Skip to contentSkip to conversation

crypto: caam: Change the i.MX8MQ check support all i.MX8M variants

PATCH In review Adam Ford 3 participants 10 messages Nov 30, 2019 – Dec 11, 2019

Subsystems (4)
lore
Share & tools
lore

Adam Ford

author
#0 lore raw

PATCH 1/2 crypto: caam: Change the i.MX8MQ check support all i.MX8M variants

The i.MX8M Mini uses the same crypto engine as the i.MX8MQ, but the driver is restricting the check to just the i.MX8MQ.

This patch lets the driver support all i.MX8M Variants if enabled.

drivers/crypto/caam/ctrl.c +1 -1
index db22777d59b4..1ce03f8961b6 100644
@@ -527,7 +527,7 @@ static const struct soc_device_attribute caam_imx_soc_table[] = {
 	{ .soc_id = "i.MX6UL", .data = &caam_imx6ul_data },
 	{ .soc_id = "i.MX6*",  .data = &caam_imx6_data },
 	{ .soc_id = "i.MX7*",  .data = &caam_imx7_data },
-	{ .soc_id = "i.MX8MQ", .data = &caam_imx7_data },
+	{ .soc_id = "i.MX8M*", .data = &caam_imx7_data },
 	{ .family = "Freescale i.MX" },
 	{ /* sentinel */ }
 };
Signature · 1 line
2.20.1
author
#1 lore raw

PATCH 2/2 arm64: dts: imx8mm: Add Crypto CAAM support

The i.MX8M Mini supports the same crypto engine as what is in the i.MX8MQ, but it is not currently present in the device tree, because it may be resricted by security features.

This patch places in into the device tree and marks it as disabled, but anyone not restricting the CAAM with secure mode functions can mark it as enabled.

arch/arm64/boot/dts/freescale/imx8mm.dtsi +31 -0
index 2ed1a3537f05..68c7c1adeb60 100644
@@ -723,6 +723,37 @@
 				status = "disabled";
 			};
 
+			crypto: crypto@30900000 {
+				compatible = "fsl,sec-v4.0";
+				#address-cells = <1>;
+				#size-cells = <1>;
+				reg = <0x30900000 0x40000>;
+				ranges = <0 0x30900000 0x40000>;
+				interrupts = <GIC_SPI 91 IRQ_TYPE_LEVEL_HIGH>;
+				clocks = <&clk IMX8MM_CLK_AHB>,
+					 <&clk IMX8MM_CLK_IPG_ROOT>;
+				clock-names = "aclk", "ipg";
+				status = "disabled";
+
+				sec_jr0: jr@1000 {
+					compatible = "fsl,sec-v4.0-job-ring";
+					reg = <0x1000 0x1000>;
+					interrupts = <GIC_SPI 105 IRQ_TYPE_LEVEL_HIGH>;
+				};
+
+				sec_jr1: jr@2000 {
+					compatible = "fsl,sec-v4.0-job-ring";
+					reg = <0x2000 0x1000>;
+					interrupts = <GIC_SPI 106 IRQ_TYPE_LEVEL_HIGH>;
+				};
+
+				sec_jr2: jr@3000 {
+					compatible = "fsl,sec-v4.0-job-ring";
+					reg = <0x3000 0x1000>;
+					interrupts = <GIC_SPI 114 IRQ_TYPE_LEVEL_HIGH>;
+				};
+			};
+
 			i2c1: i2c@30a20000 {
 				compatible = "fsl,imx8mm-i2c", "fsl,imx21-i2c";
 				#address-cells = <1>;
Signature · 1 line
2.20.1
maintainer
#2 lore raw

Re: [PATCH 2/2] arm64: dts: imx8mm: Add Crypto CAAM support

Quoted 4 lines · Adam Ford

The i.MX8M Mini supports the same crypto engine as what is in the i.MX8MQ, but it is not currently present in the device tree, because it may be resricted by security features.

What exactly are you referring to?

