Re: [PATCH v3 09/13] dmaengine: dw-edma-pcie: Add platform ops to match data

[sashiko-bot@kernel.org]

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

Pre-existing issues:
- [High] Double free of PCI IRQ vectors on device removal.
--

commit ea9357c364d1768529dfa87f1f7bc66276215c93
Author: Koichiro Den <den@valinux.co.jp>

dmaengine: dw-edma-pcie: Add platform ops to match data

This commit moves the platform operations pointer into the match data
structure. It updates existing matches to use the standard platform ops
pointer without introducing any functional changes.

> diff --git a/drivers/dma/dw-edma/dw-edma-pcie.c b/drivers/dma/dw-edma/dw-edma-pcie.c
> index 5249324ad6bff..96038aaca0791 100644
> --- a/drivers/dma/dw-edma/dw-edma-pcie.c
> +++ b/drivers/dma/dw-edma/dw-edma-pcie.c
[ ... ]
> @@ -597,17 +598,20 @@ static void dw_edma_pcie_remove(struct pci_dev *pdev)

[Severity: High]
This isn't a bug introduced by this patch, but looking at
dw_edma_pcie_remove(), is there a double-free risk with the IRQ vectors?

In dw_edma_pcie_probe(), the device is enabled using pcim_enable_device(),
which means IRQ vectors allocated by pci_alloc_irq_vectors() are
automatically managed and freed by devres during device teardown.

However, dw_edma_pcie_remove() explicitly frees them again:

static void dw_edma_pcie_remove(struct pci_dev *pdev)
{
        struct dw_edma_chip *chip = pci_get_drvdata(pdev);
        ...
        /* Freeing IRQs */
        pci_free_irq_vectors(pdev);
}

Does this manual call to pci_free_irq_vectors() on a devres-managed device
lead to a double-free when pcim_msi_release() is later invoked?

>  
>  static const struct dw_edma_pcie_match_data snps_edda_match_data = {
>  	.data = &snps_edda_data,
> +	.plat_ops = &dw_edma_pcie_plat_ops,
>  	.parse_caps = dw_edma_pcie_parse_synopsys_caps,
>  };

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260620170040.3756043-1-den@valinux.co.jp?part=9