On Thu, Sep 15, 2005 at 10:12:33PM +1000, Douglas Gilbert wrote: > Christoph Hellwig wrote: > > The upper layer doesn't send these down since 2.4.x (or 2.6 in > > practice), so no need to handle it. Inline sym_setup_data_pointers > > into it's only caller so we can fail gracefully in the case we'd get > > one neverless. > > Christoph, > I have no objection to this patch. > > I would like to flag that at some stage in the > future the SCSI (or whatever it may be called) > subsystem will need to support true bidirectional > commands (there are a few of these in SBC-2 and > lots of them in OSD). To do it properly I think we > need two scatter gather lists, one for data out > (to the device) and one for data in (from the device). > The SBC-2 XDWRITEREAD command indicates that the data out > is done before the data in (from the device); I'm not > sure whether that ordering is always the case. Yes, we will need to support truely bidirectional commands one day, but hopefully never on SPI HBAs ;-) At the same point we'll probably need to support > 16 byte CDBs aswell, and we'll need to switch to dynamically allocating the actual ->cmnd buffer (maybe only if it's > 12 bytes?). Sounds like a nice research projects for someone's Bacchelors thesis if anyone out there is looking for something like that ;-) - : send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
