response.\r\n\r\nYou can't. It's a bug in Netty that we'll fix with #13338" ]

[]

"2023-04-17T14:42:48Z"

[]

[ "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketServerHandshaker.java" ]

[ "codec-http/src/main/java/io/netty/handler/codec/http/websocketx/WebSocketServerHandshaker.java" ]

[]

"2023-04-17T14:02:11Z"

[ "keyword_pr_to_issue" ]

[ "You can specify a CNAME cache by calling `DnsNameResolverBuilder.cnameCache(DnsCnameCache)`, specifying `new DefaultDnsCnameCache()` with the desired min/max TTL values. Would this work for you?", "Thanks for the response, Trustin. This would work, but it feels like a workaround as opposed to allowing ttl values to be set for the other caches via the ttl fields", "@trustin I think it is a bit inconvenient to have configured either all types of caches or to leave them all default based on the TTLs. I prepared one change can you take a look at this #13386?" ]

[]

"2023-05-17T06:17:45Z"

[]

[ "resolver-dns/src/main/java/io/netty/resolver/dns/DefaultAuthoritativeDnsServerCache.java", "resolver-dns/src/main/java/io/netty/resolver/dns/DefaultDnsCnameCache.java", "resolver-dns/src/main/java/io/netty/resolver/dns/DnsNameResolverBuilder.java" ]

[ "resolver-dns/src/main/java/io/netty/resolver/dns/DefaultAuthoritativeDnsServerCache.java", "resolver-dns/src/main/java/io/netty/resolver/dns/DefaultDnsCnameCache.java", "resolver-dns/src/main/java/io/netty/resolver/dns/DnsNameResolverBuilder.java" ]

[ "resolver-dns/src/test/java/io/netty/resolver/dns/DnsNameResolverBuilderTest.java" ]

"2023-05-16T10:41:26Z"

[ "keyword_pr_to_issue" ]

[ "thanks @violetagg " ]

[]

"2023-06-02T13:06:51Z"

[]

[ "transport-classes-epoll/src/main/java/io/netty/channel/epoll/EpollDomainSocketChannel.java", "transport-classes-kqueue/src/main/java/io/netty/channel/kqueue/KQueueDomainSocketChannel.java" ]

[ "transport-classes-epoll/src/main/java/io/netty/channel/epoll/EpollDomainSocketChannel.java", "transport-classes-kqueue/src/main/java/io/netty/channel/kqueue/KQueueDomainSocketChannel.java" ]

[ "testsuite/src/main/java/io/netty/testsuite/transport/socket/SocketAddressesTest.java" ]

"2023-06-01T12:41:04Z"

[ "keyword_pr_to_issue" ]

[ "Unfortunately, those pictures are not worth even a single word. That's too less information to conclude anything. \r\n\r\nWhat kind of workload is it? How many connections? Are you running blocking operations on EventLoop?", "> Unfortunately, those pictures are not worth even a single word. That's too less information to conclude anything.\r\n> \r\n> What kind of workload is it? How many connections? Are you running blocking operations on EventLoop?\r\n\r\n5 connections only，no blocking operations on EventLoop", "the cause i already found\r\n\r\n\"nioEventLoopGroup-3-1\" Id=14 cpuUsage=79.15% deltaTime=171ms time=9015ms RUNNABLE\r\n at io.netty.handler.codec.DelimiterBasedFrameDecoder.indexOf(DelimiterBasedFrameDecoder.java:314)\r\n at io.netty.handler.codec.DelimiterBasedFrameDecoder.decode(DelimiterBasedFrameDecoder.java:236)\r\n at io.netty.handler.codec.DelimiterBasedFrameDecoder.decode(DelimiterBasedFrameDecoder.java:214)\r\n at io.netty.handler.codec.ByteToMessageDecoder.decodeRemovalReentryProtection(ByteToMessageDecoder.java:501)\r\n at io.netty.handler.codec.ByteToMessageDecoder.callDecode(ByteToMessageDecoder.java:440)\r\n at io.netty.handler.codec.ByteToMessageDecoder.channelRead(ByteToMessageDecoder.java:276)\r\n at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:379)\r\n at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:365)\r\n at io.netty.channel.AbstractChannelHandlerContext.fireChannelRead(AbstractChannelHandlerContext.java:357)\r\n at io.netty.handler.timeout.IdleStateHandler.channelRead(IdleStateHandler.java:286)\r\n at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:379)\r\n at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:365)\r\n at io.netty.channel.AbstractChannelHandlerContext.fireChannelRead(AbstractChannelHandlerContext.java:357)\r\n at io.netty.channel.DefaultChannelPipeline$HeadContext.channelRead(DefaultChannelPipeline.java:1410)\r\n at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:379)\r\n at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:365)\r\n at io.netty.channel.DefaultChannelPipeline.fireChannelRead(DefaultChannelPipeline.java:919)\r\n at io.netty.channel.nio.AbstractNioByteChannel$NioByteUnsafe.read(AbstractNioByteChannel.java:163)\r\n at io.netty.channel.nio.NioEventLoop.processSelectedKey(NioEventLoop.java:714)\r\n at io.netty.channel.nio.NioEventLoop.processSelectedKeysOptimized(NioEventLoop.java:650)\r\n at io.netty.channel.nio.NioEventLoop.processSelectedKeys(NioEventLoop.java:576)\r\n at io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:493)\r\n at io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989)\r\n at io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)\r\n at io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)\r\n at java.lang.Thread.run(Thread.java:748)\r\n", "I put up a PR to use a faster search algorithm, but `DelimiterBasedFrameDecoder` is _always_ going to be CPU intensive because it has to search for the delimiters no matter what." ]

