vsock/internal/poll: hard vendoring go/internal/poll@74e384f50d30

Signed-off-by: Koichi Shiraishi <[email protected]>

Author: zchee

vsock/internal/poll/LICENSE

@@ -0,0 +1,27 @@
+Copyright (c) 2009 The Go Authors. All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are
+met:
+
+   * Redistributions of source code must retain the above copyright
+notice, this list of conditions and the following disclaimer.
+   * Redistributions in binary form must reproduce the above
+copyright notice, this list of conditions and the following disclaimer
+in the documentation and/or other materials provided with the
+distribution.
+   * Neither the name of Google Inc. nor the names of its
+contributors may be used to endorse or promote products derived from
+this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

vsock/internal/poll/README.md

@@ -0,0 +1,3 @@
+# poll
+
+This package forked and edit from [go/src/internal/poll@74e384f50d30](https://github.com/golang/go/tree/74e384f50d30/src/internal/poll)

vsock/internal/poll/copy_file_range_linux.go

@@ -0,0 +1,165 @@
+// Copyright 2020 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package poll
+
+import (
+	"sync/atomic"
+	"syscall"
+
+	"golang.org/x/sys/unix"
+)
+
+var copyFileRangeSupported int32 = -1 // accessed atomically
+
+const maxCopyFileRangeRound = 1 << 30
+
+func kernelVersion() (major int, minor int) {
+	var uname syscall.Utsname
+	if err := syscall.Uname(&uname); err != nil {
+		return
+	}
+
+	rl := uname.Release
+	var values [2]int
+	vi := 0
+	value := 0
+	for _, c := range rl {
+		if '0' <= c && c <= '9' {
+			value = (value * 10) + int(c-'0')
+		} else {
+			// Note that we're assuming N.N.N here.  If we see anything else we are likely to
+			// mis-parse it.
+			values[vi] = value
+			vi++
+			if vi >= len(values) {
+				break
+			}
+			value = 0
+		}
+	}
+	switch vi {
+	case 0:
+		return 0, 0
+	case 1:
+		return values[0], 0
+	case 2:
+		return values[0], values[1]
+	}
+	return
+}
+
+// CopyFileRange copies at most remain bytes of data from src to dst, using
+// the copy_file_range system call. dst and src must refer to regular files.
+func CopyFileRange(dst, src *FD, remain int64) (written int64, handled bool, err error) {
+	if supported := atomic.LoadInt32(&copyFileRangeSupported); supported == 0 {
+		return 0, false, nil
+	} else if supported == -1 {
+		major, minor := kernelVersion()
+		if major > 5 || (major == 5 && minor >= 3) {
+			atomic.StoreInt32(&copyFileRangeSupported, 1)
+		} else {
+			// copy_file_range(2) is broken in various ways on kernels older than 5.3,
+			// see issue #42400 and
+			// https://man7.org/linux/man-pages/man2/copy_file_range.2.html#VERSIONS
+			atomic.StoreInt32(&copyFileRangeSupported, 0)
+			return 0, false, nil
+		}
+	}
+	for remain > 0 {
+		max := remain
+		if max > maxCopyFileRangeRound {
+			max = maxCopyFileRangeRound
+		}
+		n, err := copyFileRange(dst, src, int(max))
+		switch err {
+		case syscall.ENOSYS:
+			// copy_file_range(2) was introduced in Linux 4.5.
