[PATCH] driver core: fw_devlink: Don't warn in fw_devlink_dev_sync_state()

[Ulf Hansson]

On Thu, 25 Sept 2025 at 14:08, Tomi Valkeinen
<tomi.valkeinen at ideasonboard.com> wrote:
>
>
>
> On 25/09/2025 14:59, Ulf Hansson wrote:
> > Due to the wider deployment of the ->sync_state() support, for PM domains
> > for example, we are receiving reports about the messages that are being
> > logged in fw_devlink_dev_sync_state(). In particular as they are at the
> > warning level, which doesn't seem correct.
> >
> > Even if it certainly is useful to know that the ->sync_state() condition
> > could not be met, there may be nothing wrong with it. For example, a driver
> > may be built as module and are still waiting to be initialized/probed.
> >
> > Ideally these messages should be at the debug level, but since the
> > ->sync_state() feature is under an ongoing deployment and the prints
> > provides valuable information, let's move to the info level for now.
> >
> > Cc: Saravana Kannan <saravanak at google.com>
> > Reported-by: Geert Uytterhoeven <geert at linux-m68k.org>
> > Reported-by: Sebin Francis <sebin.francis at ti.com>
> > Reported-by: Diederik de Haas <didi.debian at cknow.org>
> > Reported-by: Jon Hunter <jonathanh at nvidia.com>
> > Signed-off-by: Ulf Hansson <ulf.hansson at linaro.org>
> > ---
> >  drivers/base/core.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/base/core.c b/drivers/base/core.c
> > index d22d6b23e758..97eab79c2f3b 100644
> > --- a/drivers/base/core.c
> > +++ b/drivers/base/core.c
> > @@ -1784,7 +1784,7 @@ static int fw_devlink_dev_sync_state(struct device *dev, void *data)
> >               return 0;
> >
> >       if (fw_devlink_sync_state == FW_DEVLINK_SYNC_STATE_STRICT) {
> > -             dev_warn(sup, "sync_state() pending due to %s\n",
> > +             dev_info(sup, "sync_state() pending due to %s\n",
> >                        dev_name(link->consumer));
> >               return 0;
> >       }
> > @@ -1792,7 +1792,7 @@ static int fw_devlink_dev_sync_state(struct device *dev, void *data)
> >       if (!list_empty(&sup->links.defer_sync))
> >               return 0;
> >
> > -     dev_warn(sup, "Timed out. Forcing sync_state()\n");
> > +     dev_info(sup, "Timed out. Forcing sync_state()\n");
>
> I have no issue with this, but I also think that while the pending print
> above could well be dev_dbg, this one is perhaps a bit more warning-ish.
> It may be harmless to get the time-out, but it would be better not to
> time-out (i.e. everything was already sync_stated, or startup scripts
> handled forcing the sync state).

I agree. Perhaps we should consider moving the above "sync_state()
pending.." to dev_dbg at some later point, then we should probably
keep the "Timed out..." print at dev_info().

Anyway, I don't think using dev_warn makes sense, at least for
configurations with CONFIG_FW_DEVLINK_SYNC_STATE_TIMEOUT being set.

>
> >       sup->state_synced = true;
> >       get_device(sup);
> >       list_add_tail(&sup->links.defer_sync, data);
>
> Reviewed-by: Tomi Valkeinen <tomi.valkeinen at ideasonboard.com>
>
>  Tomi
>

Thanks for reviewing!

Kind regards
Uffe