[]

"2023-06-15T17:25:21Z"

[]

[ "codec/src/main/java/io/netty/handler/codec/DelimiterBasedFrameDecoder.java" ]

[ "codec/src/main/java/io/netty/handler/codec/DelimiterBasedFrameDecoder.java" ]

[]

"2023-03-31T03:22:05Z"

[ "keyword_issue_to_pr" ]

[ "You can use `ByteBufUtil.indexOf` to implement a faster `DelimiterBasedFrameDecoder.indexOf`, if you're up for doing a PR. There's some finicky behavior that needs to be preserved, though.", "your delimiter is a single char one @zzy444626905 ? if yes, it shouldn't be that cost if you use a recent enough version of Netty (when https://github.com/netty/netty/pull/10737 was in).", "This was fixed by #13451" ]

[]

"2023-06-15T17:25:21Z"

[]

[ "codec/src/main/java/io/netty/handler/codec/DelimiterBasedFrameDecoder.java" ]

[ "codec/src/main/java/io/netty/handler/codec/DelimiterBasedFrameDecoder.java" ]

[]

"2023-04-07T00:58:13Z"

[ "keyword_pr_to_issue" ]

[ "Good find, we'll take a look." ]

[ "should we use `array[i].equals(o)` ?", "imho this would be \"safer\". ", "Changed it to `Objects.equals` in line with the API spec recommendation – it handles `null` values correctly.", "Oh no, that's not available in the older Java versions. Fine." ]

"2023-06-15T18:49:43Z"

[]

[ "transport/src/main/java/io/netty/channel/nio/SelectedSelectionKeySet.java" ]

[ "transport/src/main/java/io/netty/channel/nio/SelectedSelectionKeySet.java" ]

[ "transport/src/test/java/io/netty/channel/nio/SelectedSelectionKeySetTest.java" ]

"2023-04-13T14:07:54Z"

[ "keyword_pr_to_issue" ]

[]

[ "can you add a comment to explain where these values come from ?", "Does this mean Java < 8 are not supported?", "@hyperxpro good point... maybe we should just add a version check as well. Let me do this " ]

"2023-08-09T01:22:31Z"

[]

[ "handler/src/main/java/io/netty/handler/ssl/SslContext.java" ]

[ "handler/src/main/java/io/netty/handler/ssl/SslContext.java" ]

[ "handler/src/test/java/io/netty/handler/ssl/SslContextTest.java", "handler/src/test/resources/io/netty/handler/ssl/generate-certs.sh", "handler/src/test/resources/io/netty/handler/ssl/rsa_pbes2_enc_pkcs8.key" ]

"2023-08-08T03:24:15Z"

[ "keyword_pr_to_issue" ]