+			// Go supports Linux >= 2.6.33, so the system call
+			// may not be present.
+			//
+			// If we see ENOSYS, we have certainly not transferred
+			// any data, so we can tell the caller that we
+			// couldn't handle the transfer and let them fall
+			// back to more generic code.
+			//
+			// Seeing ENOSYS also means that we will not try to
+			// use copy_file_range(2) again.
+			atomic.StoreInt32(&copyFileRangeSupported, 0)
+			return 0, false, nil
+		case syscall.EXDEV, syscall.EINVAL, syscall.EIO, syscall.EOPNOTSUPP, syscall.EPERM:
+			// Prior to Linux 5.3, it was not possible to
+			// copy_file_range across file systems. Similarly to
+			// the ENOSYS case above, if we see EXDEV, we have
+			// not transferred any data, and we can let the caller
+			// fall back to generic code.
+			//
+			// As for EINVAL, that is what we see if, for example,
+			// dst or src refer to a pipe rather than a regular
+			// file. This is another case where no data has been
+			// transferred, so we consider it unhandled.
+			//
+			// If src and dst are on CIFS, we can see EIO.
+			// See issue #42334.
+			//
+			// If the file is on NFS, we can see EOPNOTSUPP.
+			// See issue #40731.
+			//
+			// If the process is running inside a Docker container,
+			// we might see EPERM instead of ENOSYS. See issue
+			// #40893. Since EPERM might also be a legitimate error,
+			// don't mark copy_file_range(2) as unsupported.
+			return 0, false, nil
+		case nil:
+			if n == 0 {
+				// If we did not read any bytes at all,
+				// then this file may be in a file system
+				// where copy_file_range silently fails.
+				// https://lore.kernel.org/linux-fsdevel/[email protected]/T/#m05753578c7f7882f6e9ffe01f981bc223edef2b0
+				if written == 0 {
+					return 0, false, nil
+				}
+				// Otherwise src is at EOF, which means
+				// we are done.
+				return written, true, nil
+			}
+			remain -= n
+			written += n
+		default:
+			return written, true, err
+		}
+	}
+	return written, true, nil
+}
+
+// copyFileRange performs one round of copy_file_range(2).
+func copyFileRange(dst, src *FD, max int) (written int64, err error) {
+	// The signature of copy_file_range(2) is:
+	//
+	// ssize_t copy_file_range(int fd_in, loff_t *off_in,
+	//                         int fd_out, loff_t *off_out,
+	//                         size_t len, unsigned int flags);
+	//
+	// Note that in the call to unix.CopyFileRange below, we use nil
+	// values for off_in and off_out. For the system call, this means
+	// "use and update the file offsets". That is why we must acquire
+	// locks for both file descriptors (and why this whole machinery is
+	// in the internal/poll package to begin with).
+	if err := dst.writeLock(); err != nil {
+		return 0, err
+	}
+	defer dst.writeUnlock()
+	if err := src.readLock(); err != nil {
+		return 0, err
+	}
+	defer src.readUnlock()
+	var n int
+	for {
+		n, err = unix.CopyFileRange(src.Sysfd, nil, dst.Sysfd, nil, max, 0)
+		if err != syscall.EINTR {
+			break
+		}
+	}
+	return int64(n), err
+}

