From mboxrd@z Thu Jan  1 00:00:00 1970
From: Boris Brezillon <boris.brezillon@bootlin.com>
Date: Mon, 19 Nov 2018 12:57:51 +0100
Subject: [U-Boot] [PATCH 3/4] mtd: Delete partitions attached to the
 device when a device is deleted
In-Reply-To: <0776b30a-fce7-010f-5903-0faf9bb0d0e2@denx.de>
References: <20181116144026.3015-1-boris.brezillon@bootlin.com>
 <20181116144026.3015-3-boris.brezillon@bootlin.com>
 <20181117101912.55eaccc6@bbrezillon>
 <0776b30a-fce7-010f-5903-0faf9bb0d0e2@denx.de>
Message-ID: <20181119125751.44feb402@bbrezillon>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

On Mon, 19 Nov 2018 07:16:53 +0100
Heiko Schocher <hs@denx.de> wrote:

> Hello Boris,
> 
> Am 17.11.2018 um 10:19 schrieb Boris Brezillon:
> > On Fri, 16 Nov 2018 15:40:25 +0100
> > Boris Brezillon <boris.brezillon@bootlin.com> wrote:
> >   
> >> If we don't do that, partitions might still be exposed while the
> >> underlying device is gone.
> >>
> >> Fixes: 2a74930da57f ("mtd: mtdpart: implement proper partition handling")
> >> Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
> >> ---
> >>   drivers/mtd/mtdcore.c   |  1 +
> >>   include/linux/mtd/mtd.h | 14 ++++++++++++++
> >>   2 files changed, 15 insertions(+)
> >>
> >> diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
> >> index 7a15ded8c883..46657fe7c949 100644
> >> --- a/drivers/mtd/mtdcore.c
> >> +++ b/drivers/mtd/mtdcore.c
> >> @@ -528,6 +528,7 @@ int del_mtd_device(struct mtd_info *mtd)
> >>   	struct mtd_notifier *not;
> >>   #endif
> >>   
> >> +	del_mtd_partitions(mtd);
> >>   	mutex_lock(&mtd_table_mutex);
> >>   
> >>   	if (idr_find(&mtd_idr, mtd->index) != mtd) {
> >> diff --git a/include/linux/mtd/mtd.h b/include/linux/mtd/mtd.h
> >> index d20ebd820289..c5b58dd3f0f7 100644
> >> --- a/include/linux/mtd/mtd.h
> >> +++ b/include/linux/mtd/mtd.h
> >> @@ -562,8 +562,22 @@ unsigned mtd_mmap_capabilities(struct mtd_info *mtd);
> >>   /* drivers/mtd/mtdcore.h */
> >>   int add_mtd_device(struct mtd_info *mtd);
> >>   int del_mtd_device(struct mtd_info *mtd);
> >> +
> >> +#ifdef CONFIG_MTD_PARTITIONS
> >>   int add_mtd_partitions(struct mtd_info *, const struct mtd_partition *, int);
> >>   int del_mtd_partitions(struct mtd_info *);
> >> +#else
> >> +static inline int add_mtd_partitions(struct mtd_info *,
> >> +				     const struct mtd_partition *, int)  
> > 
> > Args should have names.
> >   
> >> +{
> >> +	return 0;
> >> +}
> >> +
> >> +static int del_mtd_partitions(struct mtd_info *)  
> > 
> > Missing inline here.
> > 
> > I'll send a v2 fixing those 2 bugs.  
> 
> Thanks!
> 
> I tried your patchset, with them "ubi part ubi" does now work, also
> after a "sf probe" ...
> 
> There is one problem, see log [1].
> 
> If you have an ubi partition attached (In my example on the NAND),
> and issue "sf probe", the following "mtd list" shows not anymore
> the SPI NOR MTD partitions. (It prints an error message
> "Partition "ubi" already in use, aborting")
> 
> If I detach UBI from the NAND MTD partition, the MTD Partitions on
> the SPI NOR are again found after a "sf probe" before "mtd list" [2]

I sent you a new version of this patchset that should address your
issue. Note that updating parts on a device that is being used is still
forbidden, but at least it does not block updates on other devices.

Oh, and I keep thinking the spi-flash code is broken, and none of this
should happen if the spi-nor MTD devs were staying around instead of
being unregistered/registered every time sf probe is called. But I
don't intend to fix it now, as the proper solution is probably to port
the spi-nor layer we have in Linux to u-boot.