[ "I have reproduced it and found that the decoded value is **not equal** to the original value whenever the `enableHuffmanEncoding` is true or false. And I found a comment in `HpackEncoder.java`: Only **ASCII** is allowed in http2 headers(https://tools.ietf.org/html/rfc7540#section-8.1.2), so if there exists unexpected characters, unexpected errors may happen. \r\n1. support charset ISO-8859-1 \r\n<img width=\"804\" alt=\"image\" src=\"https://github.com/netty/netty/assets/23135058/56b43e66-f645-4a34-bf2b-9d64fab5c421\">\r\n2. support characters with one byte \r\n<img width=\"517\" alt=\"image\" src=\"https://github.com/netty/netty/assets/23135058/8b4f6d0d-6574-4804-bb31-345278cabf39\">\r\n\r\nIn your case, your headers contains invalid character, which is not in ISO-8859-1 charset or more than one byte.\r\n\r\n\r\n", "Thanks for looking into this @tmyVvit! You're totally right that you're not supposed to use non-US-ASCII chars in the the header values but the interesting bit is that, even if you do, something in the decoder/encoder chain normally does sanitization of the string and replaces characters that are outside of allowed range with `?`. Here is a quick REPL demo:\r\n\r\n```scala\r\n@ val buf = Unpooled.buffer(10000) \r\nbuf: io.netty.buffer.ByteBuf = UnpooledByteBufAllocator$InstrumentedUnpooledUnsafeHeapByteBuf(ridx: 0, widx: 0, cap: 10000)\r\n\r\n@ headers.set(\"some-key\", \"hello 우리는에서문제를발견했습니다world\") \r\nres14: io.netty.handler.codec.http2.Http2Headers = DefaultHttp2Headers[some-key: hello 우리는에서문제를발견했습니다world]\r\n\r\n@ encoder.encodeHeaders(1, headers, buf) \r\n\r\n\r\n@ val decodedHeader = decoder.decodeHeaders(1, buf) \r\ndecodedHeader: io.netty.handler.codec.http2.Http2Headers = DefaultHttp2Headers[some-key: hello ??????????????world, some-key: hello ??????????????world]\r\n\r\n@ val decodedHeaderValue = decodedHeader.get(\"some-key\") \r\ndecodedHeaderValue: CharSequence = hello ??????????????world\r\n```\r\n\r\nSomething in that chain is obviously not doing the sanitization right when a) Huffman encoding is ON and b) the string contains certain UTF-8 chars in the certain positions.", "@tmyVvit is right: it's the `一` char at index 195 that is turning onto a `0x0`. The only reason it doesn't turn into a `0x0` when not using huffman encoding is because of the `.getBytes(IO_8859_1)` call instead of doing what huffman encoding does and doing a simple truncation via `string.charAt(i) & 0xFF`. Despite not being a `0x0`, it's a certainty that the resulting value is still not what you wanted.\r\n\r\n@vkostyukov, can you provide a complete example? I'm not getting the same results as you. It may depend on how the `DefaultHttp2Headers` is instantiated. Specifically, when I run\r\n```\r\nHttp2Headers headers = new DefaultHttp2Headers(true, true, 16);\r\nheaders.set(\"some-key\", \"hello 우리는에서문제를발견했습니다world\");\r\n\r\nDefaultHttp2HeadersEncoder encoder = new DefaultHttp2HeadersEncoder();\r\nByteBuf buf = Unpooled.buffer(10000);\r\nencoder.encodeHeaders(1, headers, buf);\r\n\r\nHttp2HeadersDecoder decoder = new DefaultHttp2HeadersDecoder();\r\nHttp2Headers decodedHeader = decoder.decodeHeaders(1, buf);\r\nthrow new Exception(\"Value:\" + decodedHeader.get(\"some-key\"));\r\n// results in java.lang.Exception: Value:hello °¬”Ð\u001c8\u001c|\u001c¬ˆµÈäworld\r\n```\r\n\r\nWhat you all but certainly want is eager validation when the header is added. Unfortunately, and it doesn't seem to be what you're seeing with the '?'s, but it does look like the default validation is broken [here](https://github.com/netty/netty/blob/4.1/codec-http/src/main/java/io/netty/handler/codec/http/HttpHeaderValidationUtil.java#L143), specifically chars wider than 1 byte will all be considered 'ok' which is wrong.", "re: https://github.com/netty/netty/issues/13540#issuecomment-1671799091\r\n\r\n@tmyVvit @bryce-anderson what I find a bit confusing in the comment referenced above - https://datatracker.ietf.org/doc/html/rfc7540#section-8.1.2 says\r\n\r\n> header **field names** are strings of ASCII\r\n\r\nwhich covers header names, but it doesn't say anything about header **field values**, which sidetracked us into thinking that field values with non-US-ASCII chars are not illegal, and there's a bug in the `HpackEncoder.java`. would it make sense to update the comment in the `HpackEncoder.java` to clarify that and add more context? after re-reading the spec - think it's covered by https://datatracker.ietf.org/doc/html/rfc7540#section-10.3 which points to the \"field-content\" ABNF rule in https://datatracker.ietf.org/doc/html/rfc7230#section-3.2, and https://datatracker.ietf.org/doc/html/rfc7230#section-3.2.4 which says:\r\n\r\n> Newly defined header fields SHOULD limit their field values to\r\n US-ASCII octets. A recipient SHOULD treat other octets in field\r\n content (obs-text) as opaque data\r\n\r\nI think some clarification would be helpful for posterity.", "It actually looks like [this code path](https://github.com/netty/netty/blob/4.1/codec-http2/src/main/java/io/netty/handler/codec/http2/HpackEncoder.java#L308C17-L310C71) (when Huffman encoding is bypassed) works in our favor, if you will.\r\n\r\n```java\r\n// Only ASCII is allowed in http2 headers, so its fine to use this.\r\n// https://tools.ietf.org/html/rfc7540#section-8.1.2\r\nout.writeCharSequence(string, CharsetUtil.ISO_8859_1);\r\n```\r\n\r\nThis end up calling into `ByteBufUtil.writeAscii` [here](https://github.com/netty/netty/blob/4.1/buffer/src/main/java/io/netty/buffer/AbstractByteBuf.java#L717), which performs sanitization with the `?` character [here](https://github.com/netty/netty/blob/4.1/buffer/src/main/java/io/netty/buffer/ByteBufUtil.java#L1222-L1229). Relevant function is `AsciiString.c2b`\r\n\r\n```java\r\npublic static byte c2b(char c) {\r\n return (byte) ((c > MAX_CHAR_VALUE) ? '?' : c);\r\n}\r\n```\r\n\r\nPerhaps we should re-use the same conversion function in `HpackHuffmanEncoder` [here](https://github.com/netty/netty/blob/4.1/codec-http2/src/main/java/io/netty/handler/codec/http2/HpackHuffmanEncoder.java#L90) (instead of `data.charAt(i) & 0xFF`) to make sanitization behavior consistent whether or not Huffman encoding is engaged. ", "> which performs sanitization with the ? character [here](https://github.com/netty/netty/blob/4.1/buffer/src/main/java/io/netty/buffer/ByteBufUtil.java#L1222-L1229)\r\n\r\n+1 to perform sanitization over making non-US-ASCII characters [prohibited](https://github.com/netty/netty/blob/7127e6074c2f4b965305738d7756d196729b493e/codec-http2/src/main/java/io/netty/handler/codec/http2/DefaultHttp2Headers.java#L91-L92) - IMO that may break Netty users who happened to put non-US-ASCII chars in the header values without hitting this issue. if there's insufficient test coverage - making chars prohibited will start failing their requests at runtime causing outages. sanitizing them seems less damaging. as a precedent - Jetty sanitizes non-US-ASCII chars by replacing them with spaces ([ref](https://github.com/eclipse/jetty.project/blob/ad0b5cfa1533e83814c74063b6c2e17b099e5796/jetty-http/src/main/java/org/eclipse/jetty/http/HttpGenerator.java#L894-L906)). given that spec uses \"SHOULD\" when suggesting to restricting chars in field values to US-ASCII octets - I think it may be a reasonable option to sanitize and limit prohibited chars validations to the chars which are prohibited for security reasons https://datatracker.ietf.org/doc/html/rfc7540#section-10.3.", "> Perhaps we should re-use the same conversion function in HpackHuffmanEncoder [here](https://github.com/netty/netty/blob/4.1/codec-http2/src/main/java/io/netty/handler/codec/http2/HpackHuffmanEncoder.java#L90) (instead of data.charAt(i) & 0xFF) to make sanitization behavior consistent whether or not Huffman encoding is engaged.\r\n\r\nI think that makes sense, [here](https://github.com/netty/netty/blob/4.1/codec-http2/src/main/java/io/netty/handler/codec/http2/HpackHuffmanEncoder.java#L136) would also need to be fixed. Would one of you be interested in putting up a PR?\r\n\r\n> +1 to perform sanitization over making non-US-ASCII characters [prohibited](https://github.com/netty/netty/blob/7127e6074c2f4b965305738d7756d196729b493e/codec-http2/src/main/java/io/netty/handler/codec/http2/DefaultHttp2Headers.java#L91-L92) - IMO that may break Netty users who happened to put non-US-ASCII chars in the header values without hitting this issue.\r\n\r\nThese are not mutually exclusive: the validation is optional (and disabled by default) and happens on header insertion which is a nice feature since it lets you figure out what code is adding illegal headers so you can fix you app. If we don't fix the validation the feature will remain forever broken.", ">I think that makes sense, [here](https://github.com/netty/netty/blob/4.1/codec-http2/src/main/java/io/netty/handler/codec/http2/HpackHuffmanEncoder.java#L136) would also need to be fixed. Would one of you be interested in putting up a PR?\r\n\r\nI will give it a shot (putting up a PR)! @bryce-anderson @vkostyukov " ]

