TCP hangs when sending

[pothos]

Hello,
actually I wanted to debug it, but need some help because it is disappears e.g. if writing to pcap or extensive logging is used.

The smoltcp server thread on tap0 will send in 64 byte steps, a Linux socket client thread will read only to ~292800 bytes instead of reading all 10000000 bytes. Smoltcp is handing out keep-alives from this point on. The minimal example is here:

https://github.com/pothos/taptest

cargo run --release should reproduce the behavior.

RUST_LOG=trace cargo run --release should let it sometimes complete (same as adding print statements in the sending server).
Uncommenting the pcap writer will let it complete reliably.

I have experienced it with other values as well, but hope that this combination also reproduces it on other machines.

Best regards,
Kai

[whitequark]

Oh yeah you hit a bug; retransmit timer gets reset by ACKs whereas it should not.

Specifically, by duplicate ACKs that arrive when the window is zero length, and pass through the duplicate ACK checks because they count as keepalives.

[pothos]

Thanks for looking at it, now I have to read the RFCs along with the code and find out how the timer is supposed to be set and kept.

[pothos]

Ok, finally got it working. Besides keeping the retransmission timer there was an overflow that needed to be fixed. Preparing a PR now.

[whitequark]

@pothos I'm finally making progress on this—see d1e2292 for one method I'm using to expose underlying bugs. I am afraid that your fix is hiding one of them.

[whitequark]

@pothos I've integrated a test quite like your taptest in the commit 44db954. I've managed to expose one other bug as well.

[whitequark]

Fixed in b247f64.

[pothos]

Thanks for including the fix - I am happy that the wrapping can be covered without distinction of cases ;)