[PATCH 0/5] usb: dwc3: xilinx: error handling and teardown fixes

[gregkh at linuxfoundation.org]

On Wed, Aug 26, 2026 at 12:22:38AM +0000, Thinh Nguyen wrote:
> On Mon, Aug 10, 2026, Radhey Shyam Pandey wrote:
> > This series improves error handling, resource teardown, and system sleep
> > PM robustness in the dwc3-xilinx glue driver for ZynqMP and Versal
> > platforms.
> > 
> > Patch 1 switches probe to device_get_match_data() and adds the direct
> > property.h include instead of relying on of.h for that helper.
> > 
> > Patch 2 introduces struct dwc3_xlnx_platdata and moves ZynqMP and
> > Versal setup behind plat->init(). The struct also defines an exit
> > callback slot used by the later teardown patches.
> > 
> > Patch 3 fixes system suspend and resume PHY handling: suspend now
> > powers off the PHY before phy_exit() and propagates errors; resume
> > disables clocks if PHY reinitialization fails after they were enabled.
> > 
> > Patch 4 fixes ZynqMP init error paths. Reset handles are saved in
> > driver private data and fall-through error labels re-assert only the
> > resets already released before unwinding the PHY when init fails
> > partway through reset deassert or PHY setup.
> > 
> > Patch 5 adds dwc3_xlnx_exit_zynqmp(), wires it as the ZynqMP platform
> > exit callback, and calls plat->exit() from probe error paths after a
> > successful plat->init() and from remove()/shutdown.
> > 
> > Radhey Shyam Pandey (5):
> >   usb: dwc3: xilinx: use device_get_match_data() in probe
> >   usb: dwc3: xilinx: add platform data struct with init callback
> >   usb: dwc3: xilinx: fix system suspend and resume PHY handling
> >   usb: dwc3: xilinx: re-assert resets on ZynqMP init error paths
> >   usb: dwc3: xilinx: unwind ZynqMP platform init on probe failure and
> >     remove
> > 
> >  drivers/usb/dwc3/dwc3-xilinx.c | 132 +++++++++++++++++++++++++--------
> >  1 file changed, 100 insertions(+), 32 deletions(-)
> > 
> > 
> > base-commit: 6b8c8af514d739d0335f5579b585e02babe8a727
> > -- 
> > 2.43.0
> > 
> 
> These patches look good to me, do you plan to backport the fixes? If so,
> can you Cc stable and re-order them so that the fixes patches go first?
> Better yet, keep the fixes and new changes in separate series.

Yes, this should be 2 separate patch series, one for 7.3-final and one
for 7.4-rc1.  The -final series should go to stable trees as well.

thanks,

greg k-h