Re: [usb:usb-testing 25/42] drivers/usb/dwc3/dwc3-octeon.c:403:10: warning: shift count >= width of type

[Nathan Chancellor <nathan@kernel.org>]

Hi Ladislav,

On Fri, Jul 28, 2023 at 08:48:20AM +0200, Ladislav Michl wrote:
> Hi Greg,
> 
> I'm sorry for late reply (travelling)
> 
> On Wed, 2023-07-26 at 06:39 +0200, Greg Kroah-Hartman wrote:
> > On Wed, Jul 26, 2023 at 05:09:56AM +0800, kernel test robot wrote:
> > > Hi Ladislav,
> > > 
> > > First bad commit (maybe != root cause):
> > > 
> > > tree:  
> > > https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb.git usb-
> > > testing
> > > head:   ac47051f0bb8ccf1eaedb296a19fcb541474db53
> > > commit: b3061c5a15fe460502e9574b4663dd7c6ed80b48 [25/42] usb: dwc3:
> > > dwc3-octeon: Convert to glue driver
> > 
> > Ladislav, I've dropped your dwc3-octeon patches from my tree now,
> > please fix up and resend when you have this resolved.
> 
> As I didn't find a better place to ask, I'm asking here:
> all registers are 64bit and driver is now using GENMASK and BIT
> macros to access their values. Reported problem comes from 32bit
> configuration, where BIT is UL(1) << (nr) and BITS_PER_LONG is 32.
> Is using GENMASK_ULL and BIT_ULL preferred way to go here? I'm not a
> bit fan of "fixing" code just to silence warning on irrelevant
> configurations, but will do it unless someone points me to better way.

I am not Greg but I see you did not get a response on this yet. Yes, you
should use the _ULL macros to ensure the values are always 64-bit, which
will avoid this issue altogether.

Cheers,
Nathan