On Tue, May 18, 2021 at 02:50:27PM +0200, Hans de Goede wrote:
> init_dell_smbios_wmi() only registers the dell_smbios_wmi_driver on systems
> where the Dell WMI interface is support. While exit_dell_smbios_wmi()
> unregisters it unconditionally, this leads to the following oops:
>
> [ 175.722921] ------------[ cut here ]------------
> [ 175.722925] Unexpected driver unregister!
> [ 175.722939] WARNING: CPU: 1 PID: 3630 at drivers/base/driver.c:194 driver_unregister+0x38/0x40
> ...
> [ 175.723089] Call Trace:
> [ 175.723094] cleanup_module+0x5/0xedd [dell_smbios]
> ...
> [ 175.723148] ---[ end trace 064c34e1ad49509d ]---
>
> Make the unregister happen on the same condition the register happens
> to fix this.
>
> Cc: Mario Limonciello <mario.limonciello@xxxxxxxxxxx>
> Fixes: 1a258e670434 ("platform/x86: dell-smbios-wmi: Add new WMI dispatcher driver")
> Signed-off-by: Hans de Goede <hdegoede@xxxxxxxxxx>
> ---
> drivers/platform/x86/dell/dell-smbios-wmi.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/platform/x86/dell/dell-smbios-wmi.c b/drivers/platform/x86/dell/dell-smbios-wmi.c
> index a1753485159c..33f823772733 100644
> --- a/drivers/platform/x86/dell/dell-smbios-wmi.c
> +++ b/drivers/platform/x86/dell/dell-smbios-wmi.c
> @@ -270,7 +270,8 @@ int init_dell_smbios_wmi(void)
>
> void exit_dell_smbios_wmi(void)
> {
> - wmi_driver_unregister(&dell_smbios_wmi_driver);
> + if (wmi_supported)
Something is fishy with this. wmi_supported is used in the
init_dell_smbios_wmi function so we shouldn't even be able to call this
function if wmi_supported == 0
is there a bug in the wmi framework where drivers that fail to init are not
removed from a list resulting in a failure to unregister? Is your fix hiding a
higher level issue?
Looking more shouldn't the prob function NOT get called if the init function
fails?
--mark
> + wmi_driver_unregister(&dell_smbios_wmi_driver);
> }
>
> MODULE_DEVICE_TABLE(wmi, dell_smbios_wmi_id_table);
> --
> 2.31.1
>
