[RFC,1/2] USB: OHCI: prepare to make ohci-hcd a library module

From: Manjunath Goudar <manjunath.goudar@linaro.org>

This patch prepares ohci-hcd for being split up into a core
library and separate platform driver modules.  A generic
ohci_hc_driver structure is created, containing all the "standard"
values, and a new mechanism is added whereby a driver module can
specify a set of overrides to those values.  In addition the
ohci_init(),ohci_restart(),ohci_hcd_init(),ohci_run() ohci_stop(),
ohci_suspend() and ohci_resume() routines need to be EXPORTed
for use by the drivers.

Signed-off-by: Manjunath Goudar <manjunath.goudar@linaro.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Greg KH <greg@kroah.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: linux-usb@vger.kernel.org
---
 drivers/usb/host/ohci-hcd.c |   87 ++++++++++++++++++++++++++++++++++++-------
 drivers/usb/host/ohci-mem.c |    3 +-
 drivers/usb/host/ohci.h     |   39 +++++++++++++++++++
 3 files changed, 115 insertions(+), 14 deletions(-)

On Tuesday 16 April 2013, manjunath.goudar@linaro.org wrote:
> From: Manjunath Goudar <manjunath.goudar@linaro.org>
> 
> This patch prepares ohci-hcd for being split up into a core
> library and separate platform driver modules.  A generic
> ohci_hc_driver structure is created, containing all the "standard"
> values, and a new mechanism is added whereby a driver module can
> specify a set of overrides to those values.  In addition the
> ohci_init(),ohci_restart(),ohci_hcd_init(),ohci_run() ohci_stop(),
> ohci_suspend() and ohci_resume() routines need to be EXPORTed
> for use by the drivers.
> 
> Signed-off-by: Manjunath Goudar <manjunath.goudar@linaro.org>
> Cc: Arnd Bergmann <arnd@arndb.de>
> Cc: Greg KH <greg@kroah.com>
> Cc: Alan Stern <stern@rowland.harvard.edu>
> Cc: linux-usb@vger.kernel.org

Hi Manjunath,

This looks nice I think, but let's see what Alan has to say about it.

I did not see any information in your description about how you
tested the changes. Are you able to use any USB-1.1 devices with
this on a PC, using ohci-pci as a loadable module?

> +
> +#if defined(CONFIG_PM) || defined(CONFIG_PCI)
> +extern int ohci_restart (struct ohci_hcd *ohci);
> +#else
> +static inline int ohci_restart (struct ohci_hcd *ohci)
> +{
> +	return;
> +}
> +#endif

I looked at who calls ohci_restart, and I don't think the #else clause
is actually needed here, or the "return" for that matter.

Just declare the function unconditionally.

I would also remove the space after the ohci_restart identifier, since
that is the common coding style.

	Arnd

On 17 April 2013 17:31, Arnd Bergmann <arnd@arndb.de> wrote:

> On Tuesday 16 April 2013, manjunath.goudar@linaro.org wrote:
> > From: Manjunath Goudar <manjunath.goudar@linaro.org>
> >
> > This patch prepares ohci-hcd for being split up into a core
> > library and separate platform driver modules.  A generic
> > ohci_hc_driver structure is created, containing all the "standard"
> > values, and a new mechanism is added whereby a driver module can
> > specify a set of overrides to those values.  In addition the
> > ohci_init(),ohci_restart(),ohci_hcd_init(),ohci_run() ohci_stop(),
> > ohci_suspend() and ohci_resume() routines need to be EXPORTed
> > for use by the drivers.
> >
> > Signed-off-by: Manjunath Goudar <manjunath.goudar@linaro.org>
> > Cc: Arnd Bergmann <arnd@arndb.de>
> > Cc: Greg KH <greg@kroah.com>
> > Cc: Alan Stern <stern@rowland.harvard.edu>
> > Cc: linux-usb@vger.kernel.org
>
> Hi Manjunath,
>
> This looks nice I think, but let's see what Alan has to say about it.
>
> I did not see any information in your description about how you
> tested the changes. Are you able to use any USB-1.1 devices with
> this on a PC, using ohci-pci as a loadable module?
>
> I did not test, I applied only Alan EHCI concept here.

