Re: [PATCH] USB / dwc3: remove the use of -ENOSYS from core.c

[Kushagra Verma]

On Fri, 2022-05-20 at 10:24 +0200, Greg KH wrote:
> On Fri, May 20, 2022 at 12:16:29PM +0530, Kushagra Verma wrote:
> > On Thu, 2022-05-19 at 17:59 +0200, Greg KH wrote:
> > > On Tue, May 17, 2022 at 08:22:00PM +0530, Kushagra Verma wrote:
> > > > This patch removes the use of -ENOSYS as it is used when users try
> > > > to call a
> > > > syscall that doesn't exist. So, we don't need to check if 'ret == -
> > > > ENOSYS'.
> > > 
> > > ENOSYS is also used internally in the kernel for other things.
> > > 
> > > > 
> > > > Signed-off-by: Kushagra Verma <kushagra765@xxxxxxxxxxx>
> > > > ---
> > > >  drivers/usb/dwc3/core.c | 4 ++--
> > > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > > > 
> > > > diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
> > > > index c78205c5e19f..3c1a877d5183 100644
> > > > --- a/drivers/usb/dwc3/core.c
> > > > +++ b/drivers/usb/dwc3/core.c
> > > > @@ -1283,7 +1283,7 @@ static int dwc3_core_get_phy(struct dwc3
> > > > *dwc)
> > > >         dwc->usb2_generic_phy = devm_phy_get(dev, "usb2-phy");
> > > >         if (IS_ERR(dwc->usb2_generic_phy)) {
> > > >                 ret = PTR_ERR(dwc->usb2_generic_phy);
> > > > -               if (ret == -ENOSYS || ret == -ENODEV)
> > > > +               if (ret == -ENODEV)
> > > 
> > > Did you validate that no callers can ever set this to ENOSYS?
> > Yes, I think no callers should ever set ret to ENOSYS unless it comes under the other things ENOSYS is used for.
> 
> Please audit the code to verify this and document it in the changelog
> text when you resend this.
I have audited the code and none of the callers return -ENOSYS. The devm_phy_get function calls phy_get to
lookup and obtain a reference to a phy. This function returns the phy driver or -ENODEV if there is no such
phy (according to the comments above it).
> 
> > > Why was this added in the first place?  What commit added it?
> > This was added by commit 57303488cd37d ("usb: dwc3: adapt dwc3 core to use Generic PHY Framework").
> 
> Great, please also include that information.
Ok.
> 
> thanks,
> 
> greg k-h

Thanks,

Kushagra