[U-Boot] [PATCH 3/3] usb: musb-new: omap2430: Fix compilation warning with USB_MUSB_GADGET

[Marek Vasut]

On 5/29/19 1:44 PM, Derald Woods wrote:
> 
> 
> On Wed, 29 May 2019, 05:07 Marek Vasut, <marek.vasut at gmail.com
> <mailto:marek.vasut at gmail.com>> wrote:
> 
>     On 5/29/19 3:34 AM, Derald Woods wrote:
>     > On Tue, May 28, 2019 at 7:49 PM Marek Vasut <marek.vasut at gmail.com
>     <mailto:marek.vasut at gmail.com>> wrote:
>     >>
>     >> On 5/29/19 2:00 AM, Derald Woods wrote:
>     >>> On Tue, May 28, 2019 at 6:27 PM Marek Vasut
>     <marek.vasut at gmail.com <mailto:marek.vasut at gmail.com>> wrote:
>     >>>>
>     >>>> On 5/29/19 1:16 AM, Derald Woods wrote:
>     >>>>> On Tue, May 28, 2019 at 4:16 PM Marek Vasut
>     <marek.vasut at gmail.com <mailto:marek.vasut at gmail.com>> wrote:
>     >>>>>>
>     >>>>>> On 5/28/19 4:22 AM, Derald D. Woods wrote:
>     >>>>>>> This commit addresses the following warning, when _NOT_
>     USB_MUSB_HOST:
>     >>>>>>>
>     >>>>>>> [...]
>     >>>>>>>   CC      drivers/usb/gadget/f_mass_storage.o
>     >>>>>>>   CC      drivers/usb/musb-new/omap2430.o
>     >>>>>>>   CC      drivers/usb/gadget/f_fastboot.o
>     >>>>>>>   CC      env/common.o
>     >>>>>>>   CC      env/env.o
>     >>>>>>>
>     /src/etinker/software/u-boot-master/drivers/usb/musb-new/omap2430.c:
>     In function ‘omap2430_musb_probe’:
>     >>>>>>>
>     /src/etinker/software/u-boot-master/drivers/usb/musb-new/omap2430.c:239:6:
>     warning: assignment to ‘int’ from ‘struct musb *’ makes integer from
>     pointer without a cast [-Wint-conversion]
>     >>>>>>>   ret = musb_register(&platdata->plat,
>     >>>>>>>       ^
>     >>>>>>>   LD      drivers/usb/host/built-in.o
>     >>>>>>>   CC      drivers/usb/gadget/f_sdp.o
>     >>>>>>>   CC      fs/ext4/ext4fs.o
>     >>>>>>> [...]
>     >>>>
>     >>>> Skip to the end first
>     >>>>
>     >>>>>>> Signed-off-by: Derald D. Woods <woods.technical at gmail.com
>     <mailto:woods.technical at gmail.com>>
>     >>>>>>> ---
>     >>>>>>>  drivers/usb/musb-new/omap2430.c | 8 ++++++--
>     >>>>>>>  1 file changed, 6 insertions(+), 2 deletions(-)
>     >>>>>>>
>     >>>>>>> diff --git a/drivers/usb/musb-new/omap2430.c
>     b/drivers/usb/musb-new/omap2430.c
>     >>>>>>> index 32743aa72c..cca1653f1e 100644
>     >>>>>>> --- a/drivers/usb/musb-new/omap2430.c
>     >>>>>>> +++ b/drivers/usb/musb-new/omap2430.c
>     >>>>>>> @@ -215,11 +215,13 @@ static int omap2430_musb_probe(struct
>     udevice *dev)
>     >>>>>>>  {
>     >>>>>>>  #ifdef CONFIG_USB_MUSB_HOST
>     >>>>>>>       struct musb_host_data *host = dev_get_priv(dev);
>     >>>>>>> +#else
>     >>>>>>> +     struct musb *musbp;
>     >>>>
>     >>>> Drop this hunk
>     >>>>
>     >>>>>>>  #endif
>     >>>>>>>       struct omap2430_musb_platdata *platdata =
>     dev_get_platdata(dev);
>     >>>>>>>       struct usb_bus_priv *priv = dev_get_uclass_priv(dev);
>     >>>>>>>       struct omap_musb_board_data *otg_board_data;
>     >>>>>>> -     int ret;
>     >>>>>>> +     int ret = 0;
>     >>>>>>>       void *base = dev_read_addr_ptr(dev);
>     >>>>>>>
>     >>>>>>>       priv->desc_before_addr = true;
>     >>>>>>> @@ -236,9 +238,11 @@ static int omap2430_musb_probe(struct
>     udevice *dev)
>     >>>>>>>
>     >>>>>>>       ret = musb_lowlevel_init(host);
>     >>>>>>>  #else
>     >>>>>>> -     ret = musb_register(&platdata->plat,
>     >>>>>>> +     musbp = musb_register(&platdata->plat,
>     >>>>
>     >>>> Replace musbp with host->host
>     >>>>
>     >>>>>>>                         (struct device *)otg_board_data,
>     >>>>>>>                         platdata->base);
>     >>>>>>> +     if (IS_ERR_OR_NULL(musbp))
>     >>>>
>     >>>> here too
>     >>>>
>     >>>>>>> +             return -EINVAL;
>     >>>>>>
>     >>>>>> For example the pic32 glue code holds the musb_host_data in
>     private data
>     >>>>>> , so it can call musb_stop() in .remove callback . Can you do
>     the same?
>     >>>>>>
>     >>>>>
>     >>>>> This was just a non-structural change to eliminate a compiler
>     warning.
>     >>>>> I agree that other things can/should be done with this driver.
>     That
>     >>>>> work would be outside of what I was trying to accomplsh here,
>     at the
>     >>>>> moment.
>     >>>>
>     >>>> See above, I believe that should fix the problem _and_ not
>     crash in the
>     >>>> .remove() .
>     >>>>
>     >>>
>     >>> As with 'ti-musb', there needs to be a clear distinction between
>     >>> 'host' and 'peripheral'/'gadget'. The issues here are much more
>     broad
>     >>> than the compile warning that I was trying to address. The required
>     >>> refactoring, for USB Gadgets, deserves a proper series for older
>     OMAP3
>     >>> devices.
>     >>>
>     >>> I would rather _DROP_ this patch than apply something that could
>     >>> result in a series of kludges.
>     >>
>     >> Does the following patch work for you ?
>     >>
>     >
>     > My use case does _not_ enable 'CONFIG_USB_MUSB_HOST'. The C
>     > preprocessor directives prevent 'host->host' from being included in
>     > compilation in that case.
> 
>     drivers/usb/musb-new/musb_uboot.h doesn't have any special stuff to
>     remove the struct musb_host_data { struct musb *musb ... } field .
>     What am I missing ?
> 
> 
> The code is guarded by CPP directives.

Can you point out the code you're talking about ? Which CPP directive do
you have in mind exactly?

> You are speaking of a more
> extensive refactoring/changing of the driver.

I am speaking of that diff I posted a few hours ago, nothing else.
This should fix this problem _and_ not fail in the remove path.
There is no refactoring necessary for that to work.

> There is a set of patches
> on the mailing list, by Adam Ford, that does just that. Or at least
> makes the proper start. I will look at those as a start. I am dropping
> my effort on this single compilation fix patch. So I will be looking at
> a different path.
OK

-- 
Best regards,
Marek Vasut