Hi Timo,


On 22 May 2015 at 14:30, Timo Kokkonen [off-list ref] wrote:

On 21.05.2015 11:32, fu.wei@linaro.org wrote:

quoted

From: Fu Wei <redacted>

Also update Documentation/watchdog/watchdog-kernel-api.txt to
introduce:
(1)the new elements in the watchdog_device and watchdog_ops struct;
(2)the new API "watchdog_init_timeouts".

Reasons:
(1)kernel already has two watchdog drivers are using "pretimeout":
        drivers/char/ipmi/ipmi_watchdog.c
        drivers/watchdog/kempld_wdt.c(but the definition is different)
(2)some other dirvers are going to use this: ARM SBSA Generic Watchdog

Hi,

As I was proposing some other API changes with my early-timeout-sec work, I
can see my work is going to collide with your API change proposal a bit. So
maybe I should ask your opinion as well..

Thank you for reminding me of that, I saw "early-timeout-sec", but
because I don't get it in kernel API or watchdog_core.c, so I did not
pay attention to it.
Sorry.

Could this pretimeout feature be something that other drivers could benefit
too?

yes , as you may know, I am making a patch for pretimeout support in
watchdog framework

I can see that it does not do anything else except call panic() before
letting the watchdog expire. This is something that could be emulated easily
by the watchdog core for drivers that don't know anything about pretimeouts
at all.

For SBSA watchdog, there are two stage timeout, and according to kernel doc:
----------------------
Pretimeouts:

Some watchdog timers can be set to have a trigger go off before the
actual time they will reset the system.  This can be done with an NMI,
interrupt, or other mechanism.  This allows Linux to record useful
information (like panic information and kernel coredumps) before it
resets.
----------------------

I think panic() is the right way to do now.  but people may also need
to config :
PANIC_TIMEOUT [!=0]
KEXEC
KDUMP
for debug reason

The way I was planning the API change there would need to be a small change
with each watchdog driver in order to let the watchdog core take over
generic behaviour on behalf of the driver. My goal was to make the change so
that each driver that gets converted to the new API extensions gets a
support for early-timeout-sec for free, without needing to enable support
for it any way. If the API was designed properly, also pretimeouts could be
handled easily and maybe even so that other drivers could have that feature
even though their hardware does not explicitly give any support for it.

could you please point out your patch , then I can learn your idea :-)
For now , I am working on a common "Pretimeouts" following the concept
in kernel doc.

Any thoughts?

my thoughts is in  my pretimeout patch , would you provide some suggestion ?

Great thanks !

Thanks,
-Timo

-- 
Best regards,

Fu Wei
Software Engineer
Red Hat Software (Beijing) Co.,Ltd.Shanghai Branch
Ph: +86 21 61221326(direct)
Ph: +86 186 2020 4684 (mobile)
Room 1512, Regus One Corporate Avenue,Level 15,
One Corporate Avenue,222 Hubin Road,Huangpu District,
Shanghai,China 200021