RFR: 8327114: Attach in Linux may have wrong behaviour when pid == ns_pid (Kubernetes debug container)
Sebastian Lövdahl
duke at openjdk.org
Thu May 2 10:33:52 UTC 2024
On Thu, 2 May 2024 10:13:51 GMT, Sebastian Lövdahl <duke at openjdk.org> wrote:
> 8327114: Attach in Linux may have wrong behaviour when pid == ns_pid (Kubernetes debug container)
Ran the following tests locally:
$ make test TEST="jtreg:test/hotspot/jtreg/containers"
...
==============================
Test summary
==============================
TEST TOTAL PASS FAIL ERROR
jtreg:test/hotspot/jtreg/containers 14 14 0 0
==============================
TEST SUCCESS
$ make test TEST="jtreg:test/hotspot/jtreg/serviceability"
...
==============================
Test summary
==============================
TEST TOTAL PASS FAIL ERROR
jtreg:test/hotspot/jtreg/serviceability 374 374 0 0
==============================
TEST SUCCESS
And also manually tested it under various conditions. Basic environment information:
slovdahl at ubuntu2204:~/reproducer$ systemd --version
systemd 249 (249.11-0ubuntu3.12)
+PAM +AUDIT +SELINUX +APPARMOR +IMA +SMACK +SECCOMP +GCRYPT +GNUTLS +OPENSSL +ACL +BLKID +CURL +ELFUTILS +FIDO2 +IDN2 -IDN +IPTC +KMOD +LIBCRYPTSETUP +LIBFDISK +PCRE2 -PWQUALITY -P11KIT -QRENCODE +BZIP2 +LZ4 +XZ +ZLIB +ZSTD -XKBCOMMON +UTMP +SYSVINIT default-hierarchy=unified
slovdahl at ubuntu2204:~/reproducer$ sudo apt-get install openjdk-17-jdk-headless
slovdahl at ubuntu2204:~/reproducer$ /usr/lib/jvm/java-17-openjdk-amd64/bin/java -version
openjdk version "17.0.10" 2024-01-16
OpenJDK Runtime Environment (build 17.0.10+7-Ubuntu-122.04.1)
OpenJDK 64-Bit Server VM (build 17.0.10+7-Ubuntu-122.04.1, mixed mode, sharing)
slovdahl at ubuntu2204:~/reproducer$ /home/slovdahl/dev/external/jdk/build/linux-x86_64-server-release/images/jdk/bin/java -version
openjdk version "23-internal" 2024-09-17
OpenJDK Runtime Environment (build 23-internal-adhoc.slovdahl.jdk)
OpenJDK 64-Bit Server VM (build 23-internal-adhoc.slovdahl.jdk, mixed mode, sharing)
Reproducer.java used for the target process:
import java.io.IOException;
import java.net.InetSocketAddress;
import java.net.ServerSocket;
public class Reproducer {
public static void main(String[] args) throws InterruptedException, IOException {
int port;
if (args.length > 0) {
port = Integer.parseInt(args[0]);
}
else {
port = 81;
}
System.out.println("Hello, World!");
try (var server = new ServerSocket()) {
server.bind(new InetSocketAddress("localhost", port));
System.out.println("Bound to port " + port);
while (true) {
Thread.sleep(1_000L);
}
}
}
}
systemd unit used for testing the fix for https://bugs.openjdk.org/browse/JDK-8226919:
slovdahl at ubuntu2204:~/reproducer$ cat reproducer.service
[Service]
Type=simple
ExecStart=/usr/lib/jvm/java-17-openjdk-amd64/bin/java /home/slovdahl/reproducer/Reproducer.java
User=slovdahl
Group=slovdahl
ReadWritePaths=/tmp
AmbientCapabilities=CAP_NET_BIND_SERVICE
slovdahl at ubuntu2204:~/reproducer$ sudo cp -a reproducer.service /etc/systemd/system/
slovdahl at ubuntu2204:~/reproducer$ sudo systemctl daemon-reload
slovdahl at ubuntu2204:~/reproducer$ sudo systemctl start reproducer.service
slovdahl at ubuntu2204:~/reproducer$ sudo systemctl status reproducer.service
● reproducer.service
Loaded: loaded (/etc/systemd/system/reproducer.service; static)
Active: active (running) since Wed 2024-05-01 20:22:01 EEST; 4s ago
Main PID: 1576835 (java)
Tasks: 26 (limit: 76968)
Memory: 69.2M
CPU: 832ms
CGroup: /system.slice/reproducer.service
└─1576835 /usr/lib/jvm/java-17-openjdk-amd64/bin/java /home/slovdahl/reproducer/Reproducer.java
maj 01 20:22:01 ubuntu2204 systemd[1]: Started reproducer.service.
