[PATCH 0/2] Qualcomm SMCCC Session ID Support

[PATCH 0/2] Qualcomm SMCCC Session ID Support

[Andy Gross]

This set of patches fixes a problem with the recent adoption of the ARM
SMCCC in the Qualcomm SCM firmware.  Qualcomm actually uses the optional
Trusted OS Session ID parameter.  When SCM calls are interrupted, the
session ID field is populated with a value that must be used when the
SCM call is resumed.

The first patch extends the arm_smccc_res structure to contain the
additional a6 result field and modifies the SMCCC ASM macro to store
register 6 in the additional field.  The second patch modifies the
Qualcomm SCM code to use the new result field.

Andy Gross (2):
  arm64: kernel: Add SMC Session ID to results
  firmware: qcom: scm: Fix interrupted SCM calls

 arch/arm64/kernel/asm-offsets.c | 1 +
 arch/arm64/kernel/smccc-call.S  | 1 +
 drivers/firmware/qcom_scm-64.c  | 6 ++++--
 include/linux/arm-smccc.h       | 4 +++-
 4 files changed, 9 insertions(+), 3 deletions(-)

-- 
1.9.1

[PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Andy Gross]

This patch adds the SMC Session ID to the results passed back from SMC
calls.  The Qualcomm SMC implementation provides for interrupted SMC
functions.  When this occurs, the SMC call will return a session ID that
is required to be used when resuming the interrupted SMC call.

Signed-off-by: Andy Gross <andy.gross@linaro.org>
---
 arch/arm64/kernel/asm-offsets.c | 1 +
 arch/arm64/kernel/smccc-call.S  | 1 +
 include/linux/arm-smccc.h       | 4 +++-
 3 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/asm-offsets.c b/arch/arm64/kernel/asm-offsets.c
index 05070b7..ff38c58 100644
--- a/arch/arm64/kernel/asm-offsets.c
+++ b/arch/arm64/kernel/asm-offsets.c
@@ -141,6 +141,7 @@ int main(void)
 #endif
   DEFINE(ARM_SMCCC_RES_X0_OFFS,	offsetof(struct arm_smccc_res, a0));
   DEFINE(ARM_SMCCC_RES_X2_OFFS,	offsetof(struct arm_smccc_res, a2));
+  DEFINE(ARM_SMCCC_RES_X6_OFFS,	offsetof(struct arm_smccc_res, a6));
   BLANK();
   DEFINE(HIBERN_PBE_ORIG,	offsetof(struct pbe, orig_address));
   DEFINE(HIBERN_PBE_ADDR,	offsetof(struct pbe, address));
diff --git a/arch/arm64/kernel/smccc-call.S b/arch/arm64/kernel/smccc-call.S
index ae0496f..278dc9a 100644
--- a/arch/arm64/kernel/smccc-call.S
+++ b/arch/arm64/kernel/smccc-call.S
@@ -20,6 +20,7 @@
 	ldr	x4, [sp]
 	stp	x0, x1, [x4, #ARM_SMCCC_RES_X0_OFFS]
 	stp	x2, x3, [x4, #ARM_SMCCC_RES_X2_OFFS]
+	str	x6, [x4, #ARM_SMCCC_RES_X6_OFFS]
 	ret
 	.cfi_endproc
 	.endm
diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
index b5abfda..82d919f 100644
--- a/include/linux/arm-smccc.h
+++ b/include/linux/arm-smccc.h
@@ -63,18 +63,20 @@
 /**
  * struct arm_smccc_res - Result from SMC/HVC call
  * @a0-a3 result values from registers 0 to 3
+ * @a6 Session ID register (optional)
  */
 struct arm_smccc_res {
 	unsigned long a0;
 	unsigned long a1;
 	unsigned long a2;
 	unsigned long a3;
+	unsigned long a6;
 };
 
 /**
  * arm_smccc_smc() - make SMC calls
  * @a0-a7: arguments passed in registers 0 to 7
- * @res: result values from registers 0 to 3
+ * @res: result values from registers 0 to 3 and optional register 6
  *
  * This function is used to make SMC calls following SMC Calling Convention.
  * The content of the supplied param are copied to registers 0 to 7 prior
-- 
1.9.1

[PATCH 2/2] firmware: qcom: scm: Fix interrupted SCM calls

[Andy Gross]

This patch fixes an issue with the SCM64 calls.  Sometimes SCM calls can
be interrupted and return early.  When this happens, the contents of
register 6 will contain a session ID that is required when resuming the
SCM call.

