On Mon, Apr 14, 2025 at 11:56 AM Dmitry Antipov <dmantipov@xxxxxxxxx> wrote:
>
> On 4/14/25 12:22 PM, Arend van Spriel wrote:
>
> > When the platform_driver_probe() fails it means that brcmfmac_pdata will be NULL
>
> Hm.
>
> platform_driver_register()
> -> probe (which is brcmf_common_pd_probe())
> ...
> brcmfmac_pdata = dev_get_platdata(&pdev->dev);
> ...
> ...
> if (!bus_for_each_dev(&platform_bus_type, NULL, &drv->driver, is_bound_to_driver)) {
> retval = -ENODEV;
> platform_driver_unregister(drv); [1]
> }
>
> If we hit platform_driver_unregister() at [1], 'brcmfmac_pdata' is not NULL and,
> if something goes wrong in brcmf_core_init() next, platform_driver_unregister()
> may be called again.
>
> Shouldn't 'brcmf_common_pd_remove()' reset 'brcmfmac_data' back to NULL?
Hi Dmitry,
This one may have gone stale a bit. It showed up on my patchwork todo
list which we setup last week. I agree that reset brcmfmac_data to
NULL is needed.
Regards,
Arend
Attachment:
smime.p7s
Description: S/MIME Cryptographic Signature