Quoted 4 lines

This patch places in into the device tree and marks it as disabled, but anyone not restricting the CAAM with secure mode functions can mark it as enabled.

Even if - due to export control regulations - CAAM is "trimmed down", it loses only the encryption capabilities (hashing etc. still working).

Again, please clarify what you mean by "secure mode functions", "security features" etc.

Horia

author
#3 lore raw

Re: [PATCH 2/2] arm64: dts: imx8mm: Add Crypto CAAM support

Quoted 7 lines · Horia Geanta
Quoted 4 lines · Adam Ford

The i.MX8M Mini supports the same crypto engine as what is in the i.MX8MQ, but it is not currently present in the device tree, because it may be resricted by security features.

What exactly are you referring to?

I don't know this hardware very well, but on a different platform, we needed to make the crypto engines as disabled if they were being accessed through secure operations which made it unavailable to Linux without using some special barriers. I didn't have the special hardware on the other platform that required it that way, so I can't really explain it well. I know on those special cases, because some people were accessing these registers through other means, the devices had to be marked as 'disabled' so to avoid breaking something. Since I wasn't sure if this was left out of the i.MX8M Mini on purpose, I let this disabled just in case this hardware platform was also affected in a similar and people wanting to use it could mark it as 'okay'

adam

Quoted 12 lines
Quoted 5 lines

This patch places in into the device tree and marks it as disabled, but anyone not restricting the CAAM with secure mode functions can mark it as enabled.

Even if - due to export control regulations - CAAM is "trimmed down", it loses only the encryption capabilities (hashing etc. still working).

Again, please clarify what you mean by "secure mode functions", "security features" etc.

Horia

#4 lore raw

Re: [PATCH 2/2] arm64: dts: imx8mm: Add Crypto CAAM support

Hi Adam,

Quoted 9 lines · Adam Ford
Quoted 7 lines · Horia Geanta
Quoted 4 lines · Adam Ford

The i.MX8M Mini supports the same crypto engine as what is in the i.MX8MQ, but it is not currently present in the device tree, because it may be resricted by security features.

What exactly are you referring to?

Quoted 12 lines

I don't know this hardware very well, but on a different platform, we needed to make the crypto engines as disabled if they were being accessed through secure operations which made it unavailable to Linux without using some special barriers. I didn't have the special hardware on the other platform that required it that way, so I can't really explain it well. I know on those special cases, because some people were accessing these registers through other means, the devices had to be marked as 'disabled' so to avoid breaking something. Since I wasn't sure if this was left out of the i.MX8M Mini on purpose, I let this disabled just in case this hardware platform was also affected in a similar and people wanting to use it could mark it as 'okay'

I don't know enough about this to understand the problem you're describing. It seems like most SoCs have the CAAM enabled by default in the devicetree. On first glance I could only find fsl-lx2160a.dtsi that has it disabled.

Quoted 10 lines

adam

Quoted 7 lines
Quoted 5 lines

This patch places in into the device tree and marks it as disabled, but anyone not restricting the CAAM with secure mode functions can mark it as enabled.

Even if - due to export control regulations - CAAM is "trimmed down", it loses only the encryption capabilities (hashing etc. still working).

I don't know much about this, but as Horia said the CAAM might have limited capabilities in some cases but would still work.

Therefore I think the CAAM should be enabled by default as it already is done for most other SoCs.

Regards,
Frieder

Quoted 11 lines
Quoted 5 lines

Again, please clarify what you mean by "secure mode functions", "security features" etc.

Horia

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#5 lore raw
In reply to Adam Ford

Hi Adam,

Quoted 6 lines · Adam Ford

The i.MX8M Mini uses the same crypto engine as the i.MX8MQ, but the driver is restricting the check to just the i.MX8MQ.

This patch lets the driver support all i.MX8M Variants if enabled.

