On Wed, Sep 25, 2013 at 03:54:39PM -0700, Andrew Morton wrote: > > +#define SETOFFSET(buf, ndsiz, offset, rec) \ > > + (*(u16 *)((u8 *)(buf) + (ndsiz) + (-2 * (rec))) = (cpu_to_be16(offset))) > > This is pretty ghastly. Can you take a look at reimplementing it in a C > function? Add any necessary comments to explain what's happening and I > expect you will find that it produces a much nicer result. Take a look at the callers - buf and ndsiz remain unchanged, rec runs from 1 to 4. IOW, what should be done is /* last 8 bytes contain offsets */ __be16 *offsets = (__be16 *)(buf + node_size); ... offset = sizeof(struct hfs_bnode_desc); *--offsets = cpu_to_be16(offset); ... offset += sizeof(struct hfs_btree_header_rec); *--offsets = cpu_to_be16(offset); offset += HFSPLUS_BTREE_HDR_USER_BYTES; *--offsets = cpu_to_be16(offset); ... offset += hdr_node_map_rec_bits / 8; *--offsets = cpu_to_be16(offset); and to hell with that macro. Speaking of another thing in the patch, there's that useful function called "memset"... -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