[ "We cannot directly use `AsciiString.c2b(...)` here because this function does 2 things:\r\n\r\n1. Sanitize the given input char.\r\n2. Cast the sanitized char to a `byte`.\r\n\r\nThen we cast the returned `byte` to `int b`. Overall, we are doing `sanitizedChar.toByte.toInt` and it could yield negative value which cannot be used as array index. For example `128.toByte.toInt == -128`.\r\n\r\nWe can also consider doing:\r\n```java\r\nint b = (char) AsciiString.c2b(data.charAt(i));\r\n```\r\nor \r\n```java\r\nint b = AsciiString.b2c(AsciiString.c2b(data.charAt(i)));\r\n```\r\n\r\nBut it does not seem to be expressive and clear on our intention and it seems awkward IMO.\r\n\r\nSo, I decided to create this `sanitizeChar` local function.", "Updated it to `AsciiString.c2b(data.charAt(i)) & 0xFF` as `0xFF` can produce unsigned int which is what we want here.", "`it's`?", "My understanding of the small value is that it wont sanitize anything since it will never yield a value greater than 255, making it not a very interesting test. The 300 case should cover what we're really interested in wrt ensuring we don't get wrap-around values of null etc.\r\nA second type of test is that would be valuable is that we should get the same result whether we perform the Huffman encoding or not. What if we just test the size 300 and test it twice: once with the encoded set to 512 (the default I believe) and once super short such to force Huffman encoding, then make sure they both yield the same results?", "Thanks Bryce for the suggestion. I think it's a good idea and I have updated this PR with this idea implemented. PTAL.", "looks good." ]

"2023-08-13T06:41:56Z"

[]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/HpackEncoder.java", "codec-http2/src/main/java/io/netty/handler/codec/http2/HpackHuffmanEncoder.java" ]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/HpackEncoder.java", "codec-http2/src/main/java/io/netty/handler/codec/http2/HpackHuffmanEncoder.java" ]

[ "codec-http2/src/test/java/io/netty/handler/codec/http2/HpackEncoderTest.java", "codec-http2/src/test/java/io/netty/handler/codec/http2/HpackHuffmanTest.java" ]

"2023-08-09T07:05:07Z"

[ "keyword_issue_to_pr" ]

[ "@fax4ever the `java.lang.ExceptionInInitializerError`should have a cause set... can you add this one as well ?", "> @fax4ever the `java.lang.ExceptionInInitializerError`should have a cause set... can you add this one as well ?\r\n\r\nDebugging the following seems to be the cause:\r\n\r\n``` bash\r\njava.lang.IllegalArgumentException: cannot use an unresolved DNS server address: fe80::21c:42ff:fe00:18%enp0s5/<unresolved>:5\r\n\r\nio.netty.resolver.dns.DnsServerAddresses.sanitize(DnsServerAddresses.java:175)\r\nio.netty.resolver.dns.DnsServerAddresses.sequential(DnsServerAddresses.java:67)\r\nio.netty.resolver.dns.UnixResolverDnsServerAddressStreamProvider.putIfAbsent(UnixResolverDnsServerAddressStreamProvider.java:261)\r\nio.netty.resolver.dns.UnixResolverDnsServerAddressStreamProvider.parse(UnixResolverDnsServerAddressStreamProvider.java:241)\r\nio.netty.resolver.dns.UnixResolverDnsServerAddressStreamProvider.<init>(UnixResolverDnsServerAddressStreamProvider.java:98)\r\nio.netty.resolver.dns.UnixResolverDnsServerAddressStreamProvider.<init>(UnixResolverDnsServerAddressStreamProvider.java:133)\r\nio.netty.resolver.dns.UnixResolverDnsServerAddressStreamProvider.parseSilently(UnixResolverDnsServerAddressStreamProvider.java:72)\r\nio.netty.resolver.dns.DnsServerAddressStreamProviders$DefaultProviderHolder$1.provider(DnsServerAddressStreamProviders.java:151)\r\nio.netty.resolver.dns.DnsServerAddressStreamProviders$DefaultProviderHolder$1.<init>(DnsServerAddressStreamProviders.java:130)\r\nio.netty.resolver.dns.DnsServerAddressStreamProviders$DefaultProviderHolder.<clinit>(DnsServerAddressStreamProviders.java:128)\r\nio.netty.resolver.dns.DnsServerAddressStreamProviders.unixDefault(DnsServerAddressStreamProviders.java:117)\r\nio.netty.resolver.dns.DnsServerAddressStreamProviders.platformDefault(DnsServerAddressStreamProviders.java:113)\r\nio.netty.resolver.dns.DnsNameResolverBuilder.<init>(DnsNameResolverBuilder.java:66)\r\n```\r\n\r\nThanks for the support!", "Maybe the address `fe80::21c:42ff:fe00:18` cannot be resolved?", "@fax4ever should be fixed by https://github.com/netty/netty/pull/13614", "> @fax4ever should be fixed by #13614\r\n\r\nthank you @normanmaurer for the very quick fix!" ]

