Re: [Bug 219362] USB SATA does not correctly shutdown the SSD upon poweroff causing data loss

[Greg KH <greg@kroah.com>]

On Tue, Oct 08, 2024 at 12:00:33PM +0000, bugzilla-daemon@kernel.org wrote:
> It's an emergency shutdown in embedded scenario, we have capacitors that are
> able to sustain the system for 1 second in case of unexpected power loss. It's
> plenty of time to shutdown filesystems and drives properly. But we simply
> cannot afford to wait for userspace. Works nicely so far.

"so far" seems to mean "data loss" so I doubt that :)

> > Yes, that is because you are shutting down things in the proper order,
> like a normal shutdown will do.
> 
> Is there a reason for sysrq shutdown not to do that? I mean... c'mon...
> shutting down SSDs cleanly is the bare minimum we should do... if we fail to do
> that, we can as well just unplug the power without trying to unmount or sync
> anything...

Because the sysrq code does NOT go through the normal device shutdown
process, you are circumventing it, because it is only for emergencies.
Don't rely on it for a normal shutdown, as that is not what it is
designed for at all.

good luck!

greg k-h