On Thu, 3 Sep 2015 17:33:16 +0100
Stefan Hajnoczi <stefa...@gmail.com> wrote:
> On Mon, Aug 17, 2015 at 10:09:34AM +0200, Marc Marí wrote:
> > +static const struct {
> > + const char *format_name;
> > + const char *protocol_name;
> > + const char *library_name;
> > + bool has_probe;
> > + bool has_probe_device;
> > +} block_driver_module[] = {
>
> Why is this list incomplete? It doesn't cover all block drivers.
> Perhaps these are the only modular block drivers.
I think we can decide on a protocol first (these patches), and then
apply the changes to all (possible) drivers. At least, that was what I
had in mind.
> Also, it ignores CONFIG_CURL and friends. Perhaps it doesn't matter
> because the module loading code will just see that there is no file
> there, but maybe conditional compilation should be used?
It's true that this patch doesn't look at the CONFIG options. But the
next one does (it takes block-obj-m from the Makefile), and also
replaces this file.
Thanks
Marc
> A plural name would more consistent (i.e. you deleted the
> plural block_modules[] variable and introduced a singular
> block_driver_module[] variable).