[ "We can reuse the `addr` variable here, instead of recreating the `InetSocketAddress` object." ]

"2023-09-15T10:33:02Z"

[]

[ "resolver-dns/src/main/java/io/netty/resolver/dns/UnixResolverDnsServerAddressStreamProvider.java" ]

[ "resolver-dns/src/main/java/io/netty/resolver/dns/UnixResolverDnsServerAddressStreamProvider.java" ]

[ "resolver-dns/src/test/java/io/netty/resolver/dns/UnixResolverDnsServerAddressStreamProviderTest.java" ]

"2023-09-15T08:50:37Z"

[ "keyword_pr_to_issue" ]

[ "Via local builds I was able to isolate the problematic commit: https://github.com/netty/netty/commit/dc69a943c3682fec16845828cb6d8f0554fad61f\r\n\r\nStill not sure what this breaks - we're using standard PEM format like:\r\n\r\n```\r\n-----BEGIN CERTIFICATE-----\r\n<a bunch of base 64>\r\n-----END CERTIFICATE-----\r\n```\r\n\r\nwith none of the newlines this commit seems to address.", "@andrewparmet I guess you can't share the pem file ?", "Unfortunately not. I've been able to narrow it down a bit:\r\n\r\n- Parsing does work on the JVM - this is only an Android issue\r\n- The text returned by [`m.group(0)` here](https://github.com/netty/netty/blob/aa59245955f5121c4230d46c41a4b7f55e0c9ba7/handler/src/main/java/io/netty/handler/ssl/PemReader.java#L92) is simply `BEGIN` - which is not right at all. I'd have a hard time believing it, but is there a difference between the regex implementations on the JVM and Android?", "hmm... sounds like an android bug then... ", "Sure does. Here's a minimal reproducer:\r\n\r\n```kotlin\r\nprivate val CERT_HEADER = Pattern.compile(\"-+BEGIN\\\\s[^-\\\\r\\\\n]*CERTIFICATE[^-\\\\r\\\\n]*-+(?:\\\\s|\\\\r|\\\\n)+\")\r\nprivate val BODY = Pattern.compile(\"[a-z0-9+/=][a-z0-9+/=\\\\r\\\\n]*\", Pattern.CASE_INSENSITIVE)\r\n\r\nclass RegexTest {\r\n @Test\r\n fun regexTest() {\r\n val content =\r\n \"\"\"\r\n |-----BEGIN CERTIFICATE-----\r\n |abcdef\r\n |-----END CERTIFICATE-----\r\n \"\"\".trimMargin()\r\n\r\n val m = CERT_HEADER.matcher(content)\r\n assertTrue(m.find(0))\r\n m.usePattern(BODY)\r\n assertTrue(m.find())\r\n val body = m.group(0)\r\n assertEquals(\"abcdef\\n\", body)\r\n }\r\n}\r\n```\r\n\r\nThis test passes on the JVM (OpenJDK Runtime Environment Corretto-11.0.18.10.1 (build 11.0.18+10-LTS)) and fails on Android 34.", "Filed: https://issuetracker.google.com/issues/293206296 and we'll see what comes back.", "@andrewparmet thanks... keep us posted ", "Hi there,\r\nI have the same issue here. \r\n\r\nI did some research and found that this is probably not an Android issue.\r\nPemReader class uses regex to parse the certificates from the InputStream, but the thing is that regex works differently in JVM and Android Runtime.\r\nhttps://github.com/netty/netty/blob/285ba74f899866ddf3e23a6770c954e350245bd7/handler/src/main/java/io/netty/handler/ssl/PemReader.java#L84C1-L90C14\r\nPemReader expects that after matching the header the matcher moves the cursor to a position after the header and it works this way in JVM but it doesn't in Android and the certificate body is matched starting from the beginning of the string as just \"BEGIN\". And this is why the conversion to an X509 certificate instance fails.\r\n\r\nThis can be fixed by calling the find() method with the next position after the header as a parameter, or by using the BouncyCastle PEMParser as it's done for the private key.\r\n", "@fishb1 if you know how to fix it please open a PR and we will review it ", "You first say:\r\n\r\n> this is probably not an Android issue\r\n\r\nThen you say:\r\n\r\n> regex works differently in JVM and Android Runtime\r\n\r\nTo me this seems like the definition of \"Android issue\" but if you have a workaround that fixes the implementation on Android while not breaking it on the JVM, then to me that seems like only mild debt (when paired with comment to undo the hack and a link to the Android bug that I filed and Google accepted).\r\n\r\nIn any case we've migrated to using binary certificate files instead of text so we're not held hostage by this bug anymore.", "> To me this seems like the definition of \"Android issue\" but if you have a workaround that fixes the implementation on Android while not breaking it on the JVM, then to me that seems like only mild debt (when paired with comment to undo the hack and a link to the Android bug that I filed and Google accepted).\r\n> \r\n\r\nAhhh, I see now! I expected that usePatten() method reset the latest match position but I read the doc for the method again and it said that the position should not be affected. \r\n\r\nThanks for submitting this! \r\n\r\nBut I think a fix needs to be done anyway to make Netty work in older SDKs, even if Google fixes the regex issue in future versions." ]