> > +
> > +#if defined(CONFIG_PM) || defined(CONFIG_PCI)
> > +extern int ohci_restart (struct ohci_hcd *ohci);
> > +#else
> > +static inline int ohci_restart (struct ohci_hcd *ohci)
> > +{
> > +     return;
> > +}
> > +#endif
>
> I looked at who calls ohci_restart, and I don't think the #else clause
> is actually needed here, or the "return" for that matter.
>
> Just declare the function unconditionally.
>
> I would also remove the space after the ohci_restart identifier, since
> that is the common coding style.
>
>
I think it is not necessary that's why I will remove in next version.

Manjunath Goudar

On Tue, 16 Apr 2013 manjunath.goudar@linaro.org wrote:

> From: Manjunath Goudar <manjunath.goudar@linaro.org>
> 
> This patch prepares ohci-hcd for being split up into a core
> library and separate platform driver modules.  A generic
> ohci_hc_driver structure is created, containing all the "standard"
> values, and a new mechanism is added whereby a driver module can
> specify a set of overrides to those values.  In addition the
> ohci_init(),ohci_restart(),ohci_hcd_init(),ohci_run() ohci_stop(),
> ohci_suspend() and ohci_resume() routines need to be EXPORTed
> for use by the drivers.

This is a good start, but it goes too far by exporting routines that
ought to be private.  ohci_hcd_init, ohci_run, and ohci_stop should not
be called by bus glue modules:

	ohci_hcd_init doesn't need to be called by glue modules at all,
	since ohci_init already calls it.

	Glue modules should not override ohci_hc_driver.start.  Stuff 
	like the code at the start of ohci_pci_start can be moved to
	the end of the module's reset routine.	

	Likewise, glue modules should not need to call ohci_stop.
	If ohci_run fails, it should call ohci_stop itself.

Lastly, I agree with Arnd that sb800_prefetch should be moved to the 
pci-quirks.c file.

Alan Stern

On 19 April 2013 00:16, Alan Stern <stern@rowland.harvard.edu> wrote:

> On Tue, 16 Apr 2013 manjunath.goudar@linaro.org wrote:
>
> > From: Manjunath Goudar <manjunath.goudar@linaro.org>
> >
> > This patch prepares ohci-hcd for being split up into a core
> > library and separate platform driver modules.  A generic
> > ohci_hc_driver structure is created, containing all the "standard"
> > values, and a new mechanism is added whereby a driver module can
> > specify a set of overrides to those values.  In addition the
> > ohci_init(),ohci_restart(),ohci_hcd_init(),ohci_run() ohci_stop(),
> > ohci_suspend() and ohci_resume() routines need to be EXPORTed
> > for use by the drivers.
>
> This is a good start, but it goes too far by exporting routines that
> ought to be private.  ohci_hcd_init, ohci_run, and ohci_stop should not
> be called by bus glue modules:
>

Yes I agree,even I made  ohci_hcd_init, ohci_run, and ohci_stop as static
function.

for ohci controller initialization written generic function as ohci_setup.

int ohci_setup(struct usb_hcd *hcd)
{
        struct ohci_hcd *ohci = hcd_to_ohci(hcd);
        int retval;

        ohci->sbrn = HCD_USB11;

        /* data structure init */
        retval = ohci_init(ohci);
        if (retval)
                return retval;
        ohci_usb_reset(ohci);
        return 0;
}


>         ohci_hcd_init doesn't need to be called by glue modules at all,
>         since ohci_init already calls it.
>
> In all ohci blus glue called ohci_hcd_init in probe function,I am planing
to remove ohci_hcd_init function from ohci.mem.c file
 and moving the below statements to ohci_init function.this is my approach.
        ohci->next_statechange = jiffies;
        spin_lock_init (&ohci->lock);
        INIT_LIST_HEAD (&ohci->pending);

What about your suggestion.

        Glue modules should not override ohci_hc_driver.start.  Stuff
>         like the code at the start of ohci_pci_start can be moved to
>         the end of the module's reset routine.
>

I agree and now I am not  override the "start" function.

        Likewise, glue modules should not need to call ohci_stop.
>         If ohci_run fails, it should call ohci_stop itself.
>
> Lastly, I agree with Arnd that sb800_prefetch should be moved to the
> pci-quirks.c file.
>


>
> Alan Stern
>
>