Re: [PATCH 2/5] drivers: mtd: nand: Add qpic_common API file

[Md Sadre Alam]

On 2/15/2024 8:30 PM, Dmitry Baryshkov wrote:
> On Thu, 15 Feb 2024 at 15:53, Md Sadre Alam <quic_mdalam@xxxxxxxxxxx> wrote:
> > Add qpic_common.c file which hold all the common
> > qpic APIs which will be used by both qpic raw nand
> > driver and qpic spi nand driver.
> >
> > Co-developed-by: Sricharan Ramabadhran <quic_srichara@xxxxxxxxxxx>
> > Signed-off-by: Sricharan Ramabadhran <quic_srichara@xxxxxxxxxxx>
> > Co-developed-by: Varadarajan Narayanan <quic_varada@xxxxxxxxxxx>
> > Signed-off-by: Varadarajan Narayanan <quic_varada@xxxxxxxxxxx>
> > Signed-off-by: Md Sadre Alam <quic_mdalam@xxxxxxxxxxx>
> > ---
> >   drivers/mtd/nand/Makefile            |    1 +
> >   drivers/mtd/nand/qpic_common.c       |  786 +++++++++++++++++
> >   drivers/mtd/nand/raw/qcom_nandc.c    | 1226 +-------------------------
> >   include/linux/mtd/nand-qpic-common.h |  488 ++++++++++
> >   4 files changed, 1291 insertions(+), 1210 deletions(-)
> >   create mode 100644 drivers/mtd/nand/qpic_common.c
> >   create mode 100644 include/linux/mtd/nand-qpic-common.h
> >
> > diff --git a/drivers/mtd/nand/Makefile b/drivers/mtd/nand/Makefile
> > index 19e1291ac4d5..131707a41293 100644
> > --- a/drivers/mtd/nand/Makefile
> > +++ b/drivers/mtd/nand/Makefile
> > @@ -12,3 +12,4 @@ nandcore-$(CONFIG_MTD_NAND_ECC) += ecc.o
> >   nandcore-$(CONFIG_MTD_NAND_ECC_SW_HAMMING) += ecc-sw-hamming.o
> >   nandcore-$(CONFIG_MTD_NAND_ECC_SW_BCH) += ecc-sw-bch.o
> >   nandcore-$(CONFIG_MTD_NAND_ECC_MXIC) += ecc-mxic.o
> > +obj-y += qpic_common.o
> > diff --git a/drivers/mtd/nand/qpic_common.c b/drivers/mtd/nand/qpic_common.c
> > new file mode 100644
> > index 000000000000..4d74ba888028
> > --- /dev/null
> > +++ b/drivers/mtd/nand/qpic_common.c
> > @@ -0,0 +1,786 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * QPIC Controller common API file.
> > + * Copyright (C) 2023  Qualcomm Inc.
> > + * Authors:    Md sadre Alam           <quic_mdalam@xxxxxxxxxxx>
> > + *             Sricharan R             <quic_srichara@xxxxxxxxxxx>
> > + *             Varadarajan Narayanan   <quic_varada@xxxxxxxxxxx>
>
> This is a bit of an exaggeration. You are moving code, not writing new
> code. Please retain the existing copyrights for the moved code.
Ok
> > + *
> > + */
> > +
> > +#include <linux/mtd/nand-qpic-common.h>
> > +
> > +struct qcom_nand_controller *
> > +get_qcom_nand_controller(struct nand_chip *chip)
> > +{
> > +       return container_of(chip->controller, struct qcom_nand_controller,
> > +                           controller);
> > +}
> > +EXPORT_SYMBOL(get_qcom_nand_controller);
>
> NAK for adding functions to the global export namespace without a
> proper driver-specific prefix.
Ok, will fix in next patch
> Also, a bunch of the code here seems not so well thought. It was fine
> for an internal interface, but it doesn't look so good as a common
> wrapper. Please consider defining a sensible common code module
> interface instead.

 QPIC controller will support both raw NAND as well Serial nand interface.
 This common API file was the part of raw NAND driver , since for serial
 nand most of the APIs from raw nand driver will be re-used that's why i
 have created this common API file, so that QPIC serial nand driver
 drivers/spi/spi-qpic-snand.c and QPIC raw NAND driver
 drivers/mtd/nand/raw/qcom_nandc.c can used these common APIs.

 Could you please suggest how I should handle this in batter way.
> At least each function that is being exported should get a kerneldoc.
Ok
> Last, but not least, please use EXPORT_SYMBOL_GPL.
Ok
> > +
> > +/*
> > + * Helper to prepare DMA descriptors for configuring registers
> > + * before reading a NAND page.
> > + */
> > +void config_nand_page_read(struct nand_chip *chip)
> > +{
> > +       struct qcom_nand_controller *nandc = get_qcom_nand_controller(chip);
> > +
> > +       write_reg_dma(nandc, NAND_ADDR0, 2, 0);
> > +       write_reg_dma(nandc, NAND_DEV0_CFG0, 3, 0);
> > +       if (!nandc->props->qpic_v2)
> > +               write_reg_dma(nandc, NAND_EBI2_ECC_BUF_CFG, 1, 0);
> > +       write_reg_dma(nandc, NAND_ERASED_CW_DETECT_CFG, 1, 0);
> > +       write_reg_dma(nandc, NAND_ERASED_CW_DETECT_CFG, 1,
> > +                     NAND_ERASED_CW_SET | NAND_BAM_NEXT_SGL);
> > +}
> > +EXPORT_SYMBOL(config_nand_page_read);
> > +
> > +/* Frees the BAM transaction memory */
> > +void free_bam_transaction(struct qcom_nand_controller *nandc)
> > +{
> > +       struct bam_transaction *bam_txn = nandc->bam_txn;
> > +
> > +       devm_kfree(nandc->dev, bam_txn);
>
> devm_kfree is usually a bad sign. Either the devm_kfree should be
> dropped (because the memory area is allocated only during probe / init
> and doesn't need to be freed manually) or use kalloc/kfree directly
> without devres wrapping.
Ok
> > +}
> > +EXPORT_SYMBOL(free_bam_transaction);
> > +
>
> [skipped the rest]
>
> > --
> > 2.34.1
>
>
> --
> With best wishes
> Dmitry

Thanks for reviewing, Will fix all the comments in next patch.

Regards,
Alam.