Re: [PATCH v2 2/5] PCI: dwc: rcar-gen4: Rework rcar_gen4_pcie_drvdata .additional_common_init into .init

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Silent failure in `rcar_gen4_pcie_ep_pre_init()` and asymmetric teardown in endpoint error paths lead to kernel panics and unbalanced clock disables.
--

--- Patch [2]: [PATCH v2 2/5] PCI: dwc: rcar-gen4: Rework rcar_gen4_pcie_drvdata .additional_common_init into .init ---
commit f4863c849aaf3cf13744faee6e2f333b5420b025
Author: Marek Vasut <marek.vasut+renesas@mailbox.org>

PCI: dwc: rcar-gen4: Rework rcar_gen4_pcie_drvdata .additional_common_init into .init

This commit simplifies the PCIe initialization callchain for R-Car Gen4 SoCs.
It replaces the additional_common_init callback with a unified init callback
for both host and endpoint configurations.

> diff --git a/drivers/pci/controller/dwc/pcie-rcar-gen4.c b/drivers/pci/controller/dwc/pcie-rcar-gen4.c
> index 5f7211b91ee5b..5724f64eadfad 100644
> --- a/drivers/pci/controller/dwc/pcie-rcar-gen4.c
> +++ b/drivers/pci/controller/dwc/pcie-rcar-gen4.c

[ ... ]

> @@ -487,7 +509,7 @@ static void rcar_gen4_pcie_ep_pre_init(struct dw_pcie_ep *ep)
>  	struct rcar_gen4_pcie *rcar = to_rcar_gen4_pcie(dw);
>  	int ret;
>  
> -	ret = rcar_gen4_pcie_common_init(rcar);
> +	ret = rcar->drvdata->init(rcar);
>  	if (ret)
>  		return;

[Severity: High]
This is a pre-existing issue, but since this area is being modified, does the
early return here lead to a kernel panic if initialization fails?

If rcar->drvdata->init() fails, it returns early but cannot propagate the
error because pre_init() returns void. Execution then continues into
dw_pcie_ep_init_registers(), which accesses unclocked DWC registers, resulting
in an SError panic.

Also, if dw_pcie_ep_init() fails before calling ep->ops->pre_init() (such as
in devm_pci_epc_create()), the error path in rcar_gen4_add_dw_pcie_ep()
unconditionally calls rcar_gen4_pcie_ep_deinit(). That function performs an
MMIO write to PCIEDMAINTSTSEN while clocks are disabled, causing another
SError panic, and warns about unbalanced clock disables.

Could these endpoint error paths be reworked so the kernel fails gracefully
instead of crashing?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260720020505.315335-1-marek.vasut+renesas@mailbox.org?part=2