Signed-off-by: Andy Gross <andy.gross@linaro.org>
---
 drivers/firmware/qcom_scm-64.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/firmware/qcom_scm-64.c b/drivers/firmware/qcom_scm-64.c
index 4a0f5ea..1851294 100644
--- a/drivers/firmware/qcom_scm-64.c
+++ b/drivers/firmware/qcom_scm-64.c
@@ -131,10 +131,12 @@ static int qcom_scm_call(struct device *dev, u32 svc_id, u32 cmd_id,
 					 qcom_smccc_convention,
 					 ARM_SMCCC_OWNER_SIP, fn_id);
 
+		res->a6 = 0;
+
 		do {
 			arm_smccc_smc(cmd, desc->arginfo, desc->args[0],
-				      desc->args[1], desc->args[2], x5, 0, 0,
-				      res);
+				      desc->args[1], desc->args[2], x5, res->a6,
+				      0, res);
 		} while (res->a0 == QCOM_SCM_INTERRUPTED);
 
 		mutex_unlock(&qcom_scm_lock);
-- 
1.9.1

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Will Deacon]

On Sat, Aug 20, 2016 at 12:51:13AM -0500, Andy Gross wrote:
> This patch adds the SMC Session ID to the results passed back from SMC
> calls.  The Qualcomm SMC implementation provides for interrupted SMC
> functions.  When this occurs, the SMC call will return a session ID that
> is required to be used when resuming the interrupted SMC call.
> 
> Signed-off-by: Andy Gross <andy.gross@linaro.org>
> ---
>  arch/arm64/kernel/asm-offsets.c | 1 +
>  arch/arm64/kernel/smccc-call.S  | 1 +
>  include/linux/arm-smccc.h       | 4 +++-
>  3 files changed, 5 insertions(+), 1 deletion(-)

[...]

> diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
> index b5abfda..82d919f 100644
> --- a/include/linux/arm-smccc.h
> +++ b/include/linux/arm-smccc.h
> @@ -63,18 +63,20 @@
>  /**
>   * struct arm_smccc_res - Result from SMC/HVC call
>   * @a0-a3 result values from registers 0 to 3
> + * @a6 Session ID register (optional)
>   */
>  struct arm_smccc_res {
>  	unsigned long a0;
>  	unsigned long a1;
>  	unsigned long a2;
>  	unsigned long a3;
> +	unsigned long a6;
>  };
>  
>  /**
>   * arm_smccc_smc() - make SMC calls
>   * @a0-a7: arguments passed in registers 0 to 7
> - * @res: result values from registers 0 to 3
> + * @res: result values from registers 0 to 3 and optional register 6

AFAICT from reading the SMCCC spec, parameter register 6 is "Unpredictable,
Scratch registers" in return state, so I don't think this is correct.

What am I missing?

Will

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Andy Gross]

On Mon, Aug 22, 2016 at 02:43:14PM +0100, Will Deacon wrote:
> On Sat, Aug 20, 2016 at 12:51:13AM -0500, Andy Gross wrote:
> > This patch adds the SMC Session ID to the results passed back from SMC
> > calls.  The Qualcomm SMC implementation provides for interrupted SMC
> > functions.  When this occurs, the SMC call will return a session ID that
> > is required to be used when resuming the interrupted SMC call.
> > 
> > Signed-off-by: Andy Gross <andy.gross@linaro.org>
> > ---
> >  arch/arm64/kernel/asm-offsets.c | 1 +
> >  arch/arm64/kernel/smccc-call.S  | 1 +
> >  include/linux/arm-smccc.h       | 4 +++-
> >  3 files changed, 5 insertions(+), 1 deletion(-)
> 
> [...]
> 
> > diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
> > index b5abfda..82d919f 100644
> > --- a/include/linux/arm-smccc.h
> > +++ b/include/linux/arm-smccc.h
> > @@ -63,18 +63,20 @@
> >  /**
> >   * struct arm_smccc_res - Result from SMC/HVC call
> >   * @a0-a3 result values from registers 0 to 3
> > + * @a6 Session ID register (optional)
> >   */
> >  struct arm_smccc_res {
> >  	unsigned long a0;
> >  	unsigned long a1;
> >  	unsigned long a2;
> >  	unsigned long a3;
> > +	unsigned long a6;
> >  };
> >  
> >  /**
> >   * arm_smccc_smc() - make SMC calls
> >   * @a0-a7: arguments passed in registers 0 to 7
> > - * @res: result values from registers 0 to 3
> > + * @res: result values from registers 0 to 3 and optional register 6
> 
> AFAICT from reading the SMCCC spec, parameter register 6 is "Unpredictable,
> Scratch registers" in return state, so I don't think this is correct.
> 
> What am I missing?