maj 01 20:22:01 ubuntu2204 java[1576835]: Hello, World!
maj 01 20:22:01 ubuntu2204 java[1576835]: Bound to port 81
slovdahl at ubuntu2204:~/reproducer$ ls -lh /proc/1576835/root
ls: cannot read symbolic link '/proc/1576835/root': Permission denied
lrwxrwxrwx 1 slovdahl slovdahl 0 maj 1 20:22 /proc/1576835/root
slovdahl at ubuntu2204:~/reproducer$ sudo ls -lh /proc/1576835/root
lrwxrwxrwx 1 slovdahl slovdahl 0 maj 1 20:22 /proc/1576835/root -> /
Fails with vanilla OpenJDK 17:
slovdahl at ubuntu2204:~/reproducer$ /usr/lib/jvm/java-17-openjdk-amd64/bin/jcmd 1576835 VM.version
1576835:
com.sun.tools.attach.AttachNotSupportedException: Unable to open socket file /proc/1576835/root/tmp/.java_pid1576835: target process 1576835 doesn't respond within 10500ms or HotSpot VM not loaded
at jdk.attach/sun.tools.attach.VirtualMachineImpl.<init>(VirtualMachineImpl.java:104)
at jdk.attach/sun.tools.attach.AttachProviderImpl.attachVirtualMachine(AttachProviderImpl.java:58)
at jdk.attach/com.sun.tools.attach.VirtualMachine.attach(VirtualMachine.java:207)
at jdk.jcmd/sun.tools.jcmd.JCmd.executeCommandForPid(JCmd.java:113)
at jdk.jcmd/sun.tools.jcmd.JCmd.main(JCmd.java:97)
Works when attaching as root with vanilla OpenJDK 17:
slovdahl at ubuntu2204:~/reproducer$ sudo /usr/lib/jvm/java-17-openjdk-amd64/bin/jcmd 1576835 VM.version
1576835:
OpenJDK 64-Bit Server VM version 17.0.10+7-Ubuntu-122.04.1
JDK 17.0.10
Still works without root with a JDK built from this PR (fixed in https://bugs.openjdk.org/browse/JDK-8226919):
slovdahl at ubuntu2204:~/reproducer$ /home/slovdahl/dev/external/jdk/build/linux-x86_64-server-release/images/jdk/bin/jcmd 1576835 VM.version
1576835:
OpenJDK 64-Bit Server VM version 17.0.10+7-Ubuntu-122.04.1
JDK 17.0.10
Attaching to a JVM inside a Docker container from the host works as before with vanilla OpenJDK 17 and the JDK built from this PR (always requires root):
slovdahl at ubuntu2204:~/reproducer$ docker run --rm -v .:/app -w /app eclipse-temurin:17 java Reproducer.java
Hello, World!
