From: linux@roeck-us.net (Guenter Roeck) Date: Tue, 30 Sep 2014 06:47:46 -0700 Subject: [PATCH 1/6] watchdog: watchdog_dev: WATCHDOG_KEEP_ON feature In-Reply-To: <542A8471.8060401@elproma.com.pl> References: <1411419350-1297-1-git-send-email-j.uzycki@elproma.com.pl> <5424E535.7060009@roeck-us.net> <542987F9.2080907@elproma.com.pl> <542A3392.8020903@roeck-us.net> <542A8471.8060401@elproma.com.pl> Message-ID: <542AB482.5060102@roeck-us.net> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 09/30/2014 03:22 AM, Janusz U?ycki wrote: > > W dniu 2014-09-30 06:37, Guenter Roeck pisze: >> On 09/29/2014 09:25 AM, Janusz U?ycki wrote: >>> >>>> This patch set is trying to solve four problems at once: >>>> >>>> 1) Auto-start watchdog when its driver registers >>>> 2) Keep watchdog running when its driver registers until userspace opens it >>>> 3) Handle watchdogs which can not be stopped after being started >>>> 4) Keep watchdog running with kernel timer after it has been closed, >>>> even if it can be stopped. >>>> >>>> The next time adds 'boot time protection', which is really another term >>>> for an initial timeout, and case 5). >>>> >>>> That is a bit too much for a single patch and, even more so, a single flag. >>> >>> OK, but I think [PATCH 3/6] could be applied. >>> Do you agree? Should I resent it separately? >> >> Yes, it looks ok. > > Can you apply it? > I don't apply watchdog patches; I only provide review feedback. >> >>> I omited in the comment >>> "The patch adds suspend/resume PM support to stmp3xxx_rtc_wdt >>> watchdog driver" because the subject is almost the same. >>> >>>> Let's look at one case after another. >>>> >>>> Auto-start watchdog when its driver registers - this makes sense as a >>>> feature just by itself. A good name for its flag might be something like >>>> WDT_AUTOSTART. A matching module parameter might also make sense. >>>> >>>> autostart: >>>> Set to 0 to disable, -1 to enable with unlimited timeout, >>>> or for an initial timeout of seconds. >>> >>> Current start(1) + keep-on(2,3,4) + boottime(5) combined. It looks OK. >>> >> Maybe for you. For me they are different cases. > > They are different. I missed some words in the first sentence :) > However autostart automatically combine them together again :) > The problem is common timer so the patches are dependent. > There is no reason to use more timers. > I did not say use multiple timers. I said use multiple flags. Guenter