[PATCH v4 0/4] Port mxs-dcp to imx6ull and imx6sll

[PATCH v4 0/4] Port mxs-dcp to imx6ull and imx6sll

[Leonard Crestez]

The DCP block is present on 6sll and 6ull but not enabled. The hardware is
mostly compatible with 6sl, the only important difference is that explicit
clock enabling is required.

There were several issues with the functionality of this driver (it didn't
even probe properly) but they are fixed in cryptodev/master by this series:
    https://lore.kernel.org/patchwork/cover/994874/

Leonard Crestez (4):
  dt-bindings: crypto: Mention clocks for mxs-dcp
  crypto: mxs-dcp - Add support for dcp clk
  ARM: dts: imx6ull: Add dcp node
  ARM: imx_v6_v7_defconfig: Enable CRYPTO_DEV_MXS_DCP

 .../devicetree/bindings/crypto/fsl-dcp.txt    |  2 ++
 arch/arm/boot/dts/imx6ull.dtsi                | 10 +++++++
 arch/arm/configs/imx_v6_v7_defconfig          |  1 +
 drivers/crypto/mxs-dcp.c                      | 28 +++++++++++++++++--
 4 files changed, 38 insertions(+), 3 deletions(-)

-- 
2.17.1

[PATCH v4 1/4] dt-bindings: crypto: Mention clocks for mxs-dcp

[Leonard Crestez]

Explicit clock enabling is required on 6sll and 6ull so mention that
standard clock bindings are used.

Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
---
 Documentation/devicetree/bindings/crypto/fsl-dcp.txt | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/Documentation/devicetree/bindings/crypto/fsl-dcp.txt b/Documentation/devicetree/bindings/crypto/fsl-dcp.txt
index 76a0b4e80e83..4e4d387e38a5 100644
--- a/Documentation/devicetree/bindings/crypto/fsl-dcp.txt
+++ b/Documentation/devicetree/bindings/crypto/fsl-dcp.txt
@@ -4,10 +4,12 @@ Required properties:
 - compatible : Should be "fsl,<soc>-dcp"
 - reg : Should contain MXS DCP registers location and length
 - interrupts : Should contain MXS DCP interrupt numbers, VMI IRQ and DCP IRQ
                must be supplied, optionally Secure IRQ can be present, but
 	       is currently not implemented and not used.
+- clocks : Clock reference (only required on some SOCs: 6ull and 6sll).
+- clock-names : Must be "dcp".
 
 Example:
 
 dcp@80028000 {
 	compatible = "fsl,imx28-dcp", "fsl,imx23-dcp";
-- 
2.17.1

[PATCH v4 2/4] crypto: mxs-dcp - Add support for dcp clk

[Leonard Crestez]

On 6ull and 6sll the DCP block has a clock which needs to be explicitly
enabled.

Add minimal handling for this at probe/remove time.

Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
---
 drivers/crypto/mxs-dcp.c | 28 +++++++++++++++++++++++++---
 1 file changed, 25 insertions(+), 3 deletions(-)

diff --git a/drivers/crypto/mxs-dcp.c b/drivers/crypto/mxs-dcp.c
index 4e6ff32f8a7e..a2105cf33abb 100644
--- a/drivers/crypto/mxs-dcp.c
+++ b/drivers/crypto/mxs-dcp.c
@@ -18,10 +18,11 @@
 #include <linux/kthread.h>
 #include <linux/module.h>
 #include <linux/of.h>
 #include <linux/platform_device.h>
 #include <linux/stmp_device.h>
+#include <linux/clk.h>
 
 #include <crypto/aes.h>
 #include <crypto/sha.h>
 #include <crypto/internal/hash.h>
 #include <crypto/internal/skcipher.h>
@@ -80,10 +81,11 @@ struct dcp {
 
 	struct completion		completion[DCP_MAX_CHANS];
 	spinlock_t			lock[DCP_MAX_CHANS];
 	struct task_struct		*thread[DCP_MAX_CHANS];
 	struct crypto_queue		queue[DCP_MAX_CHANS];
+	struct clk			*dcp_clk;
 };
 
 enum dcp_chan {
 	DCP_CHAN_HASH_SHA	= 0,
 	DCP_CHAN_CRYPTO		= 2,
@@ -1051,15 +1053,28 @@ static int mxs_dcp_probe(struct platform_device *pdev)
 		return -ENOMEM;
 
