Marten Seemann f07d6939d0 congestion: avoid overflows when calculating pacer budget (#5390) ...

* congestion: prevent uint64 overflow in pacer

This should never happen for bandwidths even remotely possible in
real-world scenarios, but it’s better to be safe.

* congestion: add a benchmark test for the pacer

* add a comment

2025-10-17 07:30:07 +02:00

..

ackhandler

use default RTT (100ms) for 0-RTT if no prior estimate (#5388)

2025-10-16 09:32:46 +02:00

congestion

congestion: avoid overflows when calculating pacer budget (#5390)

2025-10-17 07:30:07 +02:00

flowcontrol

use default RTT (100ms) for 0-RTT if no prior estimate (#5388)

2025-10-16 09:32:46 +02:00

handshake

use default RTT (100ms) for 0-RTT if no prior estimate (#5388)

2025-10-16 09:32:46 +02:00

mocks

qlog: split serializiation and event definitions, remove logging abstraction (#5356)

2025-10-08 05:53:02 +02:00

monotime

implement a memory-optimized time.Time replacement (#5334)

2025-09-14 08:12:10 +02:00

protocol

protocol: optimize ConnectionID.String (#5351)

2025-10-03 14:27:39 +08:00

qerr

use assert.AnError consistently in tests (#5066)

2025-04-21 03:52:08 +02:00

qtls

handshake: use new crypto/tls 0-RTT API (#4953)

2025-02-14 03:17:01 +01:00

synctest

use synctest to make streams map tests fully deterministic (#5296)

2025-08-23 18:11:59 +02:00

testdata

testdata: migrate tests away from Ginkgo (#4670)

2024-09-12 00:42:29 -07:00

utils

use default RTT (100ms) for 0-RTT if no prior estimate (#5388)

2025-10-16 09:32:46 +02:00

wire

use testing.B.Loop in all benchmark tests (#5285)

2025-08-16 12:44:56 +02:00