p2p.Conn.Stats is not goroutine safe

[rozag]

Issue

I've been inspecting my project with go's race detector and stumbled upon a data race caused by p2p package.

Severity

It seems that the core functionality of the library is not affected by the race, but it makes it a bit harder to inspect apps that use the lib for data races because of a lot of "false positives", so to speak.

How to reproduce

Let's start with the basic garbled example:

1. Shell 1:

   cd apps/garbled
   go run --race . -e -i 800000 examples/millionaire.mpcl

2. Shell 2:

   cd apps/garbled
   go run --race . -i 750000 examples/millionaire.mpcl

3. Output in Shell 1:

    - In1: a{1,0}i64:int64
    + In2: b{1,0}i64:int64
    - Out: %_{0,1}b1:bool1
    -  In: [800000]
   Listening for connections at :8080
   New connection from 127.0.0.1:61520
   p2p.NewConn
   ==================
   WARNING: DATA RACE
   Read at 0x00c00037e0e8 by main goroutine:
     github.com/markkurossi/mpc/circuit.Evaluator()
         /Users/rozag/workspace/mpc/circuit/evaluator.go:115 +0xf44
     main.evaluatorMode()
         /Users/rozag/workspace/mpc/apps/garbled/main.go:359 +0x28d
     main.main()
         /Users/rozag/workspace/mpc/apps/garbled/main.go:311 +0x2ce6
   
   Previous write at 0x00c00037e0e8 by goroutine 14:
     github.com/markkurossi/mpc/p2p.(*Conn).writer()
         /Users/rozag/workspace/mpc/p2p/protocol.go:94 +0x244
     github.com/markkurossi/mpc/p2p.NewConn.func1()
         /Users/rozag/workspace/mpc/p2p/protocol.go:80 +0x39
   
   Goroutine 14 (running) created at:
     github.com/markkurossi/mpc/p2p.NewConn()
         /Users/rozag/workspace/mpc/p2p/protocol.go:80 +0x304
     main.evaluatorMode()
         /Users/rozag/workspace/mpc/apps/garbled/main.go:358 +0x244
     main.main()
         /Users/rozag/workspace/mpc/apps/garbled/main.go:311 +0x2ce6
   ==================
   Result[0]: false
   

4. Output in Shell 2:

    + In1: a{1,0}i64:int64
    - In2: b{1,0}i64:int64
    - Out: %_{0,1}b1:bool1
    -  In: [750000]
   p2p.NewConn
   ==================
   WARNING: DATA RACE
   Read at 0x00c000424178 by main goroutine:
     github.com/markkurossi/mpc/circuit.Garbler()
         /Users/rozag/workspace/mpc/circuit/garbler.go:139 +0xb24
     main.garblerMode()
         /Users/rozag/workspace/mpc/apps/garbled/main.go:381 +0x13b
     main.main()
         /Users/rozag/workspace/mpc/apps/garbled/main.go:318 +0x2eb5
   
   Previous write at 0x00c000424178 by goroutine 14:
     github.com/markkurossi/mpc/p2p.(*Conn).writer()
         /Users/rozag/workspace/mpc/p2p/protocol.go:94 +0x244
     github.com/markkurossi/mpc/p2p.NewConn.func1()
         /Users/rozag/workspace/mpc/p2p/protocol.go:80 +0x39
   
   Goroutine 14 (running) created at:
     github.com/markkurossi/mpc/p2p.NewConn()
         /Users/rozag/workspace/mpc/p2p/protocol.go:80 +0x304
     main.garblerMode()
         /Users/rozag/workspace/mpc/apps/garbled/main.go:378 +0xb3
     main.main()
         /Users/rozag/workspace/mpc/apps/garbled/main.go:318 +0x2eb5
   ==================
   Result[0]: false
   Found 1 data race(s)
   

The root cause

It seems that the problem is that we read and write to p2p.Conn.Stats.(Sent|Recvd) without any synchronization from multiple goroutines. And one of the goroutines is the one we start in p2p.NewConn.

Proposed solution

Leverage sync/atomic.Uint64 instead of raw uint64 in p2p.IOStats.

[markkurossi]

Fixed by merging Alexey's merge request.