From: James Hogan <james.hogan@imgtec.com>
To: Shashidhar Hiremath <shashidharh@vayavyalabs.com>
Cc: Chris Ball <cjb@laptop.org>, Shawn Guo <shawn.guo@linaro.org>,
Wolfram Sang <w.sang@pengutronix.de>,
Philip Rakity <prakity@marvell.com>,
Zhangfei Gao <zhangfei.gao@marvell.com>,
Will Newton <will.newton@imgtec.com>,
Jaehoon Chung <jh80.chung@samsung.com>,
Kyungmin Park <kyungmin.park@samsung.com>,
linux-mmc@vger.kernel.org
Subject: Re: [PATCH v2 1/1] mmc: Support of DUAL BUFFER DESC[ring] mode for dw_mmc
Date: Tue, 30 Aug 2011 09:56:01 +0100 [thread overview]
Message-ID: <4E5CA5A1.1060901@imgtec.com> (raw)
In-Reply-To: <1314692959-4139-1-git-send-email-shashidharh@vayavyalabs.com>
Hi Shashidhar
Thanks for the updated patch.
On 08/30/2011 09:29 AM, Shashidhar Hiremath wrote:
> This Patch adds the support for Dual Buffer Descriptor mode of
> Operation for the dw_mmc driver.The patch also provides the configurability
> Option for choosing DUAL_BUFFER mode or the chained modes through menuconfig.
> The Menuconfig option for selecting Dual Buffer mode or chained mode
> is selected only if Internal DMAC is enabled.
>
> Signed-off-by: Shashidhar Hiremath <shashidharh@vayavyalabs.com>
> ---
> v2:
> * As per suggestions by Will Newton and James Hogan
> -Config symbol Names prefixed with MMC_DW
> -Added more Description for Config parameters added
> -Removed unnecessary config opion IDMAC_DESC_MODE
> -IDMAC_SET_BUFFER_SIZE chenged IDMAC_SET_BUFFER_SIZES
> -fixed typos and indented commments correctly
> -if ((i +1) <= sg_len changed to ((i +1) < sg_len
> -duplication "desc->des0 = IDMAC_DES0_OWN | IDMAC_DES0_DIC" line removed
> -fixed bug in making DSL value zero
> -removed ANDing the des1 with zero before writing buffer lengths to it
> -Added proper multiline comments
> ---
> drivers/mmc/host/Kconfig | 19 +++++++++++++++++
> drivers/mmc/host/dw_mmc.c | 49 +++++++++++++++++++++++++++++++++++---------
> drivers/mmc/host/dw_mmc.h | 2 +
> 3 files changed, 60 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/mmc/host/Kconfig b/drivers/mmc/host/Kconfig
> index 8c87096..1e20dfa 100644
> --- a/drivers/mmc/host/Kconfig
> +++ b/drivers/mmc/host/Kconfig
> @@ -534,6 +534,25 @@ config MMC_DW_IDMAC
> Designware Mobile Storage IP block. This disables the external DMA
> interface.
>
> +choice
> + prompt "select IDMAC Descriptors Mode"
> + depends on MMC_DW_IDMAC
> +
> +config MMC_DW_CHAIN_DESC
> + bool "Chain Descriptor Structure"
> + help
> + Select this option to enable Chained Mode of Operation and the
> + Chained Mode operates in a mode where only one Buffer will be used
> + for each descriptor when the transfer is happening in DMA mode.
> +
> +config MMC_DW_DUAL_BUFFER_DESC
> + bool "Dual Buffer Descriptor Structure"
> + help
> + Select this option to enable Dual Buffer Desc Mode of Operation and
> + Dual Buffer Descriptor Mode or the Ring Mode indicates that two
> + buffers can be used for each descriptor during DMA mode transfer.
> +endchoice
> +
> config MMC_SH_MMCIF
> tristate "SuperH Internal MMCIF support"
> depends on MMC_BLOCK && (SUPERH || ARCH_SHMOBILE)
> diff --git a/drivers/mmc/host/dw_mmc.c b/drivers/mmc/host/dw_mmc.c
> index ff0f714..ee02208 100644
> --- a/drivers/mmc/host/dw_mmc.c
> +++ b/drivers/mmc/host/dw_mmc.c
> @@ -63,6 +63,9 @@ struct idmac_desc {
> u32 des1; /* Buffer sizes */
> #define IDMAC_SET_BUFFER1_SIZE(d, s) \
> ((d)->des1 = ((d)->des1 & 0x03ffe000) | ((s) & 0x1fff))
> +#define IDMAC_SET_BUFFER_SIZES(d, s1, s2) \
> + ((d)->des1 = (((d)->des1) | \
> + (((s1) & 0x1fff) | ((s2) & 0x1fff << 13))))
This still doesn't look right. The fields which get OR'd in don't get
cleared first, so if there is already a value you'll get the OR of the
old and new value. I think you were correct before in masking with 0,
but that means you can entirely drop the reference to (d)->des1, since
the new value doesn't depend on the old value.
>
> u32 des2; /* buffer 1 physical address */
>
> @@ -348,17 +351,38 @@ static void dw_mci_translate_sglist(struct dw_mci *host, struct mmc_data *data,
> struct idmac_desc *desc = host->sg_cpu;
>
> for (i = 0; i < sg_len; i++, desc++) {
> - unsigned int length = sg_dma_len(&data->sg[i]);
> - u32 mem_addr = sg_dma_address(&data->sg[i]);
> + /* Length and mem_address of first buffer */
> + unsigned int length1 = sg_dma_len(&data->sg[i]);
> + u32 mem_addr1 = sg_dma_address(&data->sg[i]);
> +#ifdef CONFIG_DUAL_BUFFER_DESC
don't forget to rename the configs here too.
