[PATCH 016/347] FogBugz #516535: Fix QSPI write issues

[PATCH 016/347] FogBugz #516535: Fix QSPI write issues

[Jit Loon Lim]

From: Chee Hong Ang <chee.hong.ang@intel.com>

QSPI driver perform chip select on every flash read/write
access. The driver need to disable/enable the QSPI controller
while performing chip select. This may cause some data lost
especially the QSPI controller is configured to run at slower
speed as it may take longer time to access the flash device.
This patch prevent the driver from disable/enable the QSPI
controller too soon and inadvertently halting any ongoing flash
read/write access by ensuring the QSPI controller is always in
idle mode after each read/write access.

Signed-off-by: Chee Hong Ang <chee.hong.ang@intel.com>
---
 drivers/spi/cadence_qspi_apb.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/drivers/spi/cadence_qspi_apb.c b/drivers/spi/cadence_qspi_apb.c
index 2cdf4c9c9f..5e03495f45 100644
--- a/drivers/spi/cadence_qspi_apb.c
+++ b/drivers/spi/cadence_qspi_apb.c
@@ -858,13 +858,9 @@ cadence_qspi_apb_indirect_read_execute(struct cadence_spi_plat *plat,
 	writel(CQSPI_REG_INDIRECTRD_DONE,
 	       plat->regbase + CQSPI_REG_INDIRECTRD);
 
-	/* Check indirect done status */
-	ret = wait_for_bit_le32(plat->regbase + CQSPI_REG_INDIRECTRD,
-				CQSPI_REG_INDIRECTRD_DONE, 0, 10, 0);
-	if (ret) {
-		printf("Indirect read clear completion error (%i)\n", ret);
-		goto failrd;
-	}
+	/* Wait til QSPI is idle */
+	if (!cadence_qspi_wait_idle(plat->regbase))
+		return -EIO;
 
 	return 0;
 
@@ -1031,6 +1027,11 @@ cadence_qspi_apb_indirect_write_execute(struct cadence_spi_plat *plat,
 
 	if (bounce_buf)
 		free(bounce_buf);
+
+	/* Wait til QSPI is idle */
+	if (!cadence_qspi_wait_idle(plat->regbase))
+		return -EIO;
+
 	return 0;
 
 failwr:
-- 
2.25.1

Re: [PATCH 016/347] FogBugz #516535: Fix QSPI write issues

[Jagan Teki]

On Tue, Aug 30, 2022 at 11:50 AM Jit Loon Lim <jit.loon.lim@intel.com> wrote:
>
> From: Chee Hong Ang <chee.hong.ang@intel.com>
>
> QSPI driver perform chip select on every flash read/write
> access. The driver need to disable/enable the QSPI controller
> while performing chip select. This may cause some data lost
> especially the QSPI controller is configured to run at slower
> speed as it may take longer time to access the flash device.
> This patch prevent the driver from disable/enable the QSPI
> controller too soon and inadvertently halting any ongoing flash
> read/write access by ensuring the QSPI controller is always in
> idle mode after each read/write access.
>
> Signed-off-by: Chee Hong Ang <chee.hong.ang@intel.com>
> --

"FogBugz #516535:" This commit head is invalid for the upstream patch.

Re: [PATCH 016/347] FogBugz #516535: Fix QSPI write issues

[Vignesh Raghavendra]

