linux-kernel - Re: [PATCH 3/5] watchdog: sp805: set WDOG_HW

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:   Wed, 23 May 2018 18:15:14 +0100
From:   Robin Murphy <robin.murphy@....com>
To:     Ray Jui <ray.jui@...adcom.com>,
        Scott Branden <scott.branden@...adcom.com>,
        Guenter Roeck <linux@...ck-us.net>
Cc:     Mark Rutland <mark.rutland@....com>, devicetree@...r.kernel.org,
        linux-watchdog@...r.kernel.org,
        Catalin Marinas <catalin.marinas@....com>,
        Will Deacon <will.deacon@....com>,
        linux-kernel@...r.kernel.org, Rob Herring <robh+dt@...nel.org>,
        bcm-kernel-feedback-list@...adcom.com,
        Wim Van Sebroeck <wim@...ux-watchdog.org>,
        Frank Rowand <frowand.list@...il.com>,
        linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH 3/5] watchdog: sp805: set WDOG_HW_RUNNING when appropriate

On 23/05/18 17:29, Ray Jui wrote:
> Hi Robin,
> 
> On 5/23/2018 4:48 AM, Robin Murphy wrote:
>> On 23/05/18 08:52, Scott Branden wrote:
>>>
>>>
>>> On 18-05-22 04:24 PM, Ray Jui wrote:
>>>> Hi Guenter,
>>>>
>>>> On 5/22/2018 1:54 PM, Guenter Roeck wrote:
>>>>> On Tue, May 22, 2018 at 11:47:18AM -0700, Ray Jui wrote:
>>>>>> If the watchdog hardware is already enabled during the boot process,
>>>>>> when the Linux watchdog driver loads, it should reset the watchdog 
>>>>>> and
>>>>>> tell the watchdog framework. As a result, ping can be generated from
>>>>>> the watchdog framework, until the userspace watchdog daemon takes 
>>>>>> over
>>>>>> control
>>>>>>
>>>>>> Signed-off-by: Ray Jui <ray.jui@...adcom.com>
>>>>>> Reviewed-by: Vladimir Olovyannikov 
>>>>>> <vladimir.olovyannikov@...adcom.com>
>>>>>> Reviewed-by: Scott Branden <scott.branden@...adcom.com>
>>>>>> ---
>>>>>>   drivers/watchdog/sp805_wdt.c | 22 ++++++++++++++++++++++
>>>>>>   1 file changed, 22 insertions(+)
>>>>>>
>>>>>> diff --git a/drivers/watchdog/sp805_wdt.c 
>>>>>> b/drivers/watchdog/sp805_wdt.c
>>>>>> index 1484609..408ffbe 100644
>>>>>> --- a/drivers/watchdog/sp805_wdt.c
>>>>>> +++ b/drivers/watchdog/sp805_wdt.c
>>>>>> @@ -42,6 +42,7 @@
>>>>>>       /* control register masks */
>>>>>>       #define    INT_ENABLE    (1 << 0)
>>>>>>       #define    RESET_ENABLE    (1 << 1)
>>>>>> +    #define    ENABLE_MASK    (INT_ENABLE | RESET_ENABLE)
>>>>>>   #define WDTINTCLR        0x00C
>>>>>>   #define WDTRIS            0x010
>>>>>>   #define WDTMIS            0x014
>>>>>> @@ -74,6 +75,18 @@ module_param(nowayout, bool, 0);
>>>>>>   MODULE_PARM_DESC(nowayout,
>>>>>>           "Set to 1 to keep watchdog running after device release");
>>>>>>   +/* returns true if wdt is running; otherwise returns false */
>>>>>> +static bool wdt_is_running(struct watchdog_device *wdd)
>>>>>> +{
>>>>>> +    struct sp805_wdt *wdt = watchdog_get_drvdata(wdd);
>>>>>> +
>>>>>> +    if ((readl_relaxed(wdt->base + WDTCONTROL) & ENABLE_MASK) ==
>>>>>> +        ENABLE_MASK)
>>>>>> +        return true;
>>>>>> +    else
>>>>>> +        return false;
>>>>>
>>>>>     return !!(readl_relaxed(wdt->base + WDTCONTROL) & ENABLE_MASK));
>>>>>
>>>>
>>>> Note ENABLE_MASK contains two bits (INT_ENABLE and RESET_ENABLE); 
>>>> therefore, a simple !!(expression) would not work? That is, the 
>>>> masked result needs to be compared against the mask again to ensure 
>>>> both bits are set, right?
>>> Ray - your original code looks correct to me.  Easier to read and 
>>> less prone to errors as shown in the attempted translation to a 
>>> single statement.
>>
>>      if (<boolean condition>)
>>          return true;
>>      else
>>          return false;
>>
>> still looks really dumb, though, and IMO is actually harder to read 
>> than just "return <boolean condition>;" because it forces you to stop 
>> and double-check that the logic is, in fact, only doing the obvious 
>> thing.
> 
> If you can propose a way to modify my original code above to make it 
> more readable, I'm fine to make the change.

Well,

	return readl_relaxed(wdt->base + WDTCONTROL) & ENABLE_MASK == ENABLE_MASK;

would probably be reasonable to anyone other than the 80-column zealots, 
but removing the silly boolean-to-boolean translation idiom really only 
emphasises the fact that it's fundamentally a big complex statement; for 
maximum clarity I'd be inclined to separate the two logical operations 
(read and comparison), e.g.:

	u32 wdtcontrol = readl_relaxed(wdt->base + WDTCONTROL);

	return wdtcontrol & ENABLE_MASK == ENABLE_MASK;

which is still -3 lines vs. the original.

> As I mentioned, I don't think the following change proposed by Guenter 
> will work due to the reason I pointed out:
> 
> return !!(readl_relaxed(wdt->base + WDTCONTROL) & ENABLE_MASK));

FWIW, getting the desired result should only need one logical not 
swapping for a bitwise one there:

	return !(~readl_relaxed(wdt->base + WDTCONTROL) & ENABLE_MASK);

but that's well into "too clever for its own good" territory ;)

Robin.