 	/* Re-align the structure so it fits the DCP constraints. */
 	sdcp->coh = PTR_ALIGN(sdcp->coh, DCP_ALIGNMENT);
 
-	/* Restart the DCP block. */
-	ret = stmp_reset_block(sdcp->base);
+	/* DCP clock is optional, only used on some SOCs */
+	sdcp->dcp_clk = devm_clk_get(dev, "dcp");
+	if (IS_ERR(sdcp->dcp_clk)) {
+		if (sdcp->dcp_clk != ERR_PTR(-ENOENT))
+			return PTR_ERR(sdcp->dcp_clk);
+		sdcp->dcp_clk = NULL;
+	}
+	ret = clk_prepare_enable(sdcp->dcp_clk);
 	if (ret)
 		return ret;
 
+	/* Restart the DCP block. */
+	ret = stmp_reset_block(sdcp->base);
+	if (ret) {
+		dev_err(dev, "Failed reset\n");
+		goto err_disable_unprepare_clk;
+	}
+
 	/* Initialize control register. */
 	writel(MXS_DCP_CTRL_GATHER_RESIDUAL_WRITES |
 	       MXS_DCP_CTRL_ENABLE_CONTEXT_CACHING | 0xf,
 	       sdcp->base + MXS_DCP_CTRL);
 
@@ -1092,11 +1107,12 @@ static int mxs_dcp_probe(struct platform_device *pdev)
 	/* Create the SHA and AES handler threads. */
 	sdcp->thread[DCP_CHAN_HASH_SHA] = kthread_run(dcp_chan_thread_sha,
 						      NULL, "mxs_dcp_chan/sha");
 	if (IS_ERR(sdcp->thread[DCP_CHAN_HASH_SHA])) {
 		dev_err(dev, "Error starting SHA thread!\n");
-		return PTR_ERR(sdcp->thread[DCP_CHAN_HASH_SHA]);
+		ret = PTR_ERR(sdcp->thread[DCP_CHAN_HASH_SHA]);
+		goto err_disable_unprepare_clk;
 	}
 
 	sdcp->thread[DCP_CHAN_CRYPTO] = kthread_run(dcp_chan_thread_aes,
 						    NULL, "mxs_dcp_chan/aes");
 	if (IS_ERR(sdcp->thread[DCP_CHAN_CRYPTO])) {
@@ -1149,10 +1165,14 @@ static int mxs_dcp_probe(struct platform_device *pdev)
 err_destroy_aes_thread:
 	kthread_stop(sdcp->thread[DCP_CHAN_CRYPTO]);
 
 err_destroy_sha_thread:
 	kthread_stop(sdcp->thread[DCP_CHAN_HASH_SHA]);
+
+err_disable_unprepare_clk:
+	clk_disable_unprepare(sdcp->dcp_clk);
+
 	return ret;
 }
 
 static int mxs_dcp_remove(struct platform_device *pdev)
 {
@@ -1168,10 +1188,12 @@ static int mxs_dcp_remove(struct platform_device *pdev)
 		crypto_unregister_algs(dcp_aes_algs, ARRAY_SIZE(dcp_aes_algs));
 
 	kthread_stop(sdcp->thread[DCP_CHAN_HASH_SHA]);
 	kthread_stop(sdcp->thread[DCP_CHAN_CRYPTO]);
 
+	clk_disable_unprepare(sdcp->dcp_clk);
+
 	platform_set_drvdata(pdev, NULL);
 
 	global_sdcp = NULL;
 
 	return 0;
-- 
2.17.1

[PATCH v4 3/4] ARM: dts: imx6ull: Add dcp node

[Leonard Crestez]

The DCP block on 6ull has no major differences other than requiring
explicit clock enabling.

Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
---
 arch/arm/boot/dts/imx6ull.dtsi | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/arch/arm/boot/dts/imx6ull.dtsi b/arch/arm/boot/dts/imx6ull.dtsi
index 796ed35d4ac9..f3668fe69eac 100644
--- a/arch/arm/boot/dts/imx6ull.dtsi
+++ b/arch/arm/boot/dts/imx6ull.dtsi
@@ -37,10 +37,20 @@
 			#address-cells = <1>;
 			#size-cells = <1>;
 			reg = <0x02200000 0x100000>;
 			ranges;
 
+			dcp: crypto@2280000 {
+				compatible = "fsl,imx6ull-dcp", "fsl,imx28-dcp";
+				reg = <0x02280000 0x4000>;
+				interrupts = <GIC_SPI 46 IRQ_TYPE_LEVEL_HIGH>,
+					     <GIC_SPI 47 IRQ_TYPE_LEVEL_HIGH>,
+					     <GIC_SPI 48 IRQ_TYPE_LEVEL_HIGH>;
+				clocks = <&clks IMX6ULL_CLK_DCP_CLK>;
+				clock-names = "dcp";
+			};
+
 			iomuxc_snvs: iomuxc-snvs@2290000 {
 				compatible = "fsl,imx6ull-iomuxc-snvs";
 				reg = <0x02290000 0x4000>;
 			};
 
-- 
2.17.1

[PATCH v4 4/4] ARM: imx_v6_v7_defconfig: Enable CRYPTO_DEV_MXS_DCP

[Leonard Crestez]

This block is present in 6sl, 6sll and 6ull so it should be enabled in
the default imx kernel config.

Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
---
 arch/arm/configs/imx_v6_v7_defconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm/configs/imx_v6_v7_defconfig b/arch/arm/configs/imx_v6_v7_defconfig
index 1ad5736c8fa6..57928dff9bce 100644
--- a/arch/arm/configs/imx_v6_v7_defconfig
+++ b/arch/arm/configs/imx_v6_v7_defconfig
@@ -428,10 +428,11 @@ CONFIG_NLS_ISO8859_1=y
 CONFIG_NLS_ISO8859_15=m
 CONFIG_NLS_UTF8=y
 CONFIG_SECURITYFS=y
 CONFIG_CRYPTO_DEV_FSL_CAAM=y
 CONFIG_CRYPTO_DEV_SAHARA=y
+CONFIG_CRYPTO_DEV_MXS_DCP=y
 CONFIG_CRC_CCITT=m
 CONFIG_CRC_T10DIF=y
 CONFIG_CRC7=m
 CONFIG_LIBCRC32C=m
 CONFIG_FONTS=y
-- 
2.17.1

Re: [PATCH v4 2/4] crypto: mxs-dcp - Add support for dcp clk

[Fabio Estevam]

Hi Leonard,

On Wed, Oct 17, 2018 at 9:38 AM Leonard Crestez <leonard.crestez@nxp.com> wrote:
>
> On 6ull and 6sll the DCP block has a clock which needs to be explicitly
> enabled.
>
> Add minimal handling for this at probe/remove time.
>
> Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
> ---

Please always explain what changed from the previous patch version.

> +       /* Restart the DCP block. */
> +       ret = stmp_reset_block(sdcp->base);
> +       if (ret) {
> +               dev_err(dev, "Failed reset\n");
> +               goto err_disable_unprepare_clk;
> +       }

This seems like an unrelated change that should be in a separate patch.

Also, this was not present in v3, so it is better to explain that this
has been introduced in v4.

Re: [PATCH v4 2/4] crypto: mxs-dcp - Add support for dcp clk

[Leonard Crestez]

On Wed, 2018-10-17 at 09:48 -0300, Fabio Estevam wrote:
> On Wed, Oct 17, 2018 at 9:38 AM Leonard Crestez <leonard.crestez@nxp.com> wrote:
> > 
> > On 6ull and 6sll the DCP block has a clock which needs to be explicitly
> > enabled.
> > 
> > Add minimal handling for this at probe/remove time.
> > 
> > Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
> 
> Please always explain what changed from the previous patch version.

There is a changelog in the cover letter.

