> I think this is right. However we may have the same issue already, > w/o my patches. In i2c_del_adapter() the following isn't protected: > device_for_each_child(&adap->dev, NULL, __unregister_client); The drivers have been removed already at that time. By a block of code which got removed in your patch: 1761 mutex_lock(&core_lock); 1762 bus_for_each_drv(&i2c_bus_type, NULL, adap, 1763 __process_removed_adapter); 1764 mutex_unlock(&core_lock);
Attachment:
signature.asc
Description: PGP signature
- References:
- [PATCH v2 0/4] i2c: Replace lists of special clients with flagging of such clients
- From: Heiner Kallweit
- [PATCH v2 2/4] i2c: Replace list-based mechanism for handling auto-detected clients
- From: Heiner Kallweit
- Re: [PATCH v2 2/4] i2c: Replace list-based mechanism for handling auto-detected clients
- From: Wolfram Sang
- Re: [PATCH v2 2/4] i2c: Replace list-based mechanism for handling auto-detected clients
- From: Heiner Kallweit
- [PATCH v2 0/4] i2c: Replace lists of special clients with flagging of such clients
- Prev by Date: Re: [PATCH v2 2/4] i2c: Replace list-based mechanism for handling auto-detected clients
- Next by Date: Re: [PATCH v2 2/4] i2c: Replace list-based mechanism for handling auto-detected clients
- Previous by thread: Re: [PATCH v2 2/4] i2c: Replace list-based mechanism for handling auto-detected clients
- Next by thread: Re: [PATCH v2 2/4] i2c: Replace list-based mechanism for handling auto-detected clients
- Index(es):