linux-mmc.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ulf Hansson <ulf.hansson@linaro.org>
To: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Cc: Russell King <linux@arm.linux.org.uk>,
	linux-mmc <linux-mmc@vger.kernel.org>,
	Chris Ball <chris@printf.net>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-arm-msm@vger.kernel.org" <linux-arm-msm@vger.kernel.org>,
	Linus Walleij <linus.walleij@linaro.org>
Subject: Re: [PATCH v4 13/13] mmc: mmci: Add Qcom specific pio_read function.
Date: Fri, 30 May 2014 13:27:13 +0200	[thread overview]
Message-ID: <CAPDyKFpto10x7ZFbozkmnPhmTH3dkENEJetD_c08bgzytKa_xw@mail.gmail.com> (raw)
In-Reply-To: <1401284886-16978-1-git-send-email-srinivas.kandagatla@linaro.org>

On 28 May 2014 15:48,  <srinivas.kandagatla@linaro.org> wrote:
> From: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
>
> MCIFIFOCNT register behaviour on Qcom chips is very different than the other
> pl180 integrations. MCIFIFOCNT register contains the number of
> words that are still waiting to be transferred through the FIFO. It keeps
> decrementing once the host CPU reads the MCIFIFO. With the existing logic and
> the MCIFIFOCNT behaviour, mmci_pio_read will loop forever, as the FIFOCNT
> register will always return transfer size before reading the FIFO.
>
> Also the data sheet states that "This register is only useful for debug
> purposes and should not be used for normal operation since it does not reflect
> data which may or may not be in the pipeline".
>
> This patch implements qcom_pio_read function so as existing mmci_pio_read is
> not suitable for Qcom SOCs. qcom_pio_read function is only selected
> based on qcom_fifo flag in variant data structure.
>
> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
> ---
>  drivers/mmc/host/mmci.c | 45 ++++++++++++++++++++++++++++++++++++++++++++-
>  drivers/mmc/host/mmci.h |  1 +
>  2 files changed, 45 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/mmc/host/mmci.c b/drivers/mmc/host/mmci.c
> index 6eb0a29..b68223a 100644
> --- a/drivers/mmc/host/mmci.c
> +++ b/drivers/mmc/host/mmci.c
> @@ -73,6 +73,7 @@ static unsigned int fmax = 515633;
>   * @busy_detect: true if busy detection on dat0 is supported
>   * @pwrreg_nopower: bits in MMCIPOWER don't controls ext. power supply
>   * @explicit_mclk_control: enable explicit mclk control in driver.
> + * @qcom_fifo: enables qcom specific fifo pio read function.
>   */
>  struct variant_data {
>         unsigned int            clkreg;
> @@ -95,6 +96,7 @@ struct variant_data {
>         bool                    busy_detect;
>         bool                    pwrreg_nopower;
>         bool                    explicit_mclk_control;
> +       bool                    qcom_fifo;
>  };
>
>  static struct variant_data variant_arm = {
> @@ -202,6 +204,7 @@ static struct variant_data variant_qcom = {
>         .pwrreg_powerup         = MCI_PWR_UP,
>         .f_max                  = 208000000,
>         .explicit_mclk_control  = true,
> +       .qcom_fifo              = true,
>  };
>
>  static int mmci_card_busy(struct mmc_host *mmc)
> @@ -1006,6 +1009,40 @@ mmci_cmd_irq(struct mmci_host *host, struct mmc_command *cmd,
>         }
>  }
>
> +static int mmci_qcom_pio_read(struct mmci_host *host, char *buffer,
> +                       unsigned int remain)
> +{
> +       u32 *ptr = (u32 *) buffer;

Instead of u32 ptr above, I suggest to use a char *ptr. You need it
anyway further down below where you move in step of bytes instead of
words.

> +       unsigned int count = 0;
> +       unsigned int words, bytes;
> +       unsigned int fsize = host->variant->fifosize;
> +
> +       words = remain >> 2;
> +       bytes = remain % 4;
> +       /* read full words followed by leftover bytes */
> +       if (words) {
> +               while (readl(host->base + MMCISTATUS) & MCI_RXDATAAVLBL) {

How about while (words && (readl(host->base + MMCISTATUS) & MCI_RXDATAAVLBL)

That would make it possible the remove the "if", both above and below.

> +                       *ptr = readl(host->base + MMCIFIFO + (count % fsize));

This looks strange. :-) Depending on the count you will read an offset
into the FIFO? Seems like a very awkward implementation of a FIFO in
the HW. :-)

BTW, what does "MCI_RXDATAAVLBL" actually mean for the Qcom variant?
How much data could you expect in the FIFO when this status is
triggered?

Are there no option of reading a number of words, depending on what
type FIFO IRQ that was raised?

