On 04/01/2015 08:47 AM, Joel Stanley wrote:

Hi Andrew,

On Wed, Apr 1, 2015 at 9:09 AM, Andrew Morton [off-list ref] wrote:

quoted

quoted

On Mon, 30 Mar 2015 12:45:32 +1030 Joel Stanley [off-list ref] wrote:

quoted

quoted

The kernel has orderly_poweroff which allows the kernel to initiate a
graceful shutdown of userspace, by running /sbin/poweroff. This adds
orderly_reboot that will cause userspace to shut itself down by calling
/sbin/reboot.

This will be used for shutdown initiated by a system controller on
platforms that do not use ACPI.

gee.  There are a lot of callers of emergency_restart().  Why is the
BMC reboot special, and how many of the emergency_restart() callers
really be using orderly_reboot()?

The BMC reboot is intended to be a graceful shutdown - let userspace
do it's thing before the system goes down.

Userspace may chose to stop and perform some long, slow teardown
before it gets around to shutting down. We don't want to move callers
over orderly_reboot() if they're shutting the system down due to a
critical failure, eg. printer on fire.

I had a read of the emergency_restart() callers and I didn't see any
obvious cases for moving over to orderly_reboot().

quoted

quoted

We have /proc/sys/kernel/poweroff_cmd.  Should we have
/proc/sys/kernel/reboot_cmd as well?  If not,
kernel/reboot.c:reboot_cmd[] can be made static ;)

I don't think we need it. I'll make reboot_cmd[] static.

Just to have parity with power off command, /proc/sys/kernel/reboot_cmd would
be nice to have.