linux-kernel - Re: [PATCH 07/13] scpi: ignore init_versions failure if reported not supported

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]

Message-ID: <584cd7fc-c6dd-d380-0842-ca663fa298ab@arm.com>
Date:   Tue, 23 Aug 2016 16:01:24 +0100
From:   Sudeep Holla <sudeep.holla@....com>
To:     Neil Armstrong <narmstrong@...libre.com>,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Cc:     Sudeep Holla <sudeep.holla@....com>,
        linux-amlogic@...ts.infradead.org, khilman@...libre.com,
        heiko@...ech.de, wxt@...k-chips.com, frank.wang@...k-chips.com
Subject: Re: [PATCH 07/13] scpi: ignore init_versions failure if reported not
 supported



On 23/08/16 15:55, Neil Armstrong wrote:
> On 08/23/2016 04:54 PM, Sudeep Holla wrote:
>>
>>
>> On 23/08/16 09:23, Neil Armstrong wrote:
>>> On 08/19/2016 06:46 PM, Sudeep Holla wrote:
>>>>
>>>>
>>>> On 18/08/16 11:11, Neil Armstrong wrote:
>>>>> In Amlogic GXBB Legacy SCPI, the LEGACY_SCPI_CMD_SCPI_CAPABILITIES report
>>>>> as SCPI_ERR_SUPPORT, so do not fail if this command is not supported.
>>>>>
>>>>> Signed-off-by: Neil Armstrong <narmstrong@...libre.com>
>>>>> ---
>>>>>  drivers/firmware/arm_scpi.c | 12 +++++++-----
>>>>>  1 file changed, 7 insertions(+), 5 deletions(-)
>>>>>
>>>>> diff --git a/drivers/firmware/arm_scpi.c b/drivers/firmware/arm_scpi.c
>>>>> index 3fe39fe..d3be4c5 100644
>>>>> --- a/drivers/firmware/arm_scpi.c
>>>>> +++ b/drivers/firmware/arm_scpi.c
>>>>> @@ -1111,12 +1111,13 @@ err:
>>>>>          ret = scpi_info->ops->init_versions(scpi_info);
>>>>>      else
>>>>>      ret = scpi_init_versions(scpi_info);
>>>>> -    if (ret) {
>>>>> +    if (ret && ret != -EOPNOTSUPP) {
>>>>>          dev_err(dev, "incorrect or no SCP firmware found\n");
>>>>>          scpi_remove(pdev);
>>>>>          return ret;
>>>>>      }
>>>>>
>>>>
>>>> Why not deal it in init_versions itself.
>>>>
>>>>> +    if (ret != -EOPNOTSUPP) {
>>>>>      _dev_info(dev, "SCP Protocol %d.%d Firmware %d.%d.%d version\n",
>>>>>            PROTOCOL_REV_MAJOR(scpi_info->protocol_version),
>>>>>            PROTOCOL_REV_MINOR(scpi_info->protocol_version),
>>>>
>>>> Why not have default value like 0.0 ? Just add a comment. Since get
>>>> version is exported out, IMO having default value makes more sense. What
>>>> do you think ?
>>>>
>>>>> @@ -1124,15 +1125,16 @@ err:
>>>>>            FW_REV_MINOR(scpi_info->firmware_version),
>>>>>            FW_REV_PATCH(scpi_info->firmware_version));
>>>>>
>>>>> +        ret = sysfs_create_groups(&dev->kobj, versions_groups);
>>>>> +        if (ret)
>>>>> +            dev_err(dev, "unable to create sysfs version group\n");
>>>>> +    }
>>>>> +
>>>>
>>>> Again this can stay as is if we have default.
>>>>
>>>
>>> Printing version 0.0 firmware 0.0.0 is a nonsense for me...
>>>
>>
>> OK 0.0 was a wrong example. May be 0.1 ?
>>
>> Since the driver has already exposed, hypothetically user-space can use
>> that information, so IMO, we need to expose some static version for pre-v1.0
>>
>> I am surprised that capability is not supported as this was present even
>> in that legacy SCPI. Do you know what happens if you send that command ?
>> Have you done some experiments on that ?
>>
>
> I've experimented and returns EOPNOTSUPP, Amlogic confirmed to us the command was not implemented.
>
> This a clearly a corner-case.
>

OK, thanks for the confirmation.
Not exporting anything could be kind of breaking ABI as it was not made
optional when introduced :( (you can blame me ;))

-- 
Regards,
Sudeep