linux-kernel - Re: [PATCH] applesmc: add sysfs file to report OSK

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]

Message-Id: <4BC00520-E08A-43D0-AA2B-EA8154F3D160@suse.de>
Date:	Mon, 10 Dec 2012 21:19:51 +0100
From:	Alexander Graf <agraf@...e.de>
To:	"Henrik Rydberg" <rydberg@...omail.se>
Cc:	Guenter Roeck <groeck-dsl@...global.net>,
	"Gabriel L. Somlo" <somlo@....edu>, khali@...ux-fr.org,
	linux@...ck-us.net, lm-sensors@...sensors.org,
	linux-kernel@...r.kernel.org, rene@...ctcode.com
Subject: Re: [PATCH] applesmc: add sysfs file to report OSK


On 10.12.2012, at 20:54, Henrik Rydberg wrote:

> Hi Guenter,
> 
>> On Mon, Dec 10, 2012 at 09:51:35AM -0500, Gabriel L. Somlo wrote:
>>> The AppleSMC contains two char[32] keys, OSK0 and OSK1, which are not
>>> reported in the key count and index by default. These keys are used by
>>> the OS X boot sequence, and normally don't matter when running Linux.
>>> 
>>> This patch creates a sysfs entry which reports the value of these keys
>>> as an ASCII string, to help emulators (such as QEMU) load OS X when
>>> running on genuine Apple hardware.
>>> 
>>> Signed-off-by: Gabriel L. Somlo <somlo@....edu>
>>> ---
>>> 
>>> For extra context: To boot OS X as a guest, QEMU must (among others)
>>> emulate the AppleSMC. To boot successfully, OS X insists on querying
>>> the (emulated) SMC for the value of OSK0 and OSK1. Currently, these
>>> values must be supplied on the QEMU command line as
>>> 
>>>  -device applesmc,osk="...concatenated values of OSK0 and OSK1..."
>>> 
>>> With the availability of /sys/devices/platform/applesmc.768/osk, the
>>> emulated QEMU AppleSMC could acquire this string directly from the
>>> (Apple-manufactured) host machine.
>>> 
>> Hmm ... this is a non-hwmon attribute which doesn't really belong into hwmon
>> in the first place ... like several other attributes in the same driver.
>> 
>> So I'll leave it up to the maintainer to decide if we should accept it. Henrik ?
> 
> Indeed, the reaons against this patch are too many. I was just about
> to reply with the below:
> 
> Gabriel,
> 
> The OSK string seems constant accross machines, which renders the
> patch rather pointless, no? And even if the OSK differs between a
> couple of machines, the emulator could easily handle it gracefully.

The point is that the return value of the OSK is a copyrighted string, we can not include in any other layer. The only way to make this legally savvy is to read the key from the host.

> 
> There are also some technical issues with the patch below, to keep in
> mind for future submissions.

Sigh - most of the comments below go back to earlier review from me. He basically had a version almost exactly like what you're asking him to do :). Funny how code style taste differs.


Alex

> 
>> drivers/hwmon/applesmc.c |   18 ++++++++++++++++++
>> 1 files changed, 18 insertions(+), 0 deletions(-)
>> 
>> diff --git a/drivers/hwmon/applesmc.c b/drivers/hwmon/applesmc.c
>> index b41baff..0c7cc71 100644
>> --- a/drivers/hwmon/applesmc.c
>> +++ b/drivers/hwmon/applesmc.c
>> @@ -1013,6 +1013,23 @@ static ssize_t applesmc_key_at_index_store(struct device *dev,
>> 	return count;
>> }
>> 
>> +static ssize_t applesmc_osk_show(struct device *dev,
>> +				struct device_attribute *attr, char *sysfsbuf)
>> +{
>> +	int fail;
> 
> All other functions use 'ret' here...
> 
>> +
>> +	mutex_lock(&smcreg.mutex);
>> +	fail = read_smc(APPLESMC_READ_CMD, "OSK0", sysfsbuf, 32) ||
>> +	       read_smc(APPLESMC_READ_CMD, "OSK1", sysfsbuf + 32, 32);
> 
> The read function should propagate error messages, i.e., keep the
> return values here. And please read to buffers instead.
> 
>> +	mutex_unlock(&smcreg.mutex);
>> +	if (fail)
>> +		return -1;
> 
> Return error here.
> 
>> +
>> +	sysfsbuf[64] = '\n';
>> +	sysfsbuf[65] = '\0';
>> +	return 65;
> 
> A snprintf here, please.
> 
>> +}
>> +
>> static struct led_classdev applesmc_backlight = {
>> 	.name			= "smc::kbd_backlight",
>> 	.default_trigger	= "nand-disk",
>> @@ -1027,6 +1044,7 @@ static struct applesmc_node_group info_group[] = {
>> 	{ "key_at_index_type", applesmc_key_at_index_type_show },
>> 	{ "key_at_index_data_length", applesmc_key_at_index_data_length_show },
>> 	{ "key_at_index_data", applesmc_key_at_index_read_show },
>> +	{ "osk", applesmc_osk_show },
> 
> Unfortunately this is not a good place to put random things going
> forward.
> 
>> 	{ }
>> };
>> 
>> -- 
>> 1.7.7.6
>> 
> 
> Given the above issues together with the weak rationale for the patch
> in the first place, this patch will not be applied.
> 
> Thanks.
> Henrik

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/