fix(storage): revert optimize gRPC writer with zero-copy

[cpriti-os]

This reverts commit df64147.

[gemini-code-assist]

Code Review

This pull request simplifies the gRPCWriter by removing the zero-copy write path and transitioning from lazy to immediate buffer allocation. Feedback from the reviewer highlights that immediate allocation increases memory overhead and that relying on cap(w.buf) instead of a dedicated chunkSize field could result in a division-by-zero panic if the chunk size is zero. It is recommended to re-introduce lazy allocation and the chunkSize field to ensure robustness and efficiency.

[gemini-code-assist]

This logic assumes cap(w.buf) is non-zero. If chunkSize is 0 (which is a valid setting in the storage package), this could lead to a division-by-zero panic in the handle method. Re-introducing lazy allocation with a chunkSize field and ensuring a minimum value would be safer.

Suggested change

	if len(w.buf)+len(v.p) <= cap(w.buf) {
	// We have not started sending yet, and we can stage all data without
	// starting a send. Compare against cap(w.buf) instead of
	// w.writeQuantum: that way we can perform a oneshot upload for objects
	// which fit in one chunk, even though we will cut the request into
	// w.writeQuantum units when we do start sending.
	origLen := len(w.buf)
	w.buf = w.buf[:origLen+len(v.p)]
	copy(w.buf[origLen:], v.p)
	close(v.done)
	} else {
	// Too large. Handle it in writeLoop.
	w.currentCommand = cmd
	return nil
	}
	if w.buf == nil {
	w.buf = make([]byte, 0, w.chunkSize)
	}
	if len(w.buf)+len(v.p) <= w.chunkSize {
	// We have not started sending yet, and we can stage all data without
	// starting a send. Compare against w.chunkSize instead of
	// w.writeQuantum: that way we can perform a oneshot upload for objects
	// which fit in one chunk, even though we will cut the request into
	// w.writeQuantum units when we do start sending.
	origLen := len(w.buf)
	w.buf = w.buf[:origLen+len(v.p)]
	copy(w.buf[origLen:], v.p)
	close(v.done)
	} else {
	// Too large. Handle it in writeLoop.
	w.currentCommand = cmd
	return nil
	}

[gemini-code-assist]

Pre-allocating the buffer here increases memory overhead for every writer instance. Consider keeping the lazy allocation pattern (initializing buf to nil and storing chunkSize in the struct) to improve memory efficiency, especially when many writers are opened or for small uploads.

Suggested change

	buf: make([]byte, 0, chunkSize),
	buf: nil,
	chunkSize: chunkSize,

[gemini-code-assist]

Removing the chunkSize field forces the implementation to rely on cap(w.buf) for chunking logic. Keeping this field is necessary to support lazy buffer allocation and provides a more robust way to track the intended chunk size.

Suggested change

	buf []byte
	buf []byte
	chunkSize int