> + unsigned int length2;
> + u32 mem_addr2;
>
> - /* Set the OWN bit and disable interrupts for this descriptor */
> + /*
> + * Set the OWN bit and disable interrupts
> + * for this descriptor
> + */
> + desc->des0 = IDMAC_DES0_OWN | IDMAC_DES0_DIC
> + if ((i+1) < sg_len) {
> + length2 = sg_dma_len(&data->sg[i+1]);
> + mem_addr2 = sg_dma_address(&data->sg[i+1]);
> + /* Buffer length being set for Buffer1 and Buffer2 */
> + IDMAC_SET_BUFFER_SIZES(desc, length1, length2);
> + desc->des3 = mem_addr2;
> + /* Incrementing for the second buffer */
> + i++;
> + } else {
> + /* Buffer length being set for Buffer1 and Buffer2 */
> + IDMAC_SET_BUFFER_SIZES(desc, length1, 0);
> + }
> +#elif CONFIG_CHAIN_DESC
here too
> + /* Set OWN bit and disable interrupts for this descriptor */
> desc->des0 = IDMAC_DES0_OWN | IDMAC_DES0_DIC | IDMAC_DES0_CH;
> -
> - /* Buffer length */
> - IDMAC_SET_BUFFER1_SIZE(desc, length);
> -
> + /* Buffer length for Buffer1 */
> + IDMAC_SET_BUFFER1_SIZE(desc, length1);
> +#endif
> /* Physical address to DMA to/from */
> - desc->des2 = mem_addr;
> + desc->des2 = mem_addr1;
> }
>
> /* Set first descriptor */
> @@ -369,6 +393,10 @@ static void dw_mci_translate_sglist(struct dw_mci *host, struct mmc_data *data,
> desc = host->sg_cpu + (i - 1) * sizeof(struct idmac_desc);
> desc->des0 &= ~(IDMAC_DES0_CH | IDMAC_DES0_DIC);
> desc->des0 |= IDMAC_DES0_LD;
> +#ifdef CONFIG_DUAL_BUFFER_DESC
and here
> + /* Set the End of Ring bit */
> + desc->des0 |= IDMAC_DES0_ER;
> +#endif
>
> wmb();
> }
> @@ -388,7 +416,8 @@ static void dw_mci_idmac_start_dma(struct dw_mci *host, unsigned int sg_len)
>
> /* Enable the IDMAC */
> temp = mci_readl(host, BMOD);
> - temp |= SDMMC_IDMAC_ENABLE | SDMMC_IDMAC_FB;
> + /* The Descriptor Skip length is made zero */
> + temp |= SDMMC_IDMAC_ENABLE | SDMMC_IDMAC_FB | SDMMC_BMOD_DSL(0);
this won't actually set the DSL to zero (you cannot clear bits with a |=
operator. You probably want to clear the DSL bits in temp, and then OR
in the new DSL value (or not in the case of setting it to 0).
> mci_writel(host, BMOD, temp);
>
> /* Start it running */
> diff --git a/drivers/mmc/host/dw_mmc.h b/drivers/mmc/host/dw_mmc.h
> index 027d377..0520dc8 100644
> --- a/drivers/mmc/host/dw_mmc.h
> +++ b/drivers/mmc/host/dw_mmc.h
> @@ -72,6 +72,8 @@
> /* Clock Enable register defines */
> #define SDMMC_CLKEN_LOW_PWR BIT(16)
> #define SDMMC_CLKEN_ENABLE BIT(0)
> +/* BMOD register defines */
> +#define SDMMC_BMOD_DSL(n) _SBF(2, (n))
> /* time-out register defines */
> #define SDMMC_TMOUT_DATA(n) _SBF(8, (n))
> #define SDMMC_TMOUT_DATA_MSK 0xFFFFFF00
Cheers
James
next prev parent reply other threads:[~2011-08-30 8:56 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-08-30 8:29 [PATCH v2 1/1] mmc: Support of DUAL BUFFER DESC[ring] mode for dw_mmc Shashidhar Hiremath
2011-08-30 8:35 ` Kyungmin Park
2011-08-30 9:23 ` Shashidhar Hiremath
2011-08-30 13:35 ` James Hogan
2011-08-30 13:55 ` Shashidhar Hiremath
2011-08-30 8:56 ` James Hogan [this message]
2011-08-30 9:40 ` Shashidhar Hiremath
2011-08-30 9:47 ` James Hogan
2011-08-30 9:50 ` Shashidhar Hiremath
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=4E5CA5A1.1060901@imgtec.com \
--to=james.hogan@imgtec.com \
--cc=cjb@laptop.org \
--cc=jh80.chung@samsung.com \
--cc=kyungmin.park@samsung.com \
--cc=linux-mmc@vger.kernel.org \
--cc=prakity@marvell.com \
--cc=shashidharh@vayavyalabs.com \
--cc=shawn.guo@linaro.org \
--cc=w.sang@pengutronix.de \
--cc=will.newton@imgtec.com \
--cc=zhangfei.gao@marvell.com \
/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