On Tue, Apr 04, 2023 at 10:01:42AM +0200, Marco Felsch wrote:
> On 23-04-04, Sascha Hauer wrote:
> > On Mon, Mar 13, 2023 at 02:41:41PM +0100, Marco Felsch wrote:
> > > It is possible that one barebox can rule multiple platforms e.g.
> > > multiple platform revision each has a dedicated device tree and of
> > > compatible. While it is possible to add multiple of-compatibles via
> > > IMD_USED_OF() (see: zii-imx8mq-dev) we didn't allowed it to flash it
> > > without 'force'. Fix this by check all possible IMD_TYPE_OF_COMPATIBLE
> > > image metadata types.
> > >
> > > Signed-off-by: Marco Felsch <m.felsch@xxxxxxxxxxxxxx>
> > > ---
> > > common/bbu.c | 42 +++++++++++++++++++++++++++---------------
> > > 1 file changed, 27 insertions(+), 15 deletions(-)
> > >
> > > diff --git a/common/bbu.c b/common/bbu.c
> > > index 3ec17216cb..a9b3fef7a0 100644
> > > --- a/common/bbu.c
> > > +++ b/common/bbu.c
> > > @@ -154,36 +154,44 @@ struct bbu_handler *bbu_find_handler_by_device(const char *devicepath)
> > > return NULL;
> > > }
> > >
> > > -static int bbu_check_of_compat(struct bbu_data *data)
> > > +static int bbu_check_of_compat(struct bbu_data *data, unsigned short of_compat_nr)
> > > {
> > > + const struct imd_header *imd = data->imd_data;
> > > + const struct imd_header *of_compat;
> > > struct device_node *root_node;
> > > const char *machine, *str;
> > > int ret;
> > > - const struct imd_header *of_compat;
> > >
> > > if (!IS_ENABLED(CONFIG_OFDEVICE) || !IS_ENABLED(CONFIG_IMD))
> > > return 0;
> >
> > This patch breaks barebox update when the update image doesn't contain
> > any IMD_TYPE_OF_COMPATIBLE entries. To fix this we need this here:
> >
> > if (!of_compat_nr)
> > return 0;
> >
> > I fixed this up.
>
> Argh.. thanks :) To make even more robust and forcable by board we could
> replace it with a dedicated CONFIG_IMD_OF_COMPATIBLE.
Making this a compile time option is not a good idea. Should be runtime.
> This way boards
> can select it to enforce the check or we skip the check. Since just
> checking:
>
> if (!of_comat_nr)
> return 0;
>
> could also hide board code issues.
Indeed, yes. Ahmad has a patch in the queue which automatically adds the
IMD_TYPE_OF_COMPATIBLE entries for each board based on the device tree
that is used in the image. I suggest we wait for this patch, and with it
we can re-evaluate this.
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 |
