Re: RFC drivers/usb/storage/libusual

From: Pete Zaitcev
Date: Fri Sep 30 2005 - 19:27:58 EST


On Fri, 30 Sep 2005 00:09:05 -0700, Phil Dibowitz <phil@xxxxxxxx> wrote:

> A quick look over the patch shows that there are now two kinds of
> unusual_dev entries: unusual_dev() and unusual_dev_fl(), where the
> latter is for entries that don't need to specify SC or PR (i.e., just
> had US_SC_DEVICE, US_PR_DEVICE in them). While I think that's a
> reasonable change, it's not clear to me why that's useful to the rest of
> the patch, or it's just making unusual_devs.h artificially shorter?

Greg asked that too and I skipped his question because I needed an
explanation. So here it is now.

I thought that the match list for USB core and the shadow list that
usb-storage had were separate. So, libusual receives the first but
not the second, which remains in usb-storage. Once they are split
like that, I thought it unsafe to continue to use an index to
correlate them, so I added extra fields into the list of information
specific list to usb-storage, for matching. These fields use more
memory, so I thought it would cost nothing to move flags over
into driver_info. The resultant savings make usb-storage somewhat
shorter than it was before. It is shorter even when libusual is
configured off. This is because the second list is about twice
shorter now.

I can drop that part of the patch, I just thought it would be safer.
If you prefer that, I would suggest moving "normal" devices at the
bottom of the list into the unusual_devs.h, to make sure that
indexes stay the same. In such case, we won't need dropping commas.

-- Pete
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/