Hi,

On 30/08/22 11:50 am, Jit Loon Lim wrote:
> From: Chee Hong Ang <chee.hong.ang@intel.com>
> 
> QSPI driver perform chip select on every flash read/write
> access. The driver need to disable/enable the QSPI controller
> while performing chip select. This may cause some data lost
> especially the QSPI controller is configured to run at slower
> speed as it may take longer time to access the flash device.
> This patch prevent the driver from disable/enable the QSPI
> controller too soon and inadvertently halting any ongoing flash
> read/write access by ensuring the QSPI controller is always in
> idle mode after each read/write access.
> 
> Signed-off-by: Chee Hong Ang <chee.hong.ang@intel.com>
> ---
>  drivers/spi/cadence_qspi_apb.c | 15 ++++++++-------
>  1 file changed, 8 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/spi/cadence_qspi_apb.c b/drivers/spi/cadence_qspi_apb.c
> index 2cdf4c9c9f..5e03495f45 100644
> --- a/drivers/spi/cadence_qspi_apb.c
> +++ b/drivers/spi/cadence_qspi_apb.c
> @@ -858,13 +858,9 @@ cadence_qspi_apb_indirect_read_execute(struct cadence_spi_plat *plat,
>  	writel(CQSPI_REG_INDIRECTRD_DONE,
>  	       plat->regbase + CQSPI_REG_INDIRECTRD);
>  
> -	/* Check indirect done status */
> -	ret = wait_for_bit_le32(plat->regbase + CQSPI_REG_INDIRECTRD,
> -				CQSPI_REG_INDIRECTRD_DONE, 0, 10, 0);
> -	if (ret) {
> -		printf("Indirect read clear completion error (%i)\n", ret);
> -		goto failrd;
> -	}

Why would you drop looking at CQSPI_REG_INDIRECTRD_DONE status bit? This
gives out a much granular error wrt what operation actually failed.


> +	/* Wait til QSPI is idle */
> +	if (!cadence_qspi_wait_idle(plat->regbase))
> +		return -EIO;
>  
>  	return 0;
>  
> @@ -1031,6 +1027,11 @@ cadence_qspi_apb_indirect_write_execute(struct cadence_spi_plat *plat,
>  
>  	if (bounce_buf)
>  		free(bounce_buf);
> +
> +	/* Wait til QSPI is idle */
> +	if (!cadence_qspi_wait_idle(plat->regbase))
> +		return -EIO;
> +
>  	return 0;
>  
>  failwr:

RE: [PATCH 016/347] FogBugz #516535: Fix QSPI write issues

[Chee, Tien Fong]

Hi Jit Loon,

> -----Original Message-----
> From: Lim, Jit Loon <jit.loon.lim@intel.com>
> Sent: Tuesday, 2 August, 2022 9:55 PM
> To: u-boot@lists.denx.de
> Cc: Jagan Teki <jagan@amarulasolutions.com>; Vignesh R <vigneshr@ti.com>;
> Vasut, Marek <marex@denx.de>; Simon <simon.k.r.goldschmidt@gmail.com>;
> Chaplin, Kris <kris.chaplin@intel.com>; Chee, Tien Fong
> <tien.fong.chee@intel.com>; Hea, Kok Kiang <kok.kiang.hea@intel.com>; Lim,
> Elly Siew Chin <elly.siew.chin.lim@intel.com>; Kho, Sin Hui
> <sin.hui.kho@intel.com>; Lokanathan, Raaj <raaj.lokanathan@intel.com>;
> Maniyam, Dinesh <dinesh.maniyam@intel.com>; Ng, Boon Khai
> <boon.khai.ng@intel.com>; Yuslaimi, Alif Zakuan
> <alif.zakuan.yuslaimi@intel.com>; Chong, Teik Heng
> <teik.heng.chong@intel.com>; Zamri, Muhammad Hazim Izzat
> <muhammad.hazim.izzat.zamri@intel.com>; Lim, Jit Loon
> <jit.loon.lim@intel.com>; Chee Hong Ang <chee.hong.ang@intel.com>
> Subject: [PATCH 016/347] FogBugz #516535: Fix QSPI write issues

Please drop the "[PATCH 016/347]" and replace "FogBugz #516535" with appropriate tag.

