Re: [PATCH v2] PCI: eswin: Fix resource leak on dw_pcie_host_init() failure

[Felix Gu]

On Thu, May 21, 2026 at 3:08 PM zhangsenchuan
<zhangsenchuan@xxxxxxxxxxxxxxxxxx> wrote:
> > --- a/drivers/pci/controller/dwc/pcie-eswin.c
> > +++ b/drivers/pci/controller/dwc/pcie-eswin.c
> > @@ -182,7 +182,7 @@ static int eswin_pcie_host_init(struct dw_pcie_rp *pp)
> >  {
> >       struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
> >       struct eswin_pcie *pcie = to_eswin_pcie(pci);
> > -     struct eswin_pcie_port *port, *tmp;
> > +     struct eswin_pcie_port *port;
> >       u32 val;
> >       int ret;
> >
> > @@ -252,10 +252,6 @@ static int eswin_pcie_host_init(struct dw_pcie_rp *pp)
> >       reset_control_bulk_assert(ESWIN_NUM_RSTS, pcie->resets);
> >  err_deassert:
> >       clk_bulk_disable_unprepare(pcie->num_clks, pcie->clks);
> > -     list_for_each_entry_safe(port, tmp, &pcie->ports, list) {
> > -             reset_control_put(port->perst);
> > -             list_del(&port->list);
> > -     }
>
> Hi, Felix
>
> If removed, in the dw_pcie_resume_noirq function, if the execution of
> "pci->pp.ops->init(&pci->pp)" fails, the resources will not be released.
>

Hi Senchuan,
Do we really need to release the port related resources in
dw_pcie_resume_noirq()
when pci->pp.ops->init(&pci->pp) fails?
And why not release the port related resources in the following error paths of
dw_pcie_resume_noirq()?

Best regards,
Felix
> Kind regards,
> Senchuan