lock: Moving locking to handler layer. (#3381)

This is implemented so that the issues like in the
following flow don't affect the behavior of operation.

```
GetObjectInfo()
.... --> Time window for mutation (no lock held)
.... --> Time window for mutation (no lock held)
GetObject()
```

This happens when two simultaneous uploads are made
to the same object the object has returned wrong
info to the client.

Another classic example is "CopyObject" API itself
which reads from a source object and copies to
destination object.

Fixes #3370
Fixes #2912

cmd/fs-v1.go

@@ -246,11 +246,6 @@ func (fs fsObjects) GetObject(bucket, object string, offset int64, length int64,
 		return traceError(InvalidRange{offset, length, fi.Size})
 	}
 
-	// Lock the object before reading.
-	objectLock := nsMutex.NewNSLock(bucket, object)
-	objectLock.RLock()
-	defer objectLock.RUnlock()
-
 	var totalLeft = length
 	bufSize := int64(readSizeV1)
 	if length > 0 && bufSize > length {
@@ -450,11 +445,6 @@ func (fs fsObjects) PutObject(bucket string, object string, size int64, data io.
 		}
 	}
 
-	// Lock the object before committing the object.
-	objectLock := nsMutex.NewNSLock(bucket, object)
-	objectLock.RLock()
-	defer objectLock.RUnlock()
-
 	// Entire object was written to the temp location, now it's safe to rename it to the actual location.
 	err = fs.storage.RenameFile(minioMetaTmpBucket, tempObj, bucket, object)
 	if err != nil {
@@ -484,12 +474,6 @@ func (fs fsObjects) DeleteObject(bucket, object string) error {
 		return err
 	}
 
-	// Lock the object before deleting so that an in progress GetObject does not return
-	// corrupt data or there is no race with a PutObject.
-	objectLock := nsMutex.NewNSLock(bucket, object)
-	objectLock.RLock()
-	defer objectLock.RUnlock()
-
 	if bucket != minioMetaBucket {
 		// We don't store fs.json for minio-S3-layer created files like policy.json,
 		// hence we don't try to delete fs.json for such files.