> > +       /* Restart the DCP block. */
> > +       ret = stmp_reset_block(sdcp->base);
> > +       if (ret) {
> > +               dev_err(dev, "Failed reset\n");
> > +               goto err_disable_unprepare_clk;
> > +       }
> 
> This seems like an unrelated change that should be in a separate patch.
> 
> Also, this was not present in v3, so it is better to explain that this
> has been introduced in v4.

This only looks slightly odd in git diff but it's not unrelated.

I placed clk get/prepare/enable just before stmp_reset_block and made
stmp_reset_block print a message and goto err_disable_unprepare_clk on
failure instead of just "if (ret) return ret;"

Re: [PATCH v4 2/4] crypto: mxs-dcp - Add support for dcp clk

[Fabio Estevam]

On Wed, Oct 17, 2018 at 9:59 AM Leonard Crestez <leonard.crestez@nxp.com> wrote:

> There is a changelog in the cover letter.

It did not show up:
https://lkml.org/lkml/2018/10/17/673

> This only looks slightly odd in git diff but it's not unrelated.
>
> I placed clk get/prepare/enable just before stmp_reset_block and made
> stmp_reset_block print a message and goto err_disable_unprepare_clk on
> failure instead of just "if (ret) return ret;"

Got it! Then it looks fine, thanks:

Reviewed-by: Fabio Estevam <festevam@gmail.com>

Re: [PATCH v4 1/4] dt-bindings: crypto: Mention clocks for mxs-dcp

[Rob Herring]

On Wed, 17 Oct 2018 12:37:51 +0000, Leonard Crestez wrote:
> Explicit clock enabling is required on 6sll and 6ull so mention that
> standard clock bindings are used.
> 
> Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
> Reviewed-by: Fabio Estevam <festevam@gmail.com>
> ---
>  Documentation/devicetree/bindings/crypto/fsl-dcp.txt | 2 ++
>  1 file changed, 2 insertions(+)
> 

Reviewed-by: Rob Herring <robh@kernel.org>

Re: [PATCH v4 3/4] ARM: dts: imx6ull: Add dcp node

[Shawn Guo]

On Wed, Oct 17, 2018 at 12:37:54PM +0000, Leonard Crestez wrote:
> The DCP block on 6ull has no major differences other than requiring
> explicit clock enabling.
> 
> Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
> Reviewed-by: Fabio Estevam <festevam@gmail.com>

Applied, thanks.

Re: [PATCH v4 4/4] ARM: imx_v6_v7_defconfig: Enable CRYPTO_DEV_MXS_DCP

[Shawn Guo]

On Wed, Oct 17, 2018 at 12:37:55PM +0000, Leonard Crestez wrote:
> This block is present in 6sl, 6sll and 6ull so it should be enabled in
> the default imx kernel config.
> 
> Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
> Reviewed-by: Fabio Estevam <festevam@gmail.com>

Applied, thanks.

Re: [PATCH v4 2/4] crypto: mxs-dcp - Add support for dcp clk

[Leonard Crestez]

On 10/17/2018 4:02 PM, Fabio Estevam wrote:
> On Wed, Oct 17, 2018 at 9:59 AM Leonard Crestez <leonard.crestez@nxp.com> wrote:
> 
>> There is a changelog in the cover letter.
> 
> It did not show up:
> https://lkml.org/lkml/2018/10/17/673

Sorry, I forgot to copy the changelog for v4. The only change is the one 
we discussed earlier in this thread.

>> This only looks slightly odd in git diff but it's not unrelated.
>>
>> I placed clk get/prepare/enable just before stmp_reset_block and made
>> stmp_reset_block print a message and goto err_disable_unprepare_clk on
>> failure instead of just "if (ret) return ret;"
> 
> Got it! Then it looks fine, thanks:
> 
> Reviewed-by: Fabio Estevam <festevam@gmail.com>

Thanks for looking at this series.

Shawn already applied patches 3/4, I guess parts 1/2 should go through 
the crypto tree? I don't know how crypto treats patches in the merge 
window, I can resend with all the ack/review tags after rc1 is out.