> +                       ptr++;
> +                       count += 4;
> +                       words--;
> +                       if (!words)
> +                               break;
> +               }
> +       }
> +
> +       if (bytes) {
> +               unsigned char buf[4];
> +               if (readl(host->base + MMCISTATUS) & MCI_RXDATAAVLBL) {
> +                       *buf = readl(host->base + MMCIFIFO + (count % fsize));
> +                       memcpy(ptr, buf, bytes);
> +                       count += bytes;
> +               }
> +       }
> +
> +       return count;
> +}
> +
>  static int mmci_pio_read(struct mmci_host *host, char *buffer, unsigned int remain)
>  {
>         void __iomem *base = host->base;
> @@ -1129,7 +1166,8 @@ static irqreturn_t mmci_pio_irq(int irq, void *dev_id)
>
>                 len = 0;
>                 if (status & MCI_RXACTIVE)
> -                       len = mmci_pio_read(host, buffer, remain);
> +                       len = host->pio_read(host, buffer, remain);
> +
>                 if (status & MCI_TXACTIVE)
>                         len = mmci_pio_write(host, buffer, remain, status);
>
> @@ -1504,6 +1542,11 @@ static int mmci_probe(struct amba_device *dev,
>         if (ret)
>                 goto host_free;
>
> +       if (variant->qcom_fifo)
> +               host->pio_read = mmci_qcom_pio_read;
> +       else
> +               host->pio_read = mmci_pio_read;
> +
>         host->plat = plat;
>         host->variant = variant;
>         host->mclk = clk_get_rate(host->clk);
> diff --git a/drivers/mmc/host/mmci.h b/drivers/mmc/host/mmci.h
> index 1882e20..dc9fe0a 100644
> --- a/drivers/mmc/host/mmci.h
> +++ b/drivers/mmc/host/mmci.h
> @@ -229,6 +229,7 @@ struct mmci_host {
>         /* pio stuff */
>         struct sg_mapping_iter  sg_miter;
>         unsigned int            size;
> +       int (*pio_read)(struct mmci_host *h, char *buf, unsigned int remain);
>
>  #ifdef CONFIG_DMA_ENGINE
>         /* DMA stuff */
> --
> 1.9.1
>

Kind regards
Ulf Hansson

  reply	other threads:[~2014-05-30 11:27 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-05-28 13:43 [PATCH v4 00/13] Add Qualcomm SD Card Controller support srinivas.kandagatla
2014-05-28 13:45 ` [PATCH v4 01/13] mmc: mmci: use NSEC_PER_SEC macro srinivas.kandagatla
2014-05-28 13:46 ` [PATCH v4 02/13] mmc: mmci: convert register bits to use BIT() macro srinivas.kandagatla
2014-05-28 13:46 ` [PATCH v4 03/13] mmc: mmci: Add Qualcomm Id to amba id table srinivas.kandagatla
2014-05-30  9:39   ` Ulf Hansson
2014-05-30  9:49     ` Srinivas Kandagatla
2014-05-28 13:46 ` [PATCH v4 04/13] mmc: mmci: Add enough delay between writes to CMD register srinivas.kandagatla
2014-05-28 13:46 ` [PATCH v4 05/13] mmc: mmci: Add Qcom datactrl register variant srinivas.kandagatla
2014-05-28 13:46 ` [PATCH v4 06/13] mmc: mmci: add ddrmode mask to variant data srinivas.kandagatla
2014-05-30  9:35   ` Ulf Hansson
2014-05-30  9:50     ` Srinivas Kandagatla
2014-05-28 13:47 ` [PATCH v4 07/13] mmc: mmci: add 8bit bus support in " srinivas.kandagatla
2014-05-28 13:47 ` [PATCH v4 08/13] mmc: mmci: add edge support to data and command out " srinivas.kandagatla
2014-05-28 13:47 ` [PATCH v4 09/13] mmc: mmci: add Qcom specifics of clk and datactrl registers srinivas.kandagatla
2014-05-30  9:55   ` Ulf Hansson
2014-05-30  9:59     ` Srinivas Kandagatla
2014-05-28 13:47 ` [PATCH v4 10/13] mmc: mmci: Add support to data commands via variant structure srinivas.kandagatla
2014-05-28 13:47 ` [PATCH v4 11/13] mmc: mmci: add f_max to " srinivas.kandagatla
2014-05-30 10:28   ` Ulf Hansson
2014-05-30 10:29     ` Srinivas Kandagatla
2014-05-28 13:47 ` [PATCH v4 12/13] mmc: mmci: add explicit clk control srinivas.kandagatla
2014-05-30 10:25   ` Ulf Hansson
2014-05-30 10:39     ` Srinivas Kandagatla
2014-05-28 13:48 ` [PATCH v4 13/13] mmc: mmci: Add Qcom specific pio_read function srinivas.kandagatla
2014-05-30 11:27   ` Ulf Hansson [this message]
2014-05-30 11:44     ` Srinivas Kandagatla
2014-05-30 16:20       ` Srinivas Kandagatla

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CAPDyKFpto10x7ZFbozkmnPhmTH3dkENEJetD_c08bgzytKa_xw@mail.gmail.com \
    --to=ulf.hansson@linaro.org \
    --cc=chris@printf.net \
    --cc=linus.walleij@linaro.org \
    --cc=linux-arm-msm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mmc@vger.kernel.org \
    --cc=linux@arm.linux.org.uk \
    --cc=srinivas.kandagatla@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).