Re: [PATCH 2/2] mtd: rawnand: use bounce buffer when vmalloced data buf detected

From: Boris Brezillon <boris.brezillon@collabora.com>
To: Kamal Dasu <kdasu.kdev@gmail.com>
Cc: Vignesh Raghavendra <vigneshr@ti.com>,
	Boris Brezillon <bbrezillon@kernel.org>,
	Richard Weinberger <richard@nod.at>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Frieder Schrempf <frieder.schrempf@kontron.de>,
	Marek Vasut <marek.vasut@gmail.com>,
	MTD Maling List <linux-mtd@lists.infradead.org>,
	Miquel Raynal <miquel.raynal@bootlin.com>,
	bcm-kernel-feedback-list@broadcom.com,
	Brian Norris <computersforpeace@gmail.com>,
	David Woodhouse <dwmw2@infradead.org>
Subject: Re: [PATCH 2/2] mtd: rawnand: use bounce buffer when vmalloced data buf detected
Date: Mon, 30 Sep 2019 18:24:58 +0200	[thread overview]
Message-ID: <20190930182458.761e8077@collabora.com> (raw)
In-Reply-To: <CAC=U0a1qvKO+t_62df_JcQBETAuNq0pwRkAb-Ofi3ski2rfdEQ@mail.gmail.com>

On Mon, 30 Sep 2019 12:01:28 -0400
Kamal Dasu <kdasu.kdev@gmail.com> wrote:

> Does anyone have any comments on this patch ?.
> 
> Kamal
> 
> On Fri, Sep 6, 2019 at 3:49 PM Kamal Dasu <kdasu.kdev@gmail.com> wrote:
> >
> > For controller drivers that use DMA and set NAND_USE_BOUNCE_BUFFER
> > option use data buffers that are not vmalloced, aligned and have
> > valid virtual address to be able to do DMA transfers. This change
> > adds additional check and makes use of data buffer allocated
> > in nand_base driver when it is passed a vmalloced data buffer for
> > DMA transfers.
> >
> > Signed-off-by: Kamal Dasu <kdasu.kdev@gmail.com>
> > ---
> >  drivers/mtd/nand/raw/nand_base.c | 14 ++++++++------
> >  1 file changed, 8 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/mtd/nand/raw/nand_base.c b/drivers/mtd/nand/raw/nand_base.c
> > index 91f046d4d452..46f6965a896a 100644
> > --- a/drivers/mtd/nand/raw/nand_base.c
> > +++ b/drivers/mtd/nand/raw/nand_base.c
> > @@ -45,6 +45,12 @@
> >
> >  #include "internals.h"
> >
> > +static int nand_need_bounce_buf(const void *buf, struct nand_chip *chip)
> > +{
> > +       return !virt_addr_valid(buf) || is_vmalloc_addr(buf) ||

I thought virt_addr_valid() was implying !is_vmalloc_addr(), are you
sure you need that test, and can you tell me on which arch(es) this is
needed.

> > +               !IS_ALIGNED((unsigned long)buf, chip->buf_align);
> > +}
> > +
> >  /* Define default oob placement schemes for large and small page devices */
> >  static int nand_ooblayout_ecc_sp(struct mtd_info *mtd, int section,
> >                                  struct mtd_oob_region *oobregion)
> > @@ -3183,9 +3189,7 @@ static int nand_do_read_ops(struct nand_chip *chip, loff_t from,
> >                 if (!aligned)
> >                         use_bufpoi = 1;
> >                 else if (chip->options & NAND_USE_BOUNCE_BUFFER)
> > -                       use_bufpoi = !virt_addr_valid(buf) ||
> > -                                    !IS_ALIGNED((unsigned long)buf,
> > -                                                chip->buf_align);
> > +                       use_bufpoi = nand_need_bounce_buf(buf, chip);
> >                 else
> >                         use_bufpoi = 0;
> >
> > @@ -4009,9 +4013,7 @@ static int nand_do_write_ops(struct nand_chip *chip, loff_t to,
> >                 if (part_pagewr)
> >                         use_bufpoi = 1;
> >                 else if (chip->options & NAND_USE_BOUNCE_BUFFER)
> > -                       use_bufpoi = !virt_addr_valid(buf) ||
> > -                                    !IS_ALIGNED((unsigned long)buf,
> > -                                                chip->buf_align);
> > +                       use_bufpoi = nand_need_bounce_buf(buf, chip);
> >                 else
> >                         use_bufpoi = 0;
> >
> > --
> > 2.17.1
> >  

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/