Re: [GIT PULL]: Generic phy updates for v5.14

[Greg KH]

On Wed, Jun 23, 2021 at 09:29:06AM +0200, Sergio Paracuellos wrote:
> Hi Greg,
> 
> On Wed, Jun 23, 2021 at 9:21 AM Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> >
> > On Wed, Jun 23, 2021 at 11:45:59AM +0530, Vinod Koul wrote:
> > > Hello Greg,
> > >
> > > Please pull to receive Generic phy subsystem updates for v5.14
> > >
> > > The following changes since commit 6efb943b8616ec53a5e444193dccf1af9ad627b5:
> > >
> > >   Linux 5.13-rc1 (2021-05-09 14:17:44 -0700)
> > >
> > > are available in the Git repository at:
> > >
> > >   git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy.git tags/phy-for-5.14
> > >
> > > for you to fetch changes up to f7eedcb8539ddcbb6fe7791f1b4ccf43f905c72f:
> > >
> > >   phy: ti: dm816x: Fix the error handling path in 'dm816x_usb_phy_probe() (2021-06-21 12:04:18 +0530)
> > >
> > > ----------------------------------------------------------------
> > > phy-for-5.14
> > >
> > >   - Updates:
> > >         - Yaml conversion for renesas,rcar-gen3 pcie phy and
> > >         rockchip-usb-phy bindings
> > >         - Support for devm_phy_get() taking NULL phy name
> > >
> > >   - New support:
> > >       - PCIe phy for Qualcomm IPQ60xx
> > >       - PCIe phy for Qualcomm SDX55
> > >       - USB phy for RK3308
> > >       - CAN transceivers phy for TI TCAN104x
> > >         - Innosilicon-based CSI dphy for rockchip
> >
> > Why is PHY_MT7621_PCI only y/n?  Shouldn't this be able to be built as a
> > module?
> >
> > The USB phy enables this driver, which now forces me to build it into my
> > kernel, which does not seem like a wise idea.  I'm not sure which commit
> > in this series does this, but that isn't a good thing.
> >
> > Ah, it's 6eded551cefe ("phy: ralink: Kconfig: convert mt7621-pci-phy
> > into 'bool'"), why is that needed?  We are working to turn more code
> > into modules, not force them to be built in, this feels like the wrong
> > way to go :(
> 
> This was turned into bool to automatically be included if CONFIG_PCI_MT7621
> is set. This should be the only requirement to add this phy driver for
> real hardware since it has nothing to do if there is no
> CONFIG_PCI_MT7621. That's why Kconfig is now:
> 
> depends on (RALINK && OF && PCI_MT7621) || COMPILE_TEST
> Am I missing something here???

It is forcing it to be built into the kernel, why can it not just be a
module?  You took that ability away here by making the value a 'bool'.

thanks,

greg k-h