linux-kernel - Re: [PATCH] eeprom: at24: don't use the default regmap locking

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:   Thu, 14 Dec 2017 21:27:31 +0100
From:   Bartosz Golaszewski <brgl@...ev.pl>
To:     Heiner Kallweit <hkallweit1@...il.com>
Cc:     linux-i2c <linux-i2c@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Bartosz Golaszewski <brgl@...ev.pl>
Subject: Re: [PATCH] eeprom: at24: don't use the default regmap locking

2017-12-10 19:57 GMT+01:00 Bartosz Golaszewski <brgl@...ev.pl>:
> 2017-12-03 22:47 GMT+01:00 Bartosz Golaszewski <brgl@...ev.pl>:
>> We need to always wait for the last write to complete and we want the
>> regmap to be protected from concurrent access during that wait. That
>> makes using the regmap's default locking impossible and we're
>> protecting it with an external mutex.
>>
>> Set an empty function as the callback for regmap's lock and unlock
>> routines so that it doesn't allocate any internal resources for
>> unnecessary locking.
>>
>> Signed-off-by: Bartosz Golaszewski <brgl@...ev.pl>
>> ---
>> This is a temporary solution - I'm working on a patch extending the
>> regmap framework with locking policy options, which will allow users
>> to manually lock & unlock regmaps if needed or disable locking at all.
>>
>>  drivers/misc/eeprom/at24.c | 9 +++++++++
>>  1 file changed, 9 insertions(+)
>>
>> diff --git a/drivers/misc/eeprom/at24.c b/drivers/misc/eeprom/at24.c
>> index 625b00166117..34bc394a9c14 100644
>> --- a/drivers/misc/eeprom/at24.c
>> +++ b/drivers/misc/eeprom/at24.c
>> @@ -498,14 +498,23 @@ static unsigned int at24_get_offset_adj(u8 flags, unsigned int byte_len)
>>         }
>>  }
>>
>> +static void at24_regmap_lock_unlock_none(void *map)
>> +{
>> +
>> +}
>> +
>>  static const struct regmap_config regmap_config_8 = {
>>         .reg_bits = 8,
>>         .val_bits = 8,
>> +       .lock = at24_regmap_lock_unlock_none,
>> +       .unlock = at24_regmap_lock_unlock_none,
>>  };
>>
>>  static const struct regmap_config regmap_config_16 = {
>>         .reg_bits = 16,
>>         .val_bits = 8,
>> +       .lock = at24_regmap_lock_unlock_none,
>> +       .unlock = at24_regmap_lock_unlock_none,
>>  };
>>
>>  static int at24_probe(struct i2c_client *client, const struct i2c_device_id *id)
>> --
>> 2.15.1
>>
>
> There were no comments. I applied this patch. These changes will be
> removed in 4.17 when we'll enable the newly added disable_locking
> variable in regmap config.
>
> Bartosz

It turned out that this may cause problem with user space initiating
operations on the regmap over debugfs entries. This has been correctly
implemented in regmap and will be available from 4.16. I backed this
patch out.

Thanks,
Bartosz