optimize performance: reduce allocations, buffered I/O, and zero-copy reads

- Read path: eliminate redundant allocation in bsPerformCommand - remove
  the pre-allocation before bs.Read() and the append loop for zero-fill,
  use direct copy and in-place zero-fill instead
- parseHeader: use command pool (getCommand) instead of direct allocation,
  reducing GC pressure on the hot path
- Unmap: use a shared 1MB zero buffer instead of allocating per-descriptor,
  dramatically reducing allocations for large unmap operations
- Network I/O: add 256KB bufio.Writer to iSCSI connections, batching
  small PDU writes into fewer syscalls. Flush after txHandler completes

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

pkg/port/iscsit/cmd.go

@@ -312,8 +312,7 @@ func parseHeader(data []byte) (*ISCSICommand, error) {
 	if len(data) != BHS_SIZE {
 		return nil, fmt.Errorf("garbled header")
 	}
-	// TODO: sync.Pool
-	m := &ISCSICommand{}
+	m := getCommand()
 	m.Immediate = 0x40&data[0] == 0x40
 	m.OpCode = OpCode(data[0] & ISCSI_OPCODE_MASK)
 	m.Final = 0x80&data[1] == 0x80

pkg/port/iscsit/conn.go

@@ -17,6 +17,7 @@ limitations under the License.
 package iscsit
 
 import (
+	"bufio"
 	"io"
 	"net"
 	"sort"
@@ -61,6 +62,7 @@ type iscsiConnection struct {
 	txIOState int
 	refcount  int
 	conn      net.Conn
+	writer    *bufio.Writer
 
 	rxBuffer []byte
 	txBuffer []byte
@@ -117,6 +119,7 @@ func (c *iscsiConnection) init() {
 	c.state = CONN_STATE_FREE
 	c.refcount = 1
 	c.readLock = new(sync.RWMutex)
+	c.writer = bufio.NewWriterSize(c.conn, 256*1024)
 	c.loginParam.sessionParam = []ISCSISessionParam{}
 	c.loginParam.tgtCSG = LoginOperationalNegotiation
 	c.loginParam.tgtNSG = LoginOperationalNegotiation
@@ -136,10 +139,15 @@ func (c *iscsiConnection) readData(buf []byte) (int, error) {
 }
 
 func (c *iscsiConnection) write(resp []byte) (int, error) {
-	return c.conn.Write(resp)
+	return c.writer.Write(resp)
+}
+
+func (c *iscsiConnection) flush() error {
+	return c.writer.Flush()
 }
 
 func (c *iscsiConnection) close() {
+	c.writer.Flush()
 	c.conn.Close()
 }
 

pkg/port/iscsit/iscsid.go

@@ -830,6 +830,13 @@ SendRemainingData:
 		}
 	}
 
+	// Flush buffered writes to the network
+	if err := conn.flush(); err != nil {
+		log.Errorf("failed to flush data to client: %v", err)
+		conn.state = CONN_STATE_CLOSE
+		return
+	}
+
 	log.Debugf("connection state: %v", conn.State())
 	switch conn.state {
 	case CONN_STATE_CLOSE, CONN_STATE_EXIT: