Re: [PATCHv2] PCI: xilinx: use fwnode_irq_get() for INTx IRQ lookup

[sashiko-bot@kernel.org]

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

Pre-existing issues:
- [High] The driver requests a shared IRQ before it masks hardware interrupts and before it allocates its IRQ domains, creating a race condition that can lead to an interrupt storm.
--

--- Patch [1]: [PATCHv2] PCI: xilinx: use fwnode_irq_get() for INTx IRQ lookup ---
commit 02b4ee51408c6b762dcc417c028ea4f662761599
Author: Rosen Penev <rosenp@gmail.com>

PCI: xilinx: use fwnode_irq_get() for INTx IRQ lookup

This commit replaces irq_of_parse_and_map() with fwnode_irq_get() for
looking up the INTx IRQ, checking the return value to properly propagate
errors such as -EPROBE_DEFER.

> diff --git a/drivers/pci/controller/pcie-xilinx.c b/drivers/pci/controller/pcie-xilinx.c
> --- a/drivers/pci/controller/pcie-xilinx.c
> +++ b/drivers/pci/controller/pcie-xilinx.c
> @@ -548,7 +548,10 @@ static int xilinx_pcie_parse_dt(struct xilinx_pcie *pcie)
>  	if (IS_ERR(pcie->reg_base))
>  		return PTR_ERR(pcie->reg_base);
>  
> -	irq = irq_of_parse_and_map(node, 0);
> +	irq = fwnode_irq_get(dev_fwnode(dev), 0);
> +	if (irq < 0)
> +		return irq;
> +
>  	err = devm_request_irq(dev, irq, xilinx_pcie_intr_handler,
>  			       IRQF_SHARED | IRQF_NO_THREAD,
>  			       "xilinx-pcie", pcie);

[Severity: High]
This is a pre-existing issue, but does requesting the shared IRQ here
create a race condition?

Looking at xilinx_pcie_probe(), xilinx_pcie_parse_dt() is called before
xilinx_pcie_init_port() masks the hardware interrupts, and before
xilinx_pcie_init_irq_domain() allocates the MSI and legacy IRQ domains.

If the shared IRQ fires before the domains are created,
xilinx_pcie_intr_handler() will execute with pcie->msi_domain and
pcie->leg_domain as NULL.

In that scenario, generic_handle_domain_irq() will return -EINVAL, leaving
the level-triggered interrupt unserviced. Because the handler returns
IRQ_HANDLED, could this immediately re-enter the handler and cause an
unhandled interrupt storm?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260716205811.800920-1-rosenp@gmail.com?part=1