Skip to content

watchdog: Add millisecond-level capabilities

[email protected] linux-watchdog 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest6 patches 7 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/6watchdog: Allow a driver to use milliseconds instead of secondsdrivers/watchdog/watchdog_core.c drivers/watchdog/watchdog_dev.c include/linux/watchdog.h +1 more+82 -257 replies
2/6watchdog: Add ioctls for millisecond timeout handlingdrivers/watchdog/watchdog_core.c drivers/watchdog/watchdog_core.h drivers/watchdog/watchdog_dev.c +1 more+98 -370 replies
3/6watchdog: Add millisecond precision device attributesdrivers/watchdog/watchdog_dev.c+24 -40 replies
4/6watchdog: Add documentation for millisecond interfacesDocumentation/watchdog/watchdog-api.rst Documentation/watchdog/watchdog-kernel-api.rst+78 -110 replies
5/6watchdog:i6300: Convert to a millisecond watchdog devicedrivers/watchdog/i6300esb.c+56 -100 replies
6/6watchdog:softdog: Convert to a millisecond watchdogdrivers/watchdog/softdog.c+7 -60 replies

Reviews trailers per patch

No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.

Read the v1 thread

Cover letter [PATCH 0/6] watchdog: Add millisecond-level capabilities

As mentioned before in the list, this is work for the ELISA (Linux in safety critical systems) group. Some use cases require tight watchdog times so a system can recover quickly on a failure.

This patch series adds millisecond level capabilites that devices can use if they wish, new ioctls to use those interfaces, new device attributes with the higher precision, documentation, and it converts the i6300 and softdog watchdog to use milliseconds.

All existing user APIs should work as before. The interface between the watchdog subsystem and the driver is converted to be modal. If a flag is set in the options then all interactions are in milliseconds. If the flag is not set then they are in seconds.

It's all pretty straightforward except for the calculations in the i6300 driver, which required some thought.

Note that checkpatch gives warnings about there being no identifiers in the function arguments in the watchdog ops structure, but I tried to keep the style the same.

Thanks,

-corey