Crash in Netty HTTP/3 when maxIdleTimeout is set to very large values

[imz87]

Hi @normanmaurer ,
When using Netty's HTTP/3 support, setting maxIdleTimeout to an extremely large value (e.g., Long.MAX_VALUE) causes the application to crash at runtime. This happens deep in the QUIC/Rust layer.

Stacktrace:

thread '<unnamed>' panicked at octets/src/lib.rs:679:9:

internal error: entered unreachable code

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

fatal runtime error: failed to initiate panic, error 5

Steps to Reproduce:

• Set maxIdleTimeout to Long.MAX_VALUE or another very large value.
• Start the application and initiate a HTTP/3 connection by curl.
• Observe the crash.

I used Oracle OpenJDK15 to test that. IMHO, the API should reject large values with a clear exception. It should not cause a native crash.

Thanks,

[normanmaurer]

@imz87 sounds like a bug in our quic implementation... Will have a look

[imz87]

thanks

[normanmaurer]

@imz87 is there an easy way to reproduce... Was not able to reproduce it yet ..

[imz87]

@imz87 is there an easy way to reproduce... Was not able to reproduce it yet ..

@normanmaurer , please run the Http3ClientExample in the branch : https://github.com/imz87/netty-incubator-codec-http3/tree/bug/crash-on-large-maxIdleTimeout
then it will show the problem.

[normanmaurer]

looking into it...

[normanmaurer]

This is a bug in quiche. Proposed a fix cloudflare/quiche#2155

[imz87]

thanks for proposing the fix in quiche

[normanmaurer]

Ok it was merged in quiche and will be part of the next release: cloudflare/quiche@1906381

[imz87]

thanks, good to hear it’s part of the next release.