Re: [PATCH V2 2/6] usb: core: Add ability to skip phy exit on suspend and init on resume

[Alan Cooper]

On Fri, Nov 9, 2018 at 3:08 AM Chunfeng Yun <chunfeng.yun@xxxxxxxxxxxx> wrote:
>
> On Thu, 2018-11-08 at 17:50 -0500, Alan Cooper wrote:
> > On Wed, Nov 7, 2018 at 8:43 PM Chunfeng Yun <chunfeng.yun@xxxxxxxxxxxx> wrote:
> > >
> > > hi,
> > > On Tue, 2018-10-30 at 18:30 -0400, Alan Cooper wrote:
> > > > On 10/17/18 9:46 PM, Chunfeng Yun wrote:> hi,
> > > > >
> > > > > On Wed, 2018-10-17 at 18:29 -0400, Al Cooper wrote:
> > > > >> Add the ability to skip calling the PHY's exit routine on suspend
> > > > >> and the PHY's init routine on resume. This is to handle a USB PHY
> > > > >> that should have it's power_off function called on suspend but
> > > > cannot
> > > > >> have it's exit function called because on exit it will disable the
> > > > >> PHY to the point where register accesses to the Host Controllers
> > > > >> using the PHY will be disabled and the host drivers will crash.
> > > > >>
> > > > >> This is enabled with the HCD flag "suspend_without_phy_exit" which
> > > > >> can be set from any HCD driver.
> > > > >>
> > > > >> Signed-off-by: Al Cooper<alcooperx@xxxxxxxxx>
> > > > >> ---
> > > > >>   drivers/usb/core/hcd.c  |  8 ++++----
> > > > >>   drivers/usb/core/phy.c  | 18 ++++++++++++------
> > > > >>   drivers/usb/core/phy.h  |  9 ++++++---
> > > > >>   include/linux/usb/hcd.h |  3 +++
> > > > >>   4 files changed, 25 insertions(+), 13 deletions(-)
> > > > >>
> > >
> > >
> > > > >>       unsigned        skip_phy_initialization:1;
> > > > >>
> > > > >> +    /* Some phys don't want the phy's exit/init called on
> > > > suspend/resume */
> > > > >> +    unsigned        suspend_without_phy_exit:1;
> > > > > As suggested before, you can skip phy's exit/init during
> > > > suspend/resume
> > > > > by enabling wakeup of hcd, so needn't add a new variable for it.
> > > >
> > > > I still need to be able to enable and disable wakeup for this driver.
> > > Just use device_init_wakeup(dev, true) instead of
> > > device_wakeup_enable(dev) for your controller driver, you can try it.
> > >
> > > Sorry for the late replay
> >
> > What about the case where sysfs is used to disable wakeup?
> Oh, It's a problem in this case.
>
> BTW: Why the host driver will crash when access host controller's
> register after power-off phy? they are usually independent.

Unfortunately, they are not independent in our design. The PHY
supplies the clock needed to read/write the OHCI, EHCI and XHCI
registers.

Al

>
> >
> > Al
> > > >
> > > > >
> > > > >> +
> > > > >>       /* The next flag is a stopgap, to be removed when all the
> > > > HCDs
> > > > >>        * support the new root-hub polling mechanism. */
> > > > >>       unsigned        uses_new_polling:1;
> > > > >
> > >
> > >
>
>