On Fri, Oct 06, 2023 at 07:43:12PM +0300, Andy Shevchenko wrote:
> The code to create the child platform device is essentially the same as
> what platform_device_register_full() does, so change over to use
> that same function to reduce duplication.
>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
> ---
> v2: (hopefully) fixed run-time NULL-dereference (LKP)
> drivers/usb/usbip/vhci_hcd.c | 55 +++++++++++++-----------------------
> 1 file changed, 20 insertions(+), 35 deletions(-)
>
> diff --git a/drivers/usb/usbip/vhci_hcd.c b/drivers/usb/usbip/vhci_hcd.c
> index 37d1fc34e8a5..f845b91848b9 100644
> --- a/drivers/usb/usbip/vhci_hcd.c
> +++ b/drivers/usb/usbip/vhci_hcd.c
> - for (i = 0; i < vhci_num_controllers; i++) {
> - void *vhci = &vhcis[i];
The void *vhci here is intended
platform_device_add_data "Add a copy of platform specific data to the
platform device's platform_data pointer"
however, vhcis is static so it is not intended to be copied (the
pdev inside struct vhci might be the cause)
> - ret = platform_device_add_data(vhcis[i].pdev, &vhci, sizeof(void *));
> - if (ret)
> - goto err_driver_register;
> - }
> -
> ret = platform_driver_register(&vhci_driver);
> if (ret)
> goto err_driver_register;
>
> for (i = 0; i < vhci_num_controllers; i++) {
> - ret = platform_device_add(vhcis[i].pdev);
> + struct platform_device_info pdevinfo = {
> + .name = driver_name,
> + .id = i,
> + .data = &vhcis[i],
here should be a &vhci
> + .size_data = sizeof(void *),
> + };
> +
> + vhcis[i].pdev = platform_device_register_full(&pdevinfo);
> + ret = PTR_ERR_OR_ZERO(vhcis[i].pdev);
> }
> --
> 2.40.0.1.gaa8946217a0b
I have reproduced the boot error reported by syzbot in
https://lore.kernel.org/linux-usb/00000000000029242706077f3145@xxxxxxxxxx/
https://lore.kernel.org/linux-usb/0000000000007634c1060793197c@xxxxxxxxxx/
and adding back void *vhci fixed it
I'll send a patch.
