linux-kernel - Re: [PATCH 7/7] platform/chrome: cros_ec_lightbar: instantiate only if the EC has a lightbar.

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

Date:   Fri, 23 Nov 2018 12:10:15 +0100
From:   Enric Balletbo i Serra <enric.balletbo@...labora.com>
To:     Guenter Roeck <linux@...ck-us.net>
Cc:     lee.jones@...aro.org, gwendal@...omium.org, drinkcat@...omium.org,
        linux-kernel@...r.kernel.org, groeck@...omium.org,
        kernel@...labora.com, bleung@...omium.org,
        Olof Johansson <olof@...om.net>
Subject: Re: [PATCH 7/7] platform/chrome: cros_ec_lightbar: instantiate only
 if the EC has a lightbar.

Hi Guenter,

On 22/11/18 20:25, Guenter Roeck wrote:
> On Thu, Nov 22, 2018 at 12:33:56PM +0100, Enric Balletbo i Serra wrote:
>> Due to the way attribute groups visibility work, the function
>> cros_ec_lightbar_attrs_are_visible is called multiple times, once per
>> attribute, and each of these calls makes an EC transaction. For what is
>> worth the EC log reports multiple errors on boot when the lightbar is
>> not available. Instead, check if the EC has a lightbar in the probe
>> function and only instantiate the device.
>>
>> Ideally we should have instantiate the driver only if the
>> EC_FEATURE_LIGHTBAR is defined, but that's not possible because that flag
>> is not in the very first Pixel Chromebook (Link), only on Samus. So, the
>> driver is instantiated by his parent always.
>>
>> This patch changes a bit the actual behaviour. Before the patch if an EC
>> doesn't have a lightbar an empty lightbar folder is created in
>> /sys/class/chromeos/<ec device>, after the patch the empty folder is not
>> created, so, the folder is only created if the lightbar exists.
>>
>> Signed-off-by: Enric Balletbo i Serra <enric.balletbo@...labora.com>
> 
> Guess this is the answer to the suggestion I had before. Maybe the two patches
> should be merged together ? Or do  others think that they should be kept
> separate ?
> 

I did in a separate patch because it changes a bit the current behaviour (i.e
after that patch the lightbar directory will not appear if is not detected).
Having in a separate patch will allow us to revert cleanly if for some weird
reason we still want the old behaviour. So in general, first I moved the
attributes and then I did a follow up patch with the probe change. This also
happens with vbc driver.

Said that, I don't mind to merge the two patches.


> Additional comment below.
> 
> Thanks,
> Guenter
> 
>> ---
>>
>>  drivers/platform/chrome/cros_ec_lightbar.c | 29 +++++++++-------------
>>  1 file changed, 12 insertions(+), 17 deletions(-)
>>
>> diff --git a/drivers/platform/chrome/cros_ec_lightbar.c b/drivers/platform/chrome/cros_ec_lightbar.c
>> index 31d22f594fac..d255264eb082 100644
>> --- a/drivers/platform/chrome/cros_ec_lightbar.c
>> +++ b/drivers/platform/chrome/cros_ec_lightbar.c
>> @@ -567,37 +567,28 @@ static struct attribute *__lb_cmds_attrs[] = {
>>  	NULL,
>>  };
>>  
>> -static bool ec_has_lightbar(struct cros_ec_dev *ec)
>> +static bool cros_ec_has_lightbar(struct cros_ec_dev *ec_dev)
>>  {
>> -	return !!get_lightbar_version(ec, NULL, NULL);
>> -}
>> -
>> -static umode_t cros_ec_lightbar_attrs_are_visible(struct kobject *kobj,
>> -						  struct attribute *a, int n)
>> -{
>> -	struct device *dev = container_of(kobj, struct device, kobj);
>> -	struct cros_ec_dev *ec = to_cros_ec_dev(dev);
>> -	struct platform_device *pdev = to_platform_device(ec->dev);
>> +	struct platform_device *pdev = to_platform_device(ec_dev->dev);
>>  	struct cros_ec_platform *pdata = pdev->dev.platform_data;
>>  	int is_cros_ec;
>>  
>>  	is_cros_ec = strcmp(pdata->ec_name, CROS_EC_DEV_NAME);
>>  
> Can this now ever be false ?
> 

Yes, this happens for example on Samus, where there are two ECs, the first one
is named "cros_ec" and the second one is "cros_pd", so will fail in the second
case. Gwendal, correct me if I am wrong, but AFAIK this is a bit of hack because
some initial versions of Samus (coded Link I guess) have no support for the
EC_LIGHTBAR_FEATURE so we can't really use the features thing.

>>  	if (is_cros_ec != 0)
>> -		return 0;
>> +		return false;
>>  
>> -	/* Only instantiate this stuff if the EC has a lightbar */
>> -	if (ec_has_lightbar(ec)) {
>> -		ec_with_lightbar = ec;
>> -		return a->mode;
>> +	if (!!get_lightbar_version(ec_dev, NULL, NULL)) {
>> +		ec_with_lightbar = ec_dev;
> 
> Is this variable (and the associated check in lb_manual_suspend_ctrl)
> still necessary ?
> 

Hmm, right, I will double check and remove in next version.

>> +		return true;
>>  	}
>> -	return 0;
>> +
>> +	return false;
>>  }
>>  
>>  struct attribute_group cros_ec_lightbar_attr_group = {
>>  	.name = "lightbar",
>>  	.attrs = __lb_cmds_attrs,
>> -	.is_visible = cros_ec_lightbar_attrs_are_visible,
>>  };
>>  
>>  static int cros_ec_lightbar_probe(struct platform_device *pd)
>> @@ -611,6 +602,10 @@ static int cros_ec_lightbar_probe(struct platform_device *pd)
>>  		return -EINVAL;
>>  	}
>>  
>> +	/* Only instantiate this stuff if the EC has a lightbar */
>> +	if (!cros_ec_has_lightbar(ec_dev))
>> +		return -ENODEV;
>> +
>>  	/* Take control of the lightbar from the EC. */
>>  	lb_manual_suspend_ctrl(ec_dev, 1);
>>  
>