Re: [PATCH 0/3] mtd: core: protect access to mtd devices while in suspend

From: Miquel Raynal <miquel.raynal@bootlin.com>
To: Sean Nyekjaer <sean@geanix.com>
Cc: Boris Brezillon <boris.brezillon@collabora.com>,
	Richard Weinberger <richard@nod.at>,
	Vignesh Raghavendra <vigneshr@ti.com>,
	Boris Brezillon <bbrezillon@kernel.org>,
	linux-mtd@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 0/3] mtd: core: protect access to mtd devices while in suspend
Date: Wed, 20 Oct 2021 09:23:31 +0200	[thread overview]
Message-ID: <20211020092331.68e46e16@xps13> (raw)
In-Reply-To: <20211020071235.in3omswo2jqrahrd@skn-laptop>

Hi Sean,

sean@geanix.com wrote on Wed, 20 Oct 2021 09:12:35 +0200:

> On Wed, Oct 20, 2021 at 09:00:58AM +0200, Boris Brezillon wrote:
> > On Wed, 20 Oct 2021 08:52:50 +0200
> > Boris Brezillon <boris.brezillon@collabora.com> wrote:
> >   
> 
> [ ... ]
> > > > 
> > > > Hi Boris and Miquel,
> > > > 
> > > > gpmi-nand.c sets NAND_SKIP_BBTSCAN so we won't get there and populate
> > > > suspend resume hooks :(
> > > > Guess there is other drivers that does the same thing...    
> > 
> > Actually, this version is even cleaner:
> > 
> > diff --git a/drivers/mtd/nand/raw/nand_base.c b/drivers/mtd/nand/raw/nand_base.c
> > index 3d6c6e880520..98c39b7f6279 100644
> > --- a/drivers/mtd/nand/raw/nand_base.c
> > +++ b/drivers/mtd/nand/raw/nand_base.c
> > @@ -6222,8 +6222,6 @@ static int nand_scan_tail(struct nand_chip *chip)
> >         mtd->_sync = nand_sync;
> >         mtd->_lock = nand_lock;
> >         mtd->_unlock = nand_unlock;
> > -       mtd->_suspend = nand_suspend;
> > -       mtd->_resume = nand_resume;
> >         mtd->_reboot = nand_shutdown;
> >         mtd->_block_isreserved = nand_block_isreserved;
> >         mtd->_block_isbad = nand_block_isbad;
> > @@ -6261,14 +6259,20 @@ static int nand_scan_tail(struct nand_chip *chip)
> >                 goto err_free_interface_config;
> >  
> >         /* Check, if we should skip the bad block table scan */
> > -       if (chip->options & NAND_SKIP_BBTSCAN)
> > -               return 0;
> > -
> > -       /* Build bad block table */
> > -       ret = nand_create_bbt(chip);
> > -       if (ret)
> > -               goto err_free_secure_regions;
> > +       if (chip->options & NAND_SKIP_BBTSCAN) {
> > +               /* Build bad block table */
> > +               ret = nand_create_bbt(chip);
> > +               if (ret)
> > +                       goto err_free_secure_regions;
> > +       }
> >  
> > +       /*
> > +        * Populate the suspend/resume hooks after the BBT has been scanned to
> > +        * avoid using the suspend lock and resume waitqueue which are only
> > +        * initialized when mtd_device_register() is called.
> > +        */
> > +       mtd->_suspend = nand_suspend;
> > +       mtd->_resume = nand_resume;
> >         return 0;
> >  
> >  err_free_secure_regions:  
> 
> Why is the gpmi-nand.c and other drivers set NAND_SKIP_BBTSCAN and then
> call nand_create_bbt() directly?
> 
> To me it looks like legacy leftover...

It is...

885d71e5838f ("mtd: gpmi: fix the NULL pointer") added this flag to
avoid a NULL pointer dereference on chip->buffer->databuf on mx23,
which IIRC does not exist anymore.

I believe we could get rid of it. But anyway, I like the "cleanup"
proposed by Boris so IMHO you can add these two patches to the series.

Thanks,
Miquèl