[]

"2023-11-10T09:00:53Z"

[]

[ "handler/src/main/java/io/netty/handler/ssl/PemReader.java" ]

[ "handler/src/main/java/io/netty/handler/ssl/PemReader.java" ]

[]

"2023-07-26T02:51:59Z"

[ "keyword_pr_to_issue" ]

[ "I've verified via https://github.com/franz1981/java-puzzles/commit/e083daa22878511c475135b5863b861471e617a6 that thanks to https://github.com/openjdk/jdk/pull/14375 the type check performed at https://github.com/netty/netty/blob/5155482afc57907a1414994ef5aca79b61ac6999/transport/src/main/java/io/netty/channel/ChannelOutboundBuffer.java#L255 can add a bimorphic guard with at best 2 concrete types before hitting the O(n) slow path type check, ,making this issue less impactful, but still showing that knowing the arity of the type checks receivers and trying reducing it via type peeling off (eg instanceof vs a known hot path concrete type) can help the JIT to have enough \"budget\" for the remaining ones, and make the JIT optimization even more effective." ]

[]

"2023-12-21T14:06:40Z"

[ "improvement" ]

[ "transport/src/main/java/io/netty/channel/AbstractCoalescingBufferQueue.java" ]

[ "transport/src/main/java/io/netty/channel/AbstractCoalescingBufferQueue.java" ]

[]

"2023-12-21T09:13:44Z"

[ "keyword_pr_to_issue" ]

[ "We are also facing the same issue.", "We are also facing the same issue" ]

[]

"2024-01-07T17:58:34Z"

[]

[ "codec-http/src/main/java/io/netty/handler/codec/http/cookie/ClientCookieDecoder.java", "codec-http/src/main/java/io/netty/handler/codec/http/cookie/CookieHeaderNames.java", "codec-http/src/main/java/io/netty/handler/codec/http/cookie/DefaultCookie.java", "codec-http/src/main/java/io/netty/handler/codec/http/cookie/ServerCookieEncoder.java" ]

[ "codec-http/src/main/java/io/netty/handler/codec/http/cookie/ClientCookieDecoder.java", "codec-http/src/main/java/io/netty/handler/codec/http/cookie/CookieHeaderNames.java", "codec-http/src/main/java/io/netty/handler/codec/http/cookie/DefaultCookie.java", "codec-http/src/main/java/io/netty/handler/codec/http/cookie/ServerCookieEncoder.java" ]

[ "codec-http/src/test/java/io/netty/handler/codec/http/cookie/ClientCookieDecoderTest.java", "codec-http/src/test/java/io/netty/handler/codec/http/cookie/ServerCookieEncoderTest.java" ]

"2023-12-19T08:47:26Z"

[ "keyword_pr_to_issue" ]

[ "Is this issue (https://github.com/grpc/grpc-java/issues/10830) resolved by this PR? ", "@njimenezotto That issue is completely different than this." ]

