Re: [PATCH v2 3/4] Documentation: bootchooser: add information about attempts_locked

[Sascha Hauer <s.hauer@pengutronix.de>]

On Tue, Jun 17, 2025 at 11:00:10AM +0200, Lars Schmidt wrote:
> 
> On 17.06.25 10:42, Sascha Hauer wrote:
> > On Mon, Jun 16, 2025 at 05:06:56PM +0200, Lars Schmidt wrote:
> >> The new attempts_locked variable does have influence on
> >> remaining_attempts counter.
> >>
> >> While at it, also make mention in the bootstate.dtsi.
> >>
> >> Signed-off-by: Lars Schmidt <l.schmidt@pengutronix.de>
> >> ---
> >>  Documentation/user/bootchooser.rst | 23 +++++++++++++++++++++++
> >>  arch/arm/dts/bootstate.dtsi        |  5 +++++
> >>  2 files changed, 28 insertions(+)
> >>
> >> diff --git a/Documentation/user/bootchooser.rst b/Documentation/user/bootchooser.rst
> >> index 351e1d14e1ead6ba8d329c03c0bc7ed28b523df0..ab0b6c3fc895226f5aa4590944a2cd675a56d985 100644
> >> --- a/Documentation/user/bootchooser.rst
> >> +++ b/Documentation/user/bootchooser.rst
> >> @@ -77,6 +77,16 @@ no remaining attempts left.
> >>  To prevent ending up in an unbootable system after a number of failed boot
> >>  attempts, there is also a built-in mechanism to reset the counters to their defaults,
> >>  controlled by the ``global.bootchooser.reset_attempts`` variable.
> >> +Alternatively, counting down the remaining attempts can be disabled by
> >> +locking bootchooser boot attempts.
> >> +This is done by defining a (32-bit) ``attempts_locked`` variable in the
> >> +bootstate and setting its value to ``1`` (usually from userspace).
> >> +
> >> +The variable affects all targets, is optional and its absence is
> >> +interpreted as ``0``, meaning that attempts are decremented normally.
> >> +
> >> +The ``attempts_locked`` value does not influence the decision on which target
> >> +to boot if any, only whether to decrement the attempts when booting.
> > This describes what it does, but not what it's good for. Could you add a
> > few sentences about it?
> 
> I was hoping this is understandable if you read in the context of the already
> existing paragraph right before that: "To prevent ending up in an unbootable system....."
> As it is meant to be an alternative to global.bootchooser.reset_attempts
> 
> I didn't want to get too repetitive, but if you want, I can try to rephrase it.

How I understand it the purpose of the attempts_locked variable is that
I can mark a slot as always-use-this-one.

There are some differences to the reset_attempts approach. With
reset_attempts barebox decreases the counter and Linux increases it
again. When a board is power cycled often enough (due to an impatient
user or weak power supply) after barebox has decreased the counter but
before Linux has increased it again then we'll fall back to another slot.
This might be undesired in some cases and the attempts_lock doesn't have
this issue, but it comes with the price that a slot that gets broken
over time (filesystem corruption or the like) won't be detected anymore
and will be booted forever.

Sascha

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |