net/http: nil pointer dereference in closeConnIfStillIdle

[reusee]

1. What version of Go are you using (go version)?
   go version devel +b75b063 Tue Jun 28 04:49:33 2016 +0000 linux/amd64
2. What operating system and processor architecture are you using (go env)?
   GOARCH="amd64"
   GOBIN=""
   GOEXE=""
   GOHOSTARCH="amd64"
   GOHOSTOS="linux"
   GOOS="linux"
   GOPATH="/home/reus"
   GORACE=""
   GOROOT="/home/reus/go"
   GOTOOLDIR="/home/reus/go/pkg/tool/linux_amd64"
   CC="gcc"
   GOGCCFLAGS="-fPIC -m64 -pthread -fmessage-length=0 -fdebug-prefix-map=/tmp/go-build483039895=/tmp/go-build -gno-record-gcc-switches"
   CXX="g++"
   CGO_ENABLED="1"
3. What did you see instead?

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x4de58f]

goroutine 54787 [running]:
panic(0x7a2f40, 0xc4200120c0)
    /home/reus/go/src/runtime/panic.go:500 +0x1a1
net/http.(*persistConn).closeConnIfStillIdle(0xc4249cfb00)
    /home/reus/go/src/net/http/transport.go:1255 +0x2f
net/http.(*persistConn).(net/http.closeConnIfStillIdle)-fm()
    /home/reus/go/src/net/http/transport.go:643 +0x2a
created by time.goFunc
    /home/reus/go/src/time/sleep.go:154 +0x44

[ALTree]

You should include a recipe to reproduce the error in your report. Point 3 in the issue report list says

3. What did you do?

If possible, provide a recipe for reproducing the error.

A complete runnable program is good.

A link on play.golang.org is best.

[bradfitz]

In addition to code, as @ALTree mentioned, does your program run under the race detector?

[reusee]

@bradfitz No, not with race detector.

I don't think it will be easy to reproduce by a short example. It's a large long-running program. I believe it's introduced by recent changes. I will try to do some git bisect to find out which commit.

[bradfitz]

No, not with race detector.

Does that mean you haven't tried, or it fails?

I will try to do some git bisect to find out which commit.

Please do. I labeled this "WaitingForInfo" since there's nothing for us to help with until we have more to go on. If you get us something soon, there's a chance it could be fixed for Go 1.7, if there's a problem.

[reusee]

@bradfitz I didn't run the program with race detector, I will give it a try.

[reusee]

Close because the program ran for several days with no crash at all, with race detector enabled / disabled. No way to reproduce.

[vadmeste]

This bug occurred with me too (minio/minio#2456). However, it is not reproducible anymore. So, just wanted to encourage you to review the code if many other people experiment the same issue.

[bradfitz]

@vadmeste, if you find a reproducible example, please file a new bug. You can reference this one. But we don't re-use old bugs.

[kidoman]

Just got this today after upgrading to Go 1.7. Try to see if it occurs multiple times and will file a new bug. As this occurs in a go routine outside our control, it ended up bringing down the entire service.

[bradfitz]

@kidoman, if you find a reproducible example, please file a new bug. You can reference this one. But we don't re-use old bugs.