linux-kernel - Re: [PATCH 4/6] arm: mach-omap2: remove "OMAP_DEVICE_NO_IDLE_ON

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: <51714CCC.9090800@ti.com>
Date:	Fri, 19 Apr 2013 19:25:24 +0530
From:	Sourav Poddar <sourav.poddar@...com>
To:	Kevin Hilman <khilman@...aro.org>
CC:	<gregkh@...uxfoundation.org>, <tony@...mide.com>,
	<rmk+kernel@....linux.org.uk>, <linux-serial@...r.kernel.org>,
	<linux-omap@...r.kernel.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 4/6] arm: mach-omap2: remove "OMAP_DEVICE_NO_IDLE_ON_SUSPEND"
 check

Hi Kevin,
On Friday 19 April 2013 03:33 AM, Kevin Hilman wrote:
> Sourav Poddar<sourav.poddar@...com>  writes:
>
>> On Thursday 18 April 2013 11:35 PM, Kevin Hilman wrote:
>>> Sourav Poddar<sourav.poddar@...com>   writes:
>>>
>>>> Remove the "OMAP_DEVICE_NO_IDLE_ON_SUSPEND" check, since UART was the only one making
>>>> use of it. Now serial core/driver takes care of the case when "no_console_suspend"
>>>> is used in the bootargs and you need to keep the clock enable for console even while suspend.
>>>>
>>>> Signed-off-by: Sourav Poddar<sourav.poddar@...com>
>>> NAK.  This patch will break many things...
>>>
>>>> ---
>>>>    arch/arm/mach-omap2/omap_device.c |    7 +------
>>>>    1 files changed, 1 insertions(+), 6 deletions(-)
>>>>
>>>> diff --git a/arch/arm/mach-omap2/omap_device.c b/arch/arm/mach-omap2/omap_device.c
>>>> index 381be7a..d6dce8f 100644
>>>> --- a/arch/arm/mach-omap2/omap_device.c
>>>> +++ b/arch/arm/mach-omap2/omap_device.c
>>>> @@ -620,11 +620,8 @@ static int _od_suspend_noirq(struct device *dev)
>>>>    	ret = pm_generic_suspend_noirq(dev);
>>>>
>>>>    	if (!ret&&   !pm_runtime_status_suspended(dev)) {
>>>> -		if (pm_generic_runtime_suspend(dev) == 0) {
>>>> -			if (!(od->flags&   OMAP_DEVICE_NO_IDLE_ON_SUSPEND))
>>>> -				omap_device_idle(pdev);
>>> Why did you remove the omap_device_idle() here?
>> This patch is used along with patch3 to get rid of the issue. I posted
>> them as a
>> seperate patch beacuse of the subject line as one goes under drivers/*
>> and the other
>> arm/mach-omap2/*..
>>
>> This check was only valid for UART, and if od->flags is set to the
>> "OMAP_DEVICE_NO_IDLE_ON_SUSPEND" flag, then UART will not be idled.
> correct, but *every other device* would be idled (if not already idle.)
>
>> But now, we no longer depend on od->flag value to prevent idling of
>> our console UART as the prepare/complete apis will take care of them.
> Right, so removing the check on od->flags is fine, but what I asked
> about is why you removed the omap_device_idle() call.
>
> Remember that this code is called for *every* omap_device during
> suspend, not just UART.
>
> What you did stops *every* device from being idled during suspend.
>
> You didn't read my whole message.  Specifically this part:
>
Yes, got your point. omap_device_idle should not be called only
for console uart.

Just did a quick testing by including the following hunk on top of my 
patch series..

diff --git a/arch/arm/mach-omap2/omap_device.c 
b/arch/arm/mach-omap2/omap_device.c
index c226946..7480e87 100644
--- a/arch/arm/mach-omap2/omap_device.c
+++ b/arch/arm/mach-omap2/omap_device.c
@@ -617,8 +617,10 @@ static int _od_suspend_noirq(struct device *dev)
         ret = pm_generic_suspend_noirq(dev);

         if (!ret && !pm_runtime_status_suspended(dev)) {
-               if (pm_generic_runtime_suspend(dev) == 0)
+               if (pm_generic_runtime_suspend(dev) == 0) {
+                       omap_device_idle(pdev);
                         od->flags |= OMAP_DEVICE_SUSPENDED;
+               }
         }

         return ret;
@@ -631,6 +633,7 @@ static int _od_resume_noirq(struct device *dev)

         if ((od->flags & OMAP_DEVICE_SUSPENDED) &&
             !pm_runtime_status_suspended(dev)) {
+               omap_device_enable(pdev);
                 od->flags &= ~OMAP_DEVICE_SUSPENDED;
                 pm_generic_runtime_resume(dev);
         }
And found the wakeup from UART is no more functional.
So, the omap_device_idle gets called for console UART also, thereby
preventing the "no_idle_on_suspend" theory.

Hence, merely putting prepare/complete callback the way I did is not 
helping.
We need to delete omap_device_idle also, which I agree is not correct.
So, we need a way to bypass this "omap_device_idle"
call for console UART. ?
What my understanding was that after modifying serial driver,
bypass the entire hunk[1] for console UART?


>>> Note that the check is for when the flag is *not* set, so this patch
>>> changes behavior for all the drivers that do not use
>>> _NO_IDLE_ON_SUSPEND.  I think that's the opposite of what you intended.
> Kevin

--
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/