What about the following lines in run_descriptor_deco0()? Does this condition also apply to i.MX8MM?

drivers/crypto/caam/ctrl.c:

	if (ctrlpriv->virt_en == 1 ||
	    /*
	     * Apparently on i.MX8MQ it doesn't matter if virt_en == 1
	     * and the following steps should be performed regardless
	     */
	    of_machine_is_compatible("fsl,imx8mq")) {
		clrsetbits_32(&ctrl->deco_rsr, 0, DECORSR_JR0);

		while (!(rd_reg32(&ctrl->deco_rsr) & DECORSR_VALID) &&
		       --timeout)
			cpu_relax();

		timeout = 100000;
	}

Regards,
Frieder

Quoted 15 lines
drivers/crypto/caam/ctrl.c +1 -1
index db22777d59b4..1ce03f8961b6 100644
@@ -527,7 +527,7 @@ static const struct soc_device_attribute caam_imx_soc_table[] = {
  	{ .soc_id = "i.MX6UL", .data = &caam_imx6ul_data },
  	{ .soc_id = "i.MX6*",  .data = &caam_imx6_data },
  	{ .soc_id = "i.MX7*",  .data = &caam_imx7_data },
-	{ .soc_id = "i.MX8MQ", .data = &caam_imx7_data },
+	{ .soc_id = "i.MX8M*", .data = &caam_imx7_data },
  	{ .family = "Freescale i.MX" },
  	{ /* sentinel */ }
  };
author
#6 lore raw
Quoted 7 lines · Schrempf Frieder

Hi Adam,

Quoted 3 lines · Adam Ford

The i.MX8M Mini uses the same crypto engine as the i.MX8MQ, but the driver is restricting the check to just the i.MX8MQ.

Quoted 6 lines
Quoted 3 lines

This patch lets the driver support all i.MX8M Variants if enabled.

What about the following lines in run_descriptor_deco0()? Does this condition also apply to i.MX8MM?

I think that's a question for NXP. I am not seeing that in the NXP Linux Release, and I don't have an 8MQ to compare.

I was able to get the driver working on the i.MXMM with the patch.

NXP Team,

Do you have any opinions on this?

adam

Quoted 36 lines

drivers/crypto/caam/ctrl.c:

        if (ctrlpriv->virt_en == 1 ||
            /*
             * Apparently on i.MX8MQ it doesn't matter if virt_en == 1
             * and the following steps should be performed regardless
             */
            of_machine_is_compatible("fsl,imx8mq")) {
                clrsetbits_32(&ctrl->deco_rsr, 0, DECORSR_JR0);

                while (!(rd_reg32(&ctrl->deco_rsr) & DECORSR_VALID) &&
                       --timeout)
                        cpu_relax();

                timeout = 100000;
        }

Regards,
Frieder

Quoted 15 lines
drivers/crypto/caam/ctrl.c +1 -1
index db22777d59b4..1ce03f8961b6 100644
@@ -527,7 +527,7 @@ static const struct soc_device_attribute caam_imx_soc_table[] = {
      { .soc_id = "i.MX6UL", .data = &caam_imx6ul_data },
      { .soc_id = "i.MX6*",  .data = &caam_imx6_data },
      { .soc_id = "i.MX7*",  .data = &caam_imx7_data },
-     { .soc_id = "i.MX8MQ", .data = &caam_imx7_data },
+     { .soc_id = "i.MX8M*", .data = &caam_imx7_data },
      { .family = "Freescale i.MX" },
      { /* sentinel */ }
  };
maintainer
#7 lore raw
Quoted 11 lines · Adam Ford
Quoted 9 lines · Schrempf Frieder

Hi Adam,

Quoted 5 lines · Adam Ford

The i.MX8M Mini uses the same crypto engine as the i.MX8MQ, but the driver is restricting the check to just the i.MX8MQ.

This patch lets the driver support all i.MX8M Variants if enabled.

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

What about the following lines in run_descriptor_deco0()? Does this condition also apply to i.MX8MM?

I think that's a question for NXP. I am not seeing that in the NXP Linux Release, and I don't have an 8MQ to compare.

IIRC the i.MX BSP releases use the JRI for initializing the RNG, and not the DECO register interface.

