delayed locks until we have started reading the body (#10474)

This is to ensure that Go contexts work properly, after some
interesting experiments I found that Go net/http doesn't
cancel the context when Body is non-zero and hasn't been
read till EOF.

The following gist explains this, this can lead to pile up
of go-routines on the server which will never be canceled
and will die at a really later point in time, which can
simply overwhelm the server.

https://gist.github.com/harshavardhana/c51dcfd055780eaeb71db54f9c589150

To avoid this refactor the locking such that we take locks after we
have started reading from the body and only take locks when needed.

Also, remove contextReader as it's not useful, doesn't work as expected
context is not canceled until the body reaches EOF so there is no point
in wrapping it with context and putting a `select {` on it which
can unnecessarily increase the CPU overhead.

We will still use the context to cancel the lockers etc.
Additional simplification in the locker code to avoid timers
as re-using them is a complicated ordeal avoid them in
the hot path, since locking is very common this may avoid
lots of allocations.

cmd/erasure-object.go

@@ -145,6 +145,32 @@ func (er erasureObjects) GetObjectNInfo(ctx context.Context, bucket, object stri
 		return nil, err
 	}
 
+	var unlockOnDefer bool
+	var nsUnlocker = func() {}
+	defer func() {
+		if unlockOnDefer {
+			nsUnlocker()
+		}
+	}()
+
+	// Acquire lock
+	if lockType != noLock {
+		lock := er.NewNSLock(ctx, bucket, object)
+		switch lockType {
+		case writeLock:
+			if err = lock.GetLock(globalOperationTimeout); err != nil {
+				return nil, err
+			}
+			nsUnlocker = lock.Unlock
+		case readLock:
+			if err = lock.GetRLock(globalOperationTimeout); err != nil {
+				return nil, err
+			}
+			nsUnlocker = lock.RUnlock
+		}
+		unlockOnDefer = true
+	}
+
 	// Handler directory request by returning a reader that
 	// returns no bytes.
 	if HasSuffix(object, SlashSeparator) {
@@ -152,7 +178,8 @@ func (er erasureObjects) GetObjectNInfo(ctx context.Context, bucket, object stri
 		if objInfo, err = er.getObjectInfoDir(ctx, bucket, object); err != nil {
 			return nil, toObjectErr(err, bucket, object)
 		}
-		return NewGetObjectReaderFromReader(bytes.NewBuffer(nil), objInfo, opts)
+		unlockOnDefer = false
+		return NewGetObjectReaderFromReader(bytes.NewBuffer(nil), objInfo, opts, nsUnlocker)
 	}
 
 	fi, metaArr, onlineDisks, err := er.getObjectFileInfo(ctx, bucket, object, opts)
@@ -173,7 +200,8 @@ func (er erasureObjects) GetObjectNInfo(ctx context.Context, bucket, object stri
 		}, toObjectErr(errMethodNotAllowed, bucket, object)
 	}
 
-	fn, off, length, nErr := NewGetObjectReader(rs, objInfo, opts)
+	unlockOnDefer = false
+	fn, off, length, nErr := NewGetObjectReader(rs, objInfo, opts, nsUnlocker)
 	if nErr != nil {
 		return nil, nErr
 	}
@@ -202,6 +230,13 @@ func (er erasureObjects) GetObject(ctx context.Context, bucket, object string, s
 		return err
 	}
 
+	// Lock the object before reading.
+	lk := er.NewNSLock(ctx, bucket, object)
+	if err := lk.GetRLock(globalOperationTimeout); err != nil {
+		return err
+	}
+	defer lk.RUnlock()
+
 	// Start offset cannot be negative.
 	if startOffset < 0 {
 		logger.LogIf(ctx, errUnexpected, logger.Application)
@@ -386,6 +421,13 @@ func (er erasureObjects) GetObjectInfo(ctx context.Context, bucket, object strin
 		return info, err
 	}
 
+	// Lock the object before reading.
+	lk := er.NewNSLock(ctx, bucket, object)
+	if err := lk.GetRLock(globalOperationTimeout); err != nil {
+		return ObjectInfo{}, err
+	}
+	defer lk.RUnlock()
+
 	if HasSuffix(object, SlashSeparator) {
 		info, err = er.getObjectInfoDir(ctx, bucket, object)
 		if err != nil {
@@ -738,6 +780,12 @@ func (er erasureObjects) putObject(ctx context.Context, bucket string, object st
 		return ObjectInfo{}, toObjectErr(err, bucket, object)
 	}
 
+	lk := er.NewNSLock(ctx, bucket, object)
+	if err := lk.GetLock(globalOperationTimeout); err != nil {
+		return ObjectInfo{}, err
+	}
+	defer lk.Unlock()
+
 	// Rename the successfully written temporary object to final location.
 	if onlineDisks, err = renameData(ctx, onlineDisks, minioMetaTmpBucket, tempObj, fi.DataDir, bucket, object, writeQuorum, nil); err != nil {
 		return ObjectInfo{}, toObjectErr(err, bucket, object)
@@ -970,6 +1018,13 @@ func (er erasureObjects) DeleteObject(ctx context.Context, bucket, object string
 		return objInfo, err
 	}
 
+	// Acquire a write lock before deleting the object.
+	lk := er.NewNSLock(ctx, bucket, object)
+	if err = lk.GetLock(globalOperationTimeout); err != nil {
+		return ObjectInfo{}, err
+	}
+	defer lk.Unlock()
+
 	storageDisks := er.getDisks()
 	writeQuorum := len(storageDisks)/2 + 1