> 
> From: Chee Hong Ang <chee.hong.ang@intel.com>
> 
> QSPI driver perform chip select on every flash read/write access. The driver need
> to disable/enable the QSPI controller while performing chip select. This may
> cause some data lost especially the QSPI controller is configured to run at
> slower speed as it may take longer time to access the flash device.
> This patch prevent the driver from disable/enable the QSPI controller too soon
> and inadvertently halting any ongoing flash read/write access by ensuring the
> QSPI controller is always in idle mode after each read/write access.
> 
> Signed-off-by: Chee Hong Ang <chee.hong.ang@intel.com>
> ---
>  drivers/spi/cadence_qspi_apb.c | 15 ++++++++-------
>  1 file changed, 8 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/spi/cadence_qspi_apb.c b/drivers/spi/cadence_qspi_apb.c
> index 2cdf4c9c9f..5e03495f45 100644
> --- a/drivers/spi/cadence_qspi_apb.c
> +++ b/drivers/spi/cadence_qspi_apb.c
> @@ -858,13 +858,9 @@ cadence_qspi_apb_indirect_read_execute(struct
> cadence_spi_plat *plat,
>  	writel(CQSPI_REG_INDIRECTRD_DONE,
>  	       plat->regbase + CQSPI_REG_INDIRECTRD);
> 
> -	/* Check indirect done status */
> -	ret = wait_for_bit_le32(plat->regbase + CQSPI_REG_INDIRECTRD,
> -				CQSPI_REG_INDIRECTRD_DONE, 0, 10, 0);
> -	if (ret) {
> -		printf("Indirect read clear completion error (%i)\n", ret);
> -		goto failrd;
> -	}
> +	/* Wait til QSPI is idle */
> +	if (!cadence_qspi_wait_idle(plat->regbase))
> +		return -EIO;
> 
>  	return 0;
> 
> @@ -1031,6 +1027,11 @@ cadence_qspi_apb_indirect_write_execute(struct
> cadence_spi_plat *plat,
> 
>  	if (bounce_buf)
>  		free(bounce_buf);
> +
> +	/* Wait til QSPI is idle */
> +	if (!cadence_qspi_wait_idle(plat->regbase))
> +		return -EIO;
> +
>  	return 0;
> 
>  failwr:
> --
> 2.25.1

Regards,
Tien Fong

[PATCH 016/347] FogBugz #516535: Fix QSPI write issues

[Jit Loon Lim]

From: Chee Hong Ang <chee.hong.ang@intel.com>

QSPI driver perform chip select on every flash read/write
access. The driver need to disable/enable the QSPI controller
while performing chip select. This may cause some data lost
especially the QSPI controller is configured to run at slower
speed as it may take longer time to access the flash device.
This patch prevent the driver from disable/enable the QSPI
controller too soon and inadvertently halting any ongoing flash
read/write access by ensuring the QSPI controller is always in
idle mode after each read/write access.

Signed-off-by: Chee Hong Ang <chee.hong.ang@intel.com>
---
 drivers/spi/cadence_qspi_apb.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/drivers/spi/cadence_qspi_apb.c b/drivers/spi/cadence_qspi_apb.c
index 2cdf4c9c9f..5e03495f45 100644
--- a/drivers/spi/cadence_qspi_apb.c
+++ b/drivers/spi/cadence_qspi_apb.c
@@ -858,13 +858,9 @@ cadence_qspi_apb_indirect_read_execute(struct cadence_spi_plat *plat,
 	writel(CQSPI_REG_INDIRECTRD_DONE,
 	       plat->regbase + CQSPI_REG_INDIRECTRD);
 
-	/* Check indirect done status */
-	ret = wait_for_bit_le32(plat->regbase + CQSPI_REG_INDIRECTRD,
-				CQSPI_REG_INDIRECTRD_DONE, 0, 10, 0);
-	if (ret) {
-		printf("Indirect read clear completion error (%i)\n", ret);
-		goto failrd;
-	}
+	/* Wait til QSPI is idle */
+	if (!cadence_qspi_wait_idle(plat->regbase))
+		return -EIO;
 
 	return 0;
 
@@ -1031,6 +1027,11 @@ cadence_qspi_apb_indirect_write_execute(struct cadence_spi_plat *plat,
 
 	if (bounce_buf)
 		free(bounce_buf);
+
+	/* Wait til QSPI is idle */
+	if (!cadence_qspi_wait_idle(plat->regbase))
+		return -EIO;
+
 	return 0;
 
 failwr:
-- 
2.25.1