[ "Does this work correctly if there are two different messages in the `input` ByteBuffer waiting to be processed? In my minds eye it will skip them both.", "I think you're right, if there's more data already buffered it'll skip all of it. I'm trying to think of an alternative but if the length of the payload doesn't match the actual payload not sure how we can skip the correct number of bytes to read the next frame correctly. Maybe adding length checks while cumulating the data in the ByteToMessageDecoder?", "Instead of using a boolean flag we could instead do something like passing `input.readSlice(payloadLength)` into the `processPayloadState` method. That would require us to move the \r\n```\r\nif (input.readableBytes() < payloadLength) {\r\n // Wait until the entire payload has been read.\r\n return;\r\n}\r\n```\r\ncheck up out of the `processPayloadState` method to right before we slice and call it but tbh that feels more readable anyway.\r\n\r\nDoing that has two nice properties and one negative: it's nice because it 'slices' exactly one frame from `input` which (1) makes it hard for `processPayloadState` to corrupt other frames and (2) makes consumption of the frame from `input` happen eagerly so you don't have to worry about skipping bytes on exception. Downside is that it does cost an extra allocation. On balance, I'd happily pay the allocation of the slice wrapper: it isn't a drop in the ocean imo.", "That makes sense to me, but I think we'd hit some issues if the advertised `payloadLength` of the frame doesn't match the actual payload. In the use case I'm looking at, I see two cases where we could see issues\r\n- __Advertised Payload Length greater than payload__: More data skipped than actually included and next frame would not be properly read.\r\n- __Advertised Payload Length less than payload__: Less data skipped than actually included and data will be left over and next frame would not be properly read.", "@isaacrivriv I read description of #13788 to understand expected behavior, but it seems like this is not what RFC wants. [Section 4.2](https://datatracker.ietf.org/doc/html/rfc9113#section-4.2) says:\r\n\r\n> An endpoint MUST send an error code of [FRAME_SIZE_ERROR](https://datatracker.ietf.org/doc/html/rfc9113#FRAME_SIZE_ERROR) if a frame exceeds the size defined in [SETTINGS_MAX_FRAME_SIZE](https://datatracker.ietf.org/doc/html/rfc9113#SETTINGS_MAX_FRAME_SIZE), exceeds any limit defined for the frame type, or is too small to contain mandatory frame data. A frame size error in a frame that could alter the state of the entire connection MUST be treated as a [connection error](https://datatracker.ietf.org/doc/html/rfc9113#ConnectionErrorHandler) ([Section 5.4.1](https://datatracker.ietf.org/doc/html/rfc9113#ConnectionErrorHandler))\r\n\r\n\r\nThen section 5.4.1 describes that the state of the entire connection is expected to be corrupted at that point and we must close it. It's not expected that we could recover and process more frames on the same connection.\r\n\r\nIf we don't already have one, it will be helpful if you could implement an integration test with a full HTTP/2 server and a faked client that send a malformed frame, then assert that the server responds with `FRAME_SIZE_ERROR`, then closes connection. ", "In this case, the behavior I'm seeing with the response from a Netty HTTP2 server matches the RFC specification where in some cases even though a malformed frame comes in, it doesn't actually shut down the connection but just resets a stream and more streams can be accepted after that. [Section 6.3](https://datatracker.ietf.org/doc/html/rfc9113#name-priority) is an example\r\n\r\n> A PRIORITY frame with a length other than 5 octets MUST be treated as a [stream error](https://datatracker.ietf.org/doc/html/rfc9113#StreamErrorHandler) ([Section 5.4.2](https://datatracker.ietf.org/doc/html/rfc9113#StreamErrorHandler)) of type [FRAME_SIZE_ERROR](https://datatracker.ietf.org/doc/html/rfc9113#FRAME_SIZE_ERROR).\r\n\r\nAs it is currently working if I send a priority frame with length 4, I correctly get a reset frame back with a frame size error (which is what the RFC states) but starting another stream with a valid request after that, it is not parsed correctly due to leftover data from the previous request.\r\n\r\nThe error really happens in those cases where the RFC states that a stream reset is used rather than a connection error because when a connection error happens everything gets cleaned up where in a stream reset there could still be leftover data", "The most important part is: \r\n\r\n> A frame size error in a frame that could alter the state of the entire connection MUST be treated as a [connection error](https://datatracker.ietf.org/doc/html/rfc9113#ConnectionErrorHandler) ([Section 5.4.1](https://datatracker.ietf.org/doc/html/rfc9113#ConnectionErrorHandler))\r\n\r\nIt's ok to send `RST_STREAM` frame first. But then, how can we trust the frame if it has incorrect format? If we can not trust it, we don't know how many bytes to skip. It could happen that we received total garbage that just happen to start with \"priority frame header\". So, if we don't know how to continue, we should send GO_AWAY and close connection. ", "The RFC states the following after that statement which confuses me\r\n\r\n> A frame size error in a frame that could alter the state of the entire connection MUST be treated as a [connection error](https://datatracker.ietf.org/doc/html/rfc9113#ConnectionErrorHandler) ([Section 5.4.1](https://datatracker.ietf.org/doc/html/rfc9113#ConnectionErrorHandler)); **this includes any frame carrying a [field block](https://datatracker.ietf.org/doc/html/rfc9113#FieldBlock) ([Section 4.3](https://datatracker.ietf.org/doc/html/rfc9113#FieldBlock)) (that is, [HEADERS](https://datatracker.ietf.org/doc/html/rfc9113#HEADERS), [PUSH_PROMISE](https://datatracker.ietf.org/doc/html/rfc9113#PUSH_PROMISE), and [CONTINUATION](https://datatracker.ietf.org/doc/html/rfc9113#CONTINUATION)), a [SETTINGS](https://datatracker.ietf.org/doc/html/rfc9113#SETTINGS) frame, and any frame with a stream identifier of 0.**\r\n\r\nSo in my head the RFC states the specifics of what \"altering the state of the connection\" is in that sentence which this particular case doesn't seem to apply. By sending a priority frame to a specific stream (not 0) we can hit this issue and doesn't really fall under those categories IMO. \r\n\r\nI think we should skip the bytes if the payload length matches the actual frame payload and send a reset stream which matches the RFC and continue reading for the cases where a stream reset is required. I also agree closing the connection on cases where the payload length doesn't actually match the frame payload considering that we are receiving garbage in those cases but I'm making sure this isn't against the RFC since it states otherwise about keeping the connection active.\r\n\r\nThe other question would be, how would we know if the frame that came in actually matches the payload length? By the time Netty start's parsing H2 data we could have more messages queued in the buffer which we're not aware of. I'm thinking maybe a check on every decode to peek at the payload length and compare to the actual frame length? I'm not aware of a way to cleanly distinguish frames from each other TBH.", "It's convenient to think of the H2 protocol as having (at least) two layers: the framing layer (the first 9 octets of a frame) and then the actual frame payloads themselves which will have some consistency requirements. What I'm suggesting leverages the fact that the length from the framing layer should be authoritative in terms of frame boundaries.