Bound to port 81
slovdahl at ubuntu2204:~/reproducer$ sudo jcmd -l | grep Reproducer
1587278 jdk.compiler/com.sun.tools.javac.launcher.Main Reproducer.java
slovdahl at ubuntu2204:~/reproducer$ sudo /usr/lib/jvm/java-17-openjdk-amd64/bin/jcmd 1587278 VM.version
1587278:
OpenJDK 64-Bit Server VM version 17.0.11+9
JDK 17.0.11
slovdahl at ubuntu2204:~/reproducer$ sudo /home/slovdahl/dev/external/jdk/build/linux-x86_64-server-release/images/jdk/bin/jcmd 1587278 VM.version
1587278:
OpenJDK 64-Bit Server VM version 17.0.11+9
JDK 17.0.11
Attaching to the same Docker container JVM from a sidecar container mounted into the same process namespace:
- Attaching works with Temurin 17, but `jcmd -l` does not list the process
- Attaching and `jcmd -l` works with Temurin 21
- Attaching and listing fails with current mainline JDK
- Attaching and listing works with the fix in this PR
slovdahl at ubuntu2204:~/reproducer$ docker run --interactive --tty --rm --pid=container:great_curie eclipse-temurin:17.0.11_9-jdk-jammy /bin/bash
root at e8fa333f7f71:/# jcmd
326 jdk.jcmd/sun.tools.jcmd.JCmd
root at e8fa333f7f71:/# jcmd 1 VM.version
1:
OpenJDK 64-Bit Server VM version 17.0.11+9
JDK 17.0.11
slovdahl at ubuntu2204:~/reproducer$ docker run --interactive --tty --rm --pid=container:great_curie eclipse-temurin:21.0.3_9-jdk-jammy /bin/bash
root at d7c84d951f3d:/# jcmd
384 jdk.jcmd/sun.tools.jcmd.JCmd
1 jdk.compiler/com.sun.tools.javac.launcher.Main Reproducer.java
root at d7c84d951f3d:/# jcmd 1 VM.version
1:
OpenJDK 64-Bit Server VM version 17.0.11+9
JDK 17.0.11
# Mainline JDK
slovdahl at ubuntu2204:~/reproducer$ docker run --interactive --tty --rm --pid=container:great_curie --volume /home/slovdahl/dev/external/jdk/build/linux-x86_64-server-release/images/jdk/:/jdk ubuntu:22.04 /bin/bash
root at 10595dab1b4d:/# /jdk/bin/jcmd
1 jdk.compiler/com.sun.tools.javac.launcher.Main Reproducer.java
443 jdk.jcmd/sun.tools.jcmd.JCmd
root at 10595dab1b4d:/# /jdk/bin/jcmd 1 VM.version
1:
com.sun.tools.attach.AttachNotSupportedException: Unable to open socket file /tmp/.java_pid1: target process 1 doesn't respond within 10500ms or HotSpot VM not loaded
at jdk.attach/sun.tools.attach.VirtualMachineImpl.<init>(VirtualMachineImpl.java:99)
at jdk.attach/sun.tools.attach.AttachProviderImpl.attachVirtualMachine(AttachProviderImpl.java:58)
at jdk.attach/com.sun.tools.attach.VirtualMachine.attach(VirtualMachine.java:207)
at jdk.jcmd/sun.tools.jcmd.JCmd.executeCommandForPid(JCmd.java:113)
at jdk.jcmd/sun.tools.jcmd.JCmd.main(JCmd.java:97)
# JDK built from this PR
slovdahl at ubuntu2204:~/reproducer$ docker run --interactive --tty --rm --pid=container:great_curie --volume /home/slovdahl/dev/external/jdk/build/linux-x86_64-server-release/images/jdk/:/jdk ubuntu:22.04 /bin/bash
root at 7b7ef4ba0b35:/# /jdk/bin/jcmd
1 jdk.compiler/com.sun.tools.javac.launcher.Main Reproducer.java
50 jdk.jcmd/sun.tools.jcmd.JCmd
root at 7b7ef4ba0b35:/# /jdk/bin/jcmd 1 VM.version
1:
OpenJDK 64-Bit Server VM version 17.0.11+9
JDK 17.0.11
-------------
PR Comment: https://git.openjdk.org/jdk/pull/19055#issuecomment-2090136676
More information about the hotspot-runtime-dev
mailing list