| 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
| ||
|
Message-ID: <4E0237CD.7060503@igalia.com>
Date: Wed, 22 Jun 2011 20:43:25 +0200
From: Carlos Alberto Lopez Perez <clopez@...lia.com>
To: Thadeu Lima de Souza Cascardo <cascardo@...oscopio.com>
CC: Daniel Oliveira Nascimento <don@...t.com.br>,
Matthew Garrett <mjg@...hat.com>,
platform-driver-x86@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/1] x86/platform: classmate-laptop: Add cmpc_keys support
for new classmate laptops
On 06/21/2011 10:50 PM, Thadeu Lima de Souza Cascardo wrote:
> On Tue, Jun 21, 2011 at 10:24:57PM +0200, Carlos Alberto Lopez Perez wrote:
>> On 06/21/2011 09:50 PM, Thadeu Lima de Souza Cascardo wrote:
>>> On Tue, Jun 21, 2011 at 09:40:01PM +0200, Carlos Alberto Lopez Perez wrote:
>>>> The cmpc_keys ACPI driver from the classmate-laptop module fails to
>>>> recognize the device on the 4th generation of classmate laptops (Atom N450)
>>>> because the device ID is named "FNBT0000" (uppercase N) on this laptops.
>>>>
>>>> MODALIAS=acpi:FNBT0000:
>>>>
>>>> This patch makes the driver cmpc_keys recognize it:
>>>>
>>>> input: cmpc_keys as /devices/LNXSYSTM:00/device:00/FNBT0000:00/input/input18
>>>>
>>>> Signed-off-by: Carlos Alberto Lopez Perez <clopez@...lia.com>
>>>> ---
>>>> drivers/platform/x86/classmate-laptop.c | 3 +++
>>>> 1 files changed, 3 insertions(+), 0 deletions(-)
>>>>
>>>> diff --git a/drivers/platform/x86/classmate-laptop.c b/drivers/platform/x86/classmate-laptop.c
>>>> index 94f93b6..03896cd 100644
>>>> --- a/drivers/platform/x86/classmate-laptop.c
>>>> +++ b/drivers/platform/x86/classmate-laptop.c
>>>> @@ -40,6 +40,7 @@ struct cmpc_accel {
>>>> #define CMPC_TABLET_HID "TBLT0000"
>>>> #define CMPC_IPML_HID "IPML200"
>>>> #define CMPC_KEYS_HID "FnBT0000"
>>>> +#define CMPC_KEYS_4GEN_HID "FNBT0000"
>>>>
>>>> /*
>>>> * Generic input device code.
>>>> @@ -683,6 +684,7 @@ static int cmpc_keys_remove(struct acpi_device *acpi, int type)
>>>>
>>>> static const struct acpi_device_id cmpc_keys_device_ids[] = {
>>>> {CMPC_KEYS_HID, 0},
>>>> + {CMPC_KEYS_4GEN_HID, 0},
>>>> {"", 0}
>>>> };
>>>>
>>>> @@ -754,6 +756,7 @@ static const struct acpi_device_id cmpc_device_ids[] = {
>>>> {CMPC_TABLET_HID, 0},
>>>> {CMPC_IPML_HID, 0},
>>>> {CMPC_KEYS_HID, 0},
>>>> + {CMPC_KEYS_4GEN_HID, 0},
>>>> {"", 0}
>>>> };
>>>>
>>>> --
>>>> 1.7.5.3
>>>>
>>>
>>> Not-Acked-By: Thadeu Lima de Souza Cascardo <cascardo@...oscopio.com>
>>>
>>> Recent Linux versions have "upper-cased" the device names, because ACPI
>>> spec says devices must have upper-case letters. This is probably related
>>> to this change, not because the device name has changed in recent
>>> versions of the device. So the correct fix would be to either revert
>>> this change in Linux or use only a single uppercase name for the device.
>>>
>>> Thanks anyway for this report. I have received one report in private a
>>> while back, but was too lazy to discuss this on the list. Any pointers
>>> on that? Easist way to go is to simply replace FnBT for FNBT.
>>>
>>> Regards,
>>> Cascardo.