Quoted 2 lines

I was able to get the driver working on the i.MXMM with the patch.

You are probably using a newer U-boot, which includes commit dfaec76029f2 ("crypto/fsl: instantiate all rng state handles")

Quoted 4 lines

NXP Team,

Do you have any opinions on this?

Since current U-boot initializes both RNG state handles, practically instantiate_rng() is a no-op.

A simple experiment is to "lie" about the state_handle_mask, to exercise the DECO acquire code (or, as mentioned above, to run with an older U-boot):

@@ -268,12 +272,19 @@ static int instantiate_rng(struct device *ctrldev, int state_handle_mask,

        struct caam_ctrl __iomem *ctrl;
        u32 *desc, status = 0, rdsta_val;
        int ret = 0, sh_idx;

+ static int force_init = 1;

        ctrl = (struct caam_ctrl __iomem *)ctrlpriv->ctrl;
        desc = kmalloc(CAAM_CMD_SZ * 7, GFP_KERNEL);
        if (!desc)
                return -ENOMEM;

+ if (force_init && (state_handle_mask == 0x3)) {

+               dev_err(ctrldev, "Forcing reinit of RNG state handle 0!\n");
+               force_init = 0;
+               state_handle_mask = 0x2;
+       }
+
        for (sh_idx = 0; sh_idx < RNG4_MAX_HANDLES; sh_idx++) {
                /*
                 * If the corresponding bit is set, this state handle

In this case boot log confirms the DECO cannot be acquired:

[    2.137101] caam 30900000.crypto: Forcing reinit of RNG state handle 0!
[    2.172293] caam 30900000.crypto: failed to acquire DECO 0
[    2.177786] caam 30900000.crypto: failed to instantiate RNG

To sum up, writing to DECORSR is mandatory.

Horia

#8 lore raw
Quoted 59 lines · Horia Geanta
Quoted 11 lines · Adam Ford
Quoted 9 lines · Schrempf Frieder

Hi Adam,

Quoted 5 lines · Adam Ford

The i.MX8M Mini uses the same crypto engine as the i.MX8MQ, but the driver is restricting the check to just the i.MX8MQ.

This patch lets the driver support all i.MX8M Variants if enabled.

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

What about the following lines in run_descriptor_deco0()? Does this condition also apply to i.MX8MM?

I think that's a question for NXP. I am not seeing that in the NXP Linux Release, and I don't have an 8MQ to compare.

IIRC the i.MX BSP releases use the JRI for initializing the RNG, and not the DECO register interface.

Quoted 2 lines

I was able to get the driver working on the i.MXMM with the patch.

You are probably using a newer U-boot, which includes commit dfaec76029f2 ("crypto/fsl: instantiate all rng state handles")

Quoted 4 lines

NXP Team,

Do you have any opinions on this?

Since current U-boot initializes both RNG state handles, practically instantiate_rng() is a no-op.

A simple experiment is to "lie" about the state_handle_mask, to exercise the DECO acquire code (or, as mentioned above, to run with an older U-boot):

(quoted diff) +7 -0
@@ -268,12 +272,19 @@ static int instantiate_rng(struct device *ctrldev, int state_handle_mask,
         struct caam_ctrl __iomem *ctrl;
         u32 *desc, status = 0, rdsta_val;
         int ret = 0, sh_idx;
+       static int force_init = 1;
 
         ctrl = (struct caam_ctrl __iomem *)ctrlpriv->ctrl;
         desc = kmalloc(CAAM_CMD_SZ * 7, GFP_KERNEL);
         if (!desc)
                 return -ENOMEM;
 
+       if (force_init && (state_handle_mask == 0x3)) {
+               dev_err(ctrldev, "Forcing reinit of RNG state handle 0!\n");
+               force_init = 0;
+               state_handle_mask = 0x2;
+       }
+
         for (sh_idx = 0; sh_idx < RNG4_MAX_HANDLES; sh_idx++) {
                 /*
                  * If the corresponding bit is set, this state handle
Quoted 6 lines

In this case boot log confirms the DECO cannot be acquired:

[    2.137101] caam 30900000.crypto: Forcing reinit of RNG state handle 0!
[    2.172293] caam 30900000.crypto: failed to acquire DECO 0
[    2.177786] caam 30900000.crypto: failed to instantiate RNG

To sum up, writing to DECORSR is mandatory.

