Re: [PATCH] PCI: Fix accessing freed memory in pci_remove_resource_files

[Bjorn Helgaas <helgaas@kernel.org>]

Hi Danijel,

Thanks for the patch.

On Fri, May 07, 2021 at 06:27:06PM +0800, Danijel Slivka wrote:
> This patch fixes segmentation fault during accessing already freed
> pci device resource files, as after freeing res_attr and res_attr_wc
> elements, in pci_remove_resource_files function, they are left as
> dangling pointers.
> 
> Signed-off-by: Danijel Slivka <danijel.slivka@amd.com>
> ---
>  drivers/pci/pci-sysfs.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/pci/pci-sysfs.c b/drivers/pci/pci-sysfs.c
> index f8afd54ca3e1..bbdf6c57fcda 100644
> --- a/drivers/pci/pci-sysfs.c
> +++ b/drivers/pci/pci-sysfs.c
> @@ -1130,12 +1130,14 @@ static void pci_remove_resource_files(struct pci_dev *pdev)
>  		if (res_attr) {
>  			sysfs_remove_bin_file(&pdev->dev.kobj, res_attr);
>  			kfree(res_attr);
> +			pdev->res_attr[i] = NULL;
>  		}
>  
>  		res_attr = pdev->res_attr_wc[i];
>  		if (res_attr) {
>  			sysfs_remove_bin_file(&pdev->dev.kobj, res_attr);
>  			kfree(res_attr);
> +			pdev->res_attr_wc[i] = NULL;

If this patch fixes something, I would expect to see a test like this
somewhere:

  if (pdev->res_attr[i])
    pdev->res_attr[i]->size = 0;

But I don't see anything like that, so I can't figure out where we
actually use res_attr[i] or res_attr_wc[i], except in
pci_remove_resource_files() itself.

Did you actually see a segmentation fault?  If so, where?

>  		}
>  	}
>  }
> -- 
> 2.20.1
>