vsock/internal/poll/errno_unix.go

@@ -0,0 +1,34 @@
+// Copyright 2019 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+//go:build aix || darwin || dragonfly || freebsd || linux || netbsd || openbsd || solaris
+// +build aix darwin dragonfly freebsd linux netbsd openbsd solaris
+
+package poll
+
+import "syscall"
+
+// Do the interface allocations only once for common
+// Errno values.
+var (
+	errEAGAIN error = syscall.EAGAIN
+	errEINVAL error = syscall.EINVAL
+	errENOENT error = syscall.ENOENT
+)
+
+// errnoErr returns common boxed Errno values, to prevent
+// allocations at runtime.
+func errnoErr(e syscall.Errno) error {
+	switch e {
+	case 0:
+		return nil
+	case syscall.EAGAIN:
+		return errEAGAIN
+	case syscall.EINVAL:
+		return errEINVAL
+	case syscall.ENOENT:
+		return errENOENT
+	}
+	return e
+}

vsock/internal/poll/errno_windows.go

@@ -0,0 +1,32 @@
+// Copyright 2019 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+//go:build windows
+// +build windows
+
+package poll
+
+import "syscall"
+
+// Do the interface allocations only once for common
+// Errno values.
+
+var (
+	errERROR_IO_PENDING error = syscall.Errno(syscall.ERROR_IO_PENDING)
+)
+
+// ErrnoErr returns common boxed Errno values, to prevent
+// allocations at runtime.
+func errnoErr(e syscall.Errno) error {
+	switch e {
+	case 0:
+		return nil
+	case syscall.ERROR_IO_PENDING:
+		return errERROR_IO_PENDING
+	}
+	// TODO: add more here, after collecting data on the common
+	// error values see on Windows. (perhaps when running
+	// all.bat?)
+	return e
+}

vsock/internal/poll/error_linux_test.go

@@ -0,0 +1,32 @@
+// Copyright 2019 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package poll_test
+
+import (
+	"errors"
+	"os"
+	"syscall"
+
+	"github.com/go-hypervisor/virtio/vsock/internal/poll"
+)
+
+func badStateFile() (*os.File, error) {
+	if os.Getuid() != 0 {
+		return nil, errors.New("must be root")
+	}
+	// Using OpenFile for a device file is an easy way to make a
+	// file attached to the runtime-integrated network poller and
+	// configured in halfway.
+	return os.OpenFile("/dev/net/tun", os.O_RDWR, 0)
+}
+
+func isBadStateFileError(err error) (string, bool) {
+	switch err {
+	case poll.ErrNotPollable, syscall.EBADFD:
+		return "", true
+	default:
+		return "not pollable or file in bad state error", false
+	}
+}

vsock/internal/poll/error_stub_test.go

@@ -0,0 +1,22 @@
+// Copyright 2019 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+//go:build !linux
+// +build !linux
+
+package poll_test
+
+import (
+	"errors"
+	"os"
+	"runtime"
+)
+
+func badStateFile() (*os.File, error) {
+	return nil, errors.New("not supported on " + runtime.GOOS)
+}
+
+func isBadStateFileError(err error) (string, bool) {
+	return "", false
+}

vsock/internal/poll/error_test.go

@@ -0,0 +1,51 @@
+// Copyright 2019 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package poll_test
+
+import (
+	"fmt"
+	"io/fs"
+	"net"
+	"os"
+	"testing"
+	"time"
+)
+
+func TestReadError(t *testing.T) {
+	t.Run("ErrNotPollable", func(t *testing.T) {
+		f, err := badStateFile()
+		if err != nil {
+			t.Skip(err)
+		}
+		defer f.Close()
+
+		// Give scheduler a chance to have two separated
+		// goroutines: an event poller and an event waiter.
+		time.Sleep(100 * time.Millisecond)
+
+		var b [1]byte
+		_, err = f.Read(b[:])
+		if perr := parseReadError(err, isBadStateFileError); perr != nil {
+			t.Fatal(perr)
+		}
+	})
+}
+
+func parseReadError(nestedErr error, verify func(error) (string, bool)) error {
+	err := nestedErr
+	if nerr, ok := err.(*net.OpError); ok {
+		err = nerr.Err
+	}
+	if nerr, ok := err.(*fs.PathError); ok {
+		err = nerr.Err
+	}
+	if nerr, ok := err.(*os.SyscallError); ok {
+		err = nerr.Err
+	}
+	if s, ok := verify(err); !ok {
+		return fmt.Errorf("got %v; want %s", nestedErr, s)
+	}
+	return nil
+}

vsock/internal/poll/export_linux_test.go

@@ -0,0 +1,22 @@
+// Copyright 2021 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Export guts for testing on linux.
+// Since testing imports os and os imports internal/poll,
+// the internal/poll tests can not be in package poll.
+
+package poll
+
+var (
+	GetPipe     = getPipe
+	PutPipe     = putPipe
+	NewPipe     = newPipe
+	DestroyPipe = destroyPipe
+)
+
+func GetPipeFds(p *SplicePipe) (int, int) {
+	return p.rfd, p.wfd
+}
+
+type SplicePipe = splicePipe