\r\nWhat to do if we detect corruption/inconsistencies in the payload (which could be that the payload is short or long, for example) will be case specific and gets into the details of what @idelpivnitskiy pointed to in his suggestions. This is actually a third advantage of what I'm suggestion: too few/extra bytes can signal corruption and if we slice we can detect that. For example, a RST_STREAM frame with size other than 4 octets should be a connection error (end of [6.4](https://datatracker.ietf.org/doc/html/rfc9113#section-6.4)). However in main if we received a RST_STREAM frame of size 5 (as determined by the frame header) where the last byte is meaningless instead of sending a connection error and potentially alerting the peer to a bug in their implementation we'll instead end up with a corrupt framing layer: we won't fail because we don't enforce the size and we won't consume that extra byte from the data stream so it will end up being interpreted as the first byte of the next frames header. In this specific example we could also just read the `payloadLength` field instead of checking that we have a buffer of size 4, but having a sliced buffer ensures decoders don't try to use more data than what the frame was advertised to contain via the header.", "Yeah that's exactly the issue I was seeing, leftover data is parsed as part of the next frame when malformed frames like that are parsed. So with @idelpivnitskiy comments and this thread I'm thinking we can implement the slice mechanism but there needs to be more checks implemented so that the payload length matches the actual payload and for those cases where we don't find a match we can alert by sending a Go Away frame? I'm trying to see what a path forward would be for this.", "See my comment above but the TL;DR is that the length field from the frame header is authoritative on size of a frames payload. What to do if that payload size is too short or too long when decoding the frame is frame type specific and is described as part of the individual frame definitions.", "That sounds right: whether to send a GOAWAY or a RST_STREAM is frame type specific and described as part of the frame definitions (usually at the ends). I don't think you have to implement all those checks as part of this PR if you don't want to: they could be done as a follow up. Finding this and making sure we keep a correct framing layer is already a huge step forward from where we are now.", "Taking a closer look it seems most of them (if not all) already verify the frame length and do the right thing wrt RST vs GOAWAY but its done as part of the `processHeaderState` state so we haven't been able to slice before the exception is thrown, and we may not even have that many bytes available, so we'd need to move the verification to a new method and perform it right before we `processPayloadState(ctx, input.slice(..), ..)`.", "All of these methods still work despite passing in a value with different meaning only because the writer index now matches the payload length. Now that you've sliced the buffer really we don't need to pass in the end length as each method could derive it from the passed in slice itself.", "fwiw, now that you've sliced the buffer I don't think you need to do this re-ordering.", "```suggestion\r\n // Stream error found, reset to reading headers once again for next frame coming in\r\n```", "Some of the verifications result in a stream error, not a connection error. If that happens I think we end up with a corrupt stream. To fix it we can simply move verification to after we've called input.readSlice(..) so that we've consumed the frame from the framing layer. We've already committed to reading frames this large so it doesn't seem like a potential attack vector. The only downside seems to be that for connection errors it can potentially delay closing the session.", "We should be able to simplify the management of this variable: we switch back to `readingHeaders = true` as soon as we consume the frame payload, so we can set it right after we `.readSlice(..)` and before we attempt to verify and process the payload and remove the mutations elsewhere.", "True, I'll update the method signatures to make this cleaner.", "Agree, I forgot that we aren't skipping the bytes anymore like the original solution if this happens. I'll move this verification after slicing", "This makes sense, I'll simplify with this in mind", "I think these can be even simpler: \r\n```suggestion\r\n int dataLength = lengthWithoutTrailingPadding(payload.readableBytes(), padding);\r\n```", "Do you think we could just adjust the writer index on `payload`?", "Agree, I'll make the changes to all of these ", "So instead of slicing the data buffer just move the writer index of the payload to dataLength?", "Yes: I think we sliced the payload previously because we didn't want to leak the parent buffer but since we've already sliced it there shouldn't be any harm in sharing that. At a minimum, I expect padding to be rare so if there isn't any padding to slice out we can certainly just send `payload` along.\r\n\r\nAnd to be more precise, I think we should do `payload.writeIndex(payload.readerIndex() + dataLength)`. `payload.readerIndex()` should always be 0 now, but it will make the code more resilient to future changes.", "Ok cool I'll make that change and saw the payload being further sliced while reading a Go Away frame so I'm updating that as well", "@bryce-anderson @idelpivnitskiy you are happy with what we have now ?", "@normanmaurer, I am. This should prevent frame stream corruption for problems that result in a stream error at the validation phase.\r\n\r\nOne behavior change that I think is worth point out is that we'll now aggregate the data for the next frame before doing any of the per-frame validations other than max frame length. I think this is safe because we've already agreed to accept frames that big such as data and headers frames but it's worth a thought.", "Consider naming it `verifyFrameState` because at the time we invoke this method, we already consumed the whole frame, not just its header. This will also align with other `verify*Frame()` methods it's calling. ", "Consider adding an assertion (ideally with a comment) that `in.readableBytes() == payloadLength`. In case in the future someone invokes this method from an incorrect state, it will rise an error.", "Consider adding a comment for future devs why we read the full header before throwing an excepting on case `payloadLength > maxFrameSize`", "@isaacrivriv thanks for pointing to the continuation of that sentence. That makes sense, for the current case we don't need to close the entire connection. Great catch!", "I agree, renaming it makes more sense I'll make this change", "Sounds good with me, I'll make this change", "I'm looking at this again and since it's a connectionError which closes the connection I don't think we have to read the full headers before throwing the exception so I'm undoing these changes" ]

"2024-01-18T05:02:39Z"

[]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/DefaultHttp2FrameReader.java" ]

[ "codec-http2/src/main/java/io/netty/handler/codec/http2/DefaultHttp2FrameReader.java" ]

[ "codec-http2/src/test/java/io/netty/handler/codec/http2/DefaultHttp2FrameReaderTest.java" ]

"2024-01-17T20:50:10Z"