Thanks Horia for providing the details.

Adam, can you update your patch to enable the code in run_descriptor_deco0() for i.MX8MM?

If I understand this correctly, this is necessary to have the RNG initialize correctly no matter what version of U-Boot is used.

Thanks,
Frieder

author
#9 lore raw
Quoted 61 lines · Schrempf Frieder
Quoted 59 lines · Horia Geanta
Quoted 11 lines · Adam Ford
Quoted 9 lines · Schrempf Frieder

Hi Adam,

Quoted 5 lines · Adam Ford

The i.MX8M Mini uses the same crypto engine as the i.MX8MQ, but the driver is restricting the check to just the i.MX8MQ.

This patch lets the driver support all i.MX8M Variants if enabled.

Quoted 8 lines
Quoted 4 lines
Quoted 1 line

What about the following lines in run_descriptor_deco0()? Does this condition also apply to i.MX8MM?

I think that's a question for NXP. I am not seeing that in the NXP Linux Release, and I don't have an 8MQ to compare.

IIRC the i.MX BSP releases use the JRI for initializing the RNG, and not the DECO register interface.

Quoted 2 lines

I was able to get the driver working on the i.MXMM with the patch.

You are probably using a newer U-boot, which includes commit dfaec76029f2 ("crypto/fsl: instantiate all rng state handles")

Quoted 4 lines

NXP Team,

Do you have any opinions on this?

Since current U-boot initializes both RNG state handles, practically instantiate_rng() is a no-op.

A simple experiment is to "lie" about the state_handle_mask, to exercise the DECO acquire code (or, as mentioned above, to run with an older U-boot):

(quoted diff) +7 -0
@@ -268,12 +272,19 @@ static int instantiate_rng(struct device *ctrldev, int state_handle_mask,
         struct caam_ctrl __iomem *ctrl;
         u32 *desc, status = 0, rdsta_val;
         int ret = 0, sh_idx;
+       static int force_init = 1;
 
         ctrl = (struct caam_ctrl __iomem *)ctrlpriv->ctrl;
         desc = kmalloc(CAAM_CMD_SZ * 7, GFP_KERNEL);
         if (!desc)
                 return -ENOMEM;
 
+       if (force_init && (state_handle_mask == 0x3)) {
+               dev_err(ctrldev, "Forcing reinit of RNG state handle 0!\n");
+               force_init = 0;
+               state_handle_mask = 0x2;
+       }
+
         for (sh_idx = 0; sh_idx < RNG4_MAX_HANDLES; sh_idx++) {
                 /*
                  * If the corresponding bit is set, this state handle
Quoted 8 lines
Quoted 6 lines

In this case boot log confirms the DECO cannot be acquired:

[    2.137101] caam 30900000.crypto: Forcing reinit of RNG state handle 0!
[    2.172293] caam 30900000.crypto: failed to acquire DECO 0
[    2.177786] caam 30900000.crypto: failed to instantiate RNG

To sum up, writing to DECORSR is mandatory.

Thanks Horia for providing the details.

I appreciate it too.

Quoted 3 lines

Adam, can you update your patch to enable the code in run_descriptor_deco0() for i.MX8MM?

I will work on that. I have been trying to get the mainline U-Boot to start correctly, because I wanted to see if/how this interacted. I'll try to get a V2 pushed today.

Quoted 3 lines

If I understand this correctly, this is necessary to have the RNG initialize correctly no matter what version of U-Boot is used.

That makes sense based on Horia's feedback. With the holidays this month, my spare time and weekends have been full.

adam

Quoted 3 lines

Thanks,
Frieder