>>
>> The ACPI specification [1] says the following:
>>
>> """
>> A valid PNP ID must be of the form “AAA####” where A is an uppercase letter
>> and # is a hex digit.
>> A valid ACPI ID must be of the form “ACPI####” where # is a hex digit.
>> """
>>
>> It don't says that the ACPI ID must be uppercase.
>>
>> I don't have an older generation of the classmate laptop to test if the
>> ACPI ID is FnBT or FNBT, but I guess that if it was working until now, is
>> because the older generations have FnBT instead of FNBT.
>>
>> However I just found a previous discussion about this issue on LKML [2] and
>> seems that the patch is included (commit: 72638f5).
>>
>> So... this means that the cmpc_keys module was not working for any
>> classmate laptop since ~2.6.29 ?
>>
>> If that is the case then replacing FnBT with FNBT will be fine.
>>
>> Thanks!
>>
>> Best regards!
>>
>
> Commit 77b23f712bc40a65160e7d02b045f1562bb43ff1 has changed this
> behaviour between 2.6.36 and 2.6.37. Could you try using 2.6.36 and see
> if the device appears as FnBT? Also, you could try using acpidump and
> look at the decompiled DSDT so we can confirm whether it is FnBT or
> FNBT. Attached is a patch that, when comparing device and driver names
> for ACPI, will do it case-insensitivily.
>
You are right. The real ACPI ID of the device is FnBT0000 but the kernel
is uppercasing it.
# acpidump | grep -i -C1 FNBT0000
0b30: 00 0a 00 0a 14 5b 82 34 46 4e 42 54 08 5f 48 49 .....[.4FNBT._HI
0b40: 44 0d 46 6e 42 54 30 30 30 30 00 08 5f 55 49 44 D.FnBT0000.._UID
0b50: 0a 00 08 46 4e 4c 4b 0a 01 14 11 46 4e 43 54 01 ...FNLK....FNCT.
[*] In kernels <= 2.6.36 (Tested in 2.6.32 and 2.6.36) the ACPI ID is
registered as FnBT0000
# cat /sys/bus/acpi/devices/FnBT0000\:00/modalias
acpi:FnBT0000:
And the cmpc_keys driver of the classmate-laptop module recognized the
device as expected:
input: cmpc_keys as /devices/LNXSYSTM:00/LNXSYBUS:00/FnBT0000:00/input/input5
[*] In kernels => 2.6.37 (Tested in 2.6.37 and 2.6.39) the ACPI ID is
registered uppercased as FNBT0000 and therefore the cmpc_keys driver fails
to recognize it.
# cat /sys/bus/acpi/devices/FNBT0000\:00/modalias
acpi:FNBT0000:
[*] With the kernel 2.6.39.1 and your patch for drivers/acpi/scan.c the
ACPI ID is registered uppercased as FNBT0000 *but* the cmpc_keys driver
recognizes it.
# cat /sys/bus/acpi/devices/FNBT0000\:00/modalias
acpi:FNBT0000:
input: cmpc_keys as /devices/LNXSYSTM:00/device:00/FNBT0000:00/input/input5
[*] With the kernel 2.6.39.1 and the commit
77b23f712bc40a65160e7d02b045f1562bb43ff1 reverted happens the same that
with kernels <= 2.6.36 (the ACPI ID is registered as FnBT0000)
> So, we have three options here. Revert an ACPICA commit, apply a
> different matching code for ACPI bus in Linux or tell all drivers they
> must use an uppper-case name instead of the device name as it appears in
> the ACPI tables.
>
Yes, we have three options here to solve the problem but I'm not sure which
one is the correct.
At least I think it has been demonstrated that the ACPI ID is *not
required* to be uppercase by the ACPI specification as you can see in the
acpidump of this device (FnBT0000).
So probably the "orthodox" way of solving this problem is reverting the
commit 77b23f712bc40a65160e7d02b045f1562bb43ff1 which is doing a wrong
assumption (its not true that the ACPI ID _must_ be uppercase).
Regards!
Download attachment "signature.asc" of type "application/pgp-signature" (263 bytes)
Powered by blists - more mailing lists