linux-kernel - Re: [PATCH] RTC: Add an alarm disable quirk

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: <51EDAF7D.3000605@linaro.org>
Date:	Mon, 22 Jul 2013 15:17:33 -0700
From:	John Stultz <john.stultz@...aro.org>
To:	Borislav Petkov <bp@...en8.de>
CC:	LKML <linux-kernel@...r.kernel.org>, Jiri Kosina <jkosina@...e.cz>,
	Borislav Petkov <bp@...e.de>,
	Thomas Gleixner <tglx@...utronix.de>,
	Rabin Vincent <rabin.vincent@...ricsson.com>
Subject: Re: [PATCH] RTC: Add an alarm disable quirk

On 07/22/2013 02:27 PM, Borislav Petkov wrote:
> On Mon, Jul 22, 2013 at 02:15:57PM -0700, John Stultz wrote:
>> On 07/22/2013 02:12 PM, Borislav Petkov wrote:
>>> On Mon, Jul 22, 2013 at 01:59:01PM -0700, John Stultz wrote:
>>>> So did this work some of the time, but not all? Or was the behavior
>>>> totally unchanged with this?
>>> Yep, some of the time. The first couple of runs it worked and I was
>>> euphoric and then it rebooted and I almost threw the box out the window
>>> :-)
>> I can understand your frustration. :)
>>
>> But its interesting it sort of worked, no?  The bit you discovered
>> earlier with the dump_stack debugging call, where we're actually
>> disabling the irq twice was interesting.
>>
>> If you use the debugging patch with this change, does it show any
>> different in logic between the working cases and the instant-reboot
>> case?
> Ok, I'm kinda confused with so many experiments I did, what we actually
> want to try:
>
> Do we want to use the filter thingy:
>
> ---
> diff --git a/drivers/rtc/rtc-cmos.c b/drivers/rtc/rtc-cmos.c
> index be06d7150de5..bb265f1651e7 100644
> --- a/drivers/rtc/rtc-cmos.c
> +++ b/drivers/rtc/rtc-cmos.c
> @@ -304,6 +304,9 @@ static void cmos_irq_enable(struct cmos_rtc *cmos, unsigned char mask)
>          rtc_control = CMOS_READ(RTC_CONTROL);
>          cmos_checkintr(cmos, rtc_control);
>
> +       if (rtc_control == mask)
> +               return;
> +
>          rtc_control |= mask;
>          CMOS_WRITE(rtc_control, RTC_CONTROL);
>          hpet_set_rtc_irq_bit(mask);
> @@ -316,6 +319,10 @@ static void cmos_irq_disable(struct cmos_rtc *cmos, unsigned char mask)
>          unsigned char   rtc_control;
>
>          rtc_control = CMOS_READ(RTC_CONTROL);
> +
> +       if (!(rtc_control & mask))
> +               return;
> +
>          rtc_control &= ~mask;
>          CMOS_WRITE(rtc_control, RTC_CONTROL);
>          hpet_mask_rtc_irq_bit(mask);
> --
>
> and also add dump_stack to see what calls cmos_irq_disable?
>
> In the run I had, the first call came from rtc_dev_ioctl so I'm guessing
> userspace and the following one was rtc workqueue rtc_timer_do_work.
>
> Just let me know what exactly we want to try and I'll do it tomorrow, on
> a clear head and not half asleep now :-)

So we probably want to do the following:
* Add your printk debug messages in rtc_cmos_read/write
* Add dump_stack to cmos_irq_disable
* Then on two machines (one working normally, the other broken): Boot 
the systems & shut them down after 5 minutes.
* Send out the full logs for both.
* Add the filtering logic above to the broken machine.
* Boot the system, shut it down after 5 minutes. Do this repeatedly 
until you get a failure (instant reboot) and a pass (stays off)
* Send out the full logs.

Hopefully from this we can sort out what exactly is going on.

Thanks again for your interest in hunting this down.

thanks
-john
--
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/