linux-kernel - Re: [PATCH 0/2] Expand oscillator stop flag (OSF) validity check to ds1341

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aGMVtfwB5gmBuW4T@redbud>
Date: Mon, 30 Jun 2025 17:54:45 -0500
From: Tyler Hicks <code@...icks.com>
To: Meagan Lloyd <meaganlloyd@...ux.microsoft.com>,
	alexandre.belloni@...tlin.com,
	Rodolfo Giometti <giometti@...eenne.com>
Cc: linux-rtc@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 0/2] Expand oscillator stop flag (OSF) validity check to
 ds1341

[Adding Rodolfo Giometti]

On 2025-06-11 11:14:14, Meagan Lloyd wrote:
> We would like to use CONFIG_RTC_HCTOSYS to sync a supercapacitor-backed
> DS1342 RTC to the kernel time early in boot. An obstacle is that the
> sync in rtc_hctosys() is unconditional as long as rtc_read_time()
> succeeds and in some power loss situations, our RTC comes up with either
> an unpredictable future time or the default 01/01/00 from the datasheet.
> Syncing a future time, followed by an NTP sync would not be desired as
> it would result in a backwards time jump. The sync feature is useful in
> boot scenarios where power is maintained so syncing only when the RTC
> data is valid would allow us to make use of the feature.
> 
> The DS1342 has the oscillator stop flag (OSF) which is a status flag
> indicating that the oscillator stopped for a period of time. It can be
> set due to power loss. Some chip types in the ds1307 driver already use
> the OSF to determine whether .read_time should provide valid data or
> return -EINVAL. This patch series expands that handling to the ds1341
> chip type (DS1341 and DS1342 share a datasheet).
> 
> These changes enable us to make use of CONFIG_RTC_HCTOSYS as they
> prevent the invalid time from getting synced to the kernel time. It will
> also prevent userspace programs from getting the invalid time as the fix
> cuts it off at the source - the .read_time function.

These two patches look good to me, although I'm not an expert in RTC drivers.
I've reviewed the DS1341/DS1342 datasheet and the approach that Meagan has
taken makes sense to me given our (Meagan and I work together) desire to use
CONFIG_RTC_HCTOSYS and the need to avoid syncing from an invalid RTC state.

I've added Rodolfo because he first added the logic to clear the Oscillator
Stop Flag, during driver initialization, way back in 2007 with v2.6.23 commit
be5f59f4b67f ("rtc-ds1307: oscillator restart for ds13{37,38,39,40}") and may
have additional context to provide.

Alexandre and Rodolfo, does this approach make sense to you? If not, do you
have any other suggestions on how to make CONFIG_RTC_HCTOSYS work with this
driver? Thanks!

Tyler

> 
> Meagan Lloyd (2):
>   rtc: ds1307: remove clear of oscillator stop flag (OSF) in probe
>   rtc: ds1307: handle oscillator stop flag (OSF) for ds1341
> 
>  drivers/rtc/rtc-ds1307.c | 15 ++++++++++++---
>  1 file changed, 12 insertions(+), 3 deletions(-)
> 
> 
> base-commit: 19272b37aa4f83ca52bdf9c16d5d81bdd1354494
> -- 
> 2.49.0
>