context.WithTimeout() yields on Done() too early with no Err() set.

[dikderoy]

The following program sample.go triggers an unexpected result

//this is except from actual code from the plugin

package traefikplugin

import (
  "net/rpc"
  ...
)

// call is called once per request in the middleware
func call() err {
  ...
  reqStart := time.Now()
  reqCtx, cancel := context.WithTimeout(context.Background(), 30*time.Milliseconds)
  defer cancel()

  done := client.Go(method, args, reply, make(chan *rpc.Call, 1)).Done
  // wait for network or timeout
  select {
    case response := <-done:
	return response.Error
    case <-reqCtx.Done():
	return fmt.Errorf(
          "rpc: call timed out: %w, net_call/duration=%s",
          reqCtx.Err(),
          time.Since(reqStart),
        )
  }
}

Expected result

rpc: call timed out: context deadline exceeded, net_call/duration=30.209943ms

Got

rpc: call timed out: %!w(<nil>), net_call/duration=1.429943ms

Yaegi Version

v0.16.1

Additional Notes

It should handle timeouts correctly, and not yield prematurely

In Traefik context, on the logs I can observe messages like this from my custom written plugin:

rpc: call timed out: %!w(<nil>), net_call/duration=1.429943ms

which effectively means that context has yielded on reqCtx.Done() chan before the timeout was reached and the reqCtx.Err() is nil.

This occurs under relatively high load instance, with ~70hit/s on particular plugin utilizing the code and global load on same Traefik instance is ~400 hit/s.

The issue is occurring for ~0.08% (out of 70hits/s on the plugin) of requests handled.

so probably it is some kind of low level race condition in yaegi.

Sorry I haven't provided the fully functional code for reproducing the bug.

If you ultimately need it, I can try to create a minimal reproducible example a bit later.