linux-kernel - Re: [PATCH 1/1] device core: Add flag to autoremove device link on supplier unbind

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <541c84e9-47fc-e7bc-8249-1477b9e63455@codeaurora.org>
Date:   Wed, 30 May 2018 18:21:02 +0530
From:   Vivek Gautam <vivek.gautam@...eaurora.org>
To:     "Rafael J. Wysocki" <rafael.j.wysocki@...el.com>
Cc:     gregkh@...uxfoundation.org, lukas@...ner.de,
        dmitry.torokhov@...il.com, aspriel@...il.com, robin.murphy@....com,
        linux-kernel@...r.kernel.org, linux-pm@...r.kernel.org
Subject: Re: [PATCH 1/1] device core: Add flag to autoremove device link on
 supplier unbind

Hi Rafael,


On 5/30/2018 4:29 PM, Rafael J. Wysocki wrote:
> On 5/30/2018 11:57 AM, Vivek Gautam wrote:
>> When using the device links without the consumers or
>> suppliers maintaining pointers to these links, a flag can
>> help in autoremoving the links on supplier driver unbind.
>> We remove these links only when the supplier's link to its
>> consumers has gone in DL_STATE_SUPPLIER_UNBIND state.
>>
>> Signed-off-by: Vivek Gautam <vivek.gautam@...eaurora.org>
>> Suggested-by: Lukas Wunner <lukas@...ner.de>
>> ---
>>
>> Lukas, as suggested in the thread [1] this change adds additional flag
>> to autoremove device links on supplier unbind.
>> For arm-smmu, we want to _not_ keep references to the device links
>> added between arm-smmu, and consumer devices.
>> Robin also pointed to [2] the need to autoremove the device link on
>> supplier unbind rather than consumer unbind.
>
> Please CC device links patches to linux-pm.

Thanks for the quick review. Sure, will keep this noted from now on.

>
>> [1] https://lkml.org/lkml/2018/3/14/390
>> [2] https://lkml.org/lkml/2018/5/21/381
>>
>>   drivers/base/core.c    | 10 ++++++++++
>>   include/linux/device.h |  2 ++
>>   2 files changed, 12 insertions(+)
>>
>> diff --git a/drivers/base/core.c b/drivers/base/core.c
>> index b610816eb887..52c7222bb3c4 100644
>> --- a/drivers/base/core.c
>> +++ b/drivers/base/core.c
>> @@ -490,6 +490,16 @@ void device_links_driver_cleanup(struct device 
>> *dev)
>>             WARN_ON(link->flags & DL_FLAG_AUTOREMOVE);
>>           WARN_ON(link->status != DL_STATE_SUPPLIER_UNBIND);
>> +
>> +        /*
>> +         * autoremove the links between this @dev and its consumer
>> +         * devices that are not active, i.e. where the link state
>> +         * has moved to DL_STATE_SUPPLIER_UNBIND.
>> +         */
>> +        if (link->status == DL_STATE_SUPPLIER_UNBIND &&
>> +            link->flags & DL_FLAG_AUTOREMOVE_S)
>> +            kref_put(&link->kref, __device_link_del);
>> +
>>           WRITE_ONCE(link->status, DL_STATE_DORMANT);
>>       }
>>   diff --git a/include/linux/device.h b/include/linux/device.h
>> index 477956990f5e..6033bf58453d 100644
>> --- a/include/linux/device.h
>> +++ b/include/linux/device.h
>> @@ -779,11 +779,13 @@ enum device_link_state {
>>    * AUTOREMOVE: Remove this link automatically on consumer driver 
>> unbind.
>>    * PM_RUNTIME: If set, the runtime PM framework will use this link.
>>    * RPM_ACTIVE: Run pm_runtime_get_sync() on the supplier during 
>> link creation.
>> + * AUTOREMOVE_S: Remove this link automatically on supplier driver 
>> unbind.
>>    */
>>   #define DL_FLAG_STATELESS    BIT(0)
>>   #define DL_FLAG_AUTOREMOVE    BIT(1)
>>   #define DL_FLAG_PM_RUNTIME    BIT(2)
>>   #define DL_FLAG_RPM_ACTIVE    BIT(3)
>> +#define DL_FLAG_AUTOREMOVE_S    BIT(4)
>
> Couldn't you invent a better name for this one?

Frankly, I wanted to have something like DL_FLAG_AUTOREMOVE_SUPPLIER, 
but that
felt a bit too long considering other flags.
Can you please consider suggesting a concise name?

Regards
Vivek
>
>>     /**
>>    * struct device_link - Device link representation.
>
>