HI, All
Please stop review of this series. It's a pity of your time.
Probably the design of USBSSP device controller will be significantly changed and simplified, so
I will have to create new driver.
The planed change include:
- simplification and change registers map
- removing command queue
- simplification in input/output endpoint context
- and other.
It was not my decision. Thanks for all your comments.
I will take them into account in new driver.
Cheers,
Pawell
>On Tue, Sep 11, 2018 at 08:48:43AM +0300, Felipe Balbi wrote:
>>
>> Hi,
>>
>> Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> writes:
>> > On Thu, Jul 19, 2018 at 06:57:35PM +0100, Pawel Laszczak wrote:
>> >> This patch add additional functions that converts some fields to string.
>> >>
>> >> For example function usbssp_trb_comp_code_string take completion
>> >> code value and return string describing completion code.
>> >>
>> >> Signed-off-by: Pawel Laszczak <pawell@xxxxxxxxxxx>
>> >> ---
>> >> drivers/usb/usbssp/gadget.h | 580 ++++++++++++++++++++++++++++++++++++
>> >> 1 file changed, 580 insertions(+)
>> >>
>> >> diff --git a/drivers/usb/usbssp/gadget.h b/drivers/usb/usbssp/gadget.h
>> >> index 49e7271187cc..b5c17603af78 100644
>> >> --- a/drivers/usb/usbssp/gadget.h
>> >> +++ b/drivers/usb/usbssp/gadget.h
>> >> @@ -930,6 +930,73 @@ struct usbssp_transfer_event {
>> >> #define COMP_UNDEFINED_ERROR 33
>> >> #define COMP_INVALID_STREAM_ID_ERROR 34
>> >>
>> >> +static inline const char *usbssp_trb_comp_code_string(u8 status)
>> >
>> > <snip>
>> >
>> >
>> > You have _giant_ inline functions here, why?
>> >
>> > Please just put this all in a .c file and let the linker properly handle
>> > things. You do not want to duplicate all of these crazy strings all
>> > over the place where ever you call these functions.
>> >
>> > And I am guessing this is only for some sort of "debugging" mode? If
>> > so, shouldn't there be a way to not even build this in? Some systems
>> > are very space constrained...
>>
>> many of them seem to be a straight copy from xhci.
>
>Which doesn't mean it's a great model to copy :)
>
>Let's learn from our past mistakes, having had to try to slim down a
>kernel for a limited memory system is a chore that no one should have to
>manually hack up the source tree just to accomplish it.
>
>thanks,
>
>greg k-h
