x/net/http2: singleUse ClientConn with AllowHTTP transport is always unusable

[ctrlaltdel121]

What version of Go are you using (go version)?

$ go version 
go version go1.20.2 darwin/arm64

Does this issue reproduce with the latest release?

Yes

What operating system and processor architecture are you using (go env)?

go env Output

$ go env
GO111MODULE="on"
GOARCH="arm64"
GOHOSTARCH="arm64"
GOHOSTOS="darwin"
GOOS="darwin"
GOPROXY="direct"
GOROOT="/opt/homebrew/Cellar/go/1.20.2/libexec"
GOSUMDB="sum.golang.org"
GOTOOLDIR="/opt/homebrew/Cellar/go/1.20.2/libexec/pkg/tool/darwin_arm64"
GOVERSION="go1.20.2"
GCCGO="gccgo"
AR="ar"
CC="cc"
CXX="c++"
CGO_ENABLED="1"
CGO_CFLAGS="-O2 -g"
CGO_CXXFLAGS="-O2 -g"
CGO_FFLAGS="-O2 -g"
CGO_LDFLAGS="-O2 -g"
PKG_CONFIG="pkg-config"
GOGCCFLAGS="-fPIC -arch arm64 -pthread -fno-caret-diagnostics -Qunused-arguments -fmessage-length=0 -fdebug-prefix-map=/var/folders/g4/47vwdw1s65929qw7smyq1vlswhhzgj/T/go-build4150723989=/tmp/go-build -gno-record-gcc-switches -fno-common"

What did you do?

Create an HTTP2 transport with AllowHTTP = true, and try to send a request using it with req.Close = true. This will hang until the internal retry mechanism in the HTTP2 roundtrip ends after exponential backoff.

https://go.dev/play/p/KhmQcQB0PPD

If you remove req.Close = true, in my contrived example you get unexpected EOF, but the point is no timeout, because its not retrying aimlessly.

What did you expect to see?

I would expect the transport to understand that AllowHTTP=true setting nextStreamID to 3 in the newClientConn() function does not mean it should be considered "not usable" before it ever makes a request.

What did you see instead?

2009/11/10 23:00:00 Starting server...
2009/11/10 23:00:00 Sending request...
<time passes because of retry backoff>
http2: client conn not usable

The issue is because of this line that sets nextStreamID to 3 before a request is ever made, and then this line that interprets a nextStreamID higher than 1 as having been "used".

I'm not sure my use case for this is exactly "up to standard" but it seems like the behavior is unintentional - the connection should be usable once even if AllowHTTP was set on its creation, and at least it shouldn't retry exponentially when there's no way it could get a different outcome in the next iteration.

Thanks for reading!

[mknyszek]

CC @neild