RE: [PATCH 1/5] cifsd: add server handler and tranport layers

["Namjae Jeon" <namjae.jeon@xxxxxxxxxxx>]

> On Mon, Mar 22, 2021 at 02:13:40PM +0900, Namjae Jeon wrote:
> > +#define RESPONSE_BUF(w)		((void *)(w)->response_buf)
> > +#define REQUEST_BUF(w)		((void *)(w)->request_buf)
> 
> Why do you do this obfuscation?
I don't remember exactly, but back then, It looked easier...
> 
> > +#define RESPONSE_BUF_NEXT(w)	\
> > +	((void *)((w)->response_buf + (w)->next_smb2_rsp_hdr_off))
> > +#define REQUEST_BUF_NEXT(w)	\
> > +	((void *)((w)->request_buf + (w)->next_smb2_rcv_hdr_off))
> 
> These obfuscations aren't even used; delete them
They are used in many place.
./smb2pdu.c:            *rsp = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:            err_rsp = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:            rsp_hdr = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:    struct smb2_hdr *hdr = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:    struct smb2_hdr *rsp = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:    rsp_hdr = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:            rsp = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:            rsp = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:            hdr = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:            hdr = RESPONSE_BUF_NEXT(work);
./smb2pdu.c:            rsp = RESPONSE_BUF_NEXT(work);

./smb2pdu.c:            *req = REQUEST_BUF_NEXT(work);
./smb2pdu.c:            rcv_hdr = REQUEST_BUF_NEXT(work);
./smb2pdu.c:    struct smb2_hdr *req_hdr = REQUEST_BUF_NEXT(work);
./smb2pdu.c:    struct smb2_hdr *req = REQUEST_BUF_NEXT(work);
./smb2pdu.c:    rcv_hdr = REQUEST_BUF_NEXT(work);
./smb2pdu.c:    hdr = REQUEST_BUF_NEXT(work);
./smb2pdu.c:            req = REQUEST_BUF_NEXT(work);
./smb2pdu.c:            req = REQUEST_BUF_NEXT(work);
./smb2pdu.c:            hdr = REQUEST_BUF_NEXT(work);
./smb2pdu.c:    req_hdr = REQUEST_BUF_NEXT(work);
./smb2pdu.c:            hdr = REQUEST_BUF_NEXT(work);
./smb2pdu.c:    req_hdr = REQUEST_BUF_NEXT(work);
> 
> > +#define RESPONSE_SZ(w)		((w)->response_sz)
> > +
> > +#define INIT_AUX_PAYLOAD(w)	((w)->aux_payload_buf = NULL)
> > +#define HAS_AUX_PAYLOAD(w)	((w)->aux_payload_sz != 0)
> 
> I mean, do you really find it clearer to write:
> 
> 	if (HAS_AUX_PAYLOAD(work))
> than
> 	if (work->aux_payload_sz)
> 
> The unobfuscated version is actually shorter!
Yep, looks better, Will fix it.

Thanks for your review!