In the case of Qualcomm's implementation, they return a value in register 6 that
may or may not be used in subsequent calls.  If I want to leverage the arm_smccc
functions, then I need to extend them to include the optional return value.  The
downside to this is that everyone who uses this is exposed to it.

Regards,

Andy

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Will Deacon]

On Mon, Aug 22, 2016 at 09:02:46AM -0500, Andy Gross wrote:
> On Mon, Aug 22, 2016 at 02:43:14PM +0100, Will Deacon wrote:
> > On Sat, Aug 20, 2016 at 12:51:13AM -0500, Andy Gross wrote:
> > > This patch adds the SMC Session ID to the results passed back from SMC
> > > calls.  The Qualcomm SMC implementation provides for interrupted SMC
> > > functions.  When this occurs, the SMC call will return a session ID that
> > > is required to be used when resuming the interrupted SMC call.
> > > 
> > > Signed-off-by: Andy Gross <andy.gross@linaro.org>
> > > ---
> > >  arch/arm64/kernel/asm-offsets.c | 1 +
> > >  arch/arm64/kernel/smccc-call.S  | 1 +
> > >  include/linux/arm-smccc.h       | 4 +++-
> > >  3 files changed, 5 insertions(+), 1 deletion(-)
> > 
> > [...]
> > 
> > > diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
> > > index b5abfda..82d919f 100644
> > > --- a/include/linux/arm-smccc.h
> > > +++ b/include/linux/arm-smccc.h
> > > @@ -63,18 +63,20 @@
> > >  /**
> > >   * struct arm_smccc_res - Result from SMC/HVC call
> > >   * @a0-a3 result values from registers 0 to 3
> > > + * @a6 Session ID register (optional)
> > >   */
> > >  struct arm_smccc_res {
> > >  	unsigned long a0;
> > >  	unsigned long a1;
> > >  	unsigned long a2;
> > >  	unsigned long a3;
> > > +	unsigned long a6;
> > >  };
> > >  
> > >  /**
> > >   * arm_smccc_smc() - make SMC calls
> > >   * @a0-a7: arguments passed in registers 0 to 7
> > > - * @res: result values from registers 0 to 3
> > > + * @res: result values from registers 0 to 3 and optional register 6
> > 
> > AFAICT from reading the SMCCC spec, parameter register 6 is "Unpredictable,
> > Scratch registers" in return state, so I don't think this is correct.
> > 
> > What am I missing?
> 
> In the case of Qualcomm's implementation, they return a value in register 6 that
> may or may not be used in subsequent calls.  If I want to leverage the arm_smccc
> functions, then I need to extend them to include the optional return value.  The
> downside to this is that everyone who uses this is exposed to it.

Yes, I'm not keen on forcing this behaviour for everybody, as you never
know what other firmware might do with unexpected a6 values. Could we
perhaps quirk it, along the lines of the completely untested patch below?

Will

--->8

diff --git a/arch/arm64/kernel/asm-offsets.c b/arch/arm64/kernel/asm-offsets.c
index 05070b72fc28..1895e87d0240 100644
--- a/arch/arm64/kernel/asm-offsets.c
+++ b/arch/arm64/kernel/asm-offsets.c
@@ -141,6 +141,8 @@ int main(void)
 #endif
   DEFINE(ARM_SMCCC_RES_X0_OFFS,	offsetof(struct arm_smccc_res, a0));
   DEFINE(ARM_SMCCC_RES_X2_OFFS,	offsetof(struct arm_smccc_res, a2));
+  DEFINE(ARM_SMCCC_RES_QUIRK_ID_OFFS,		offsetof(struct arm_smccc_res, quirk.id));
+  DEFINE(ARM_SMCCC_RES_QUIRK_STATE_OFFS,	offsetof(struct arm_smccc_res, quirk.state));
   BLANK();
   DEFINE(HIBERN_PBE_ORIG,	offsetof(struct pbe, orig_address));
   DEFINE(HIBERN_PBE_ADDR,	offsetof(struct pbe, address));
diff --git a/arch/arm64/kernel/smccc-call.S b/arch/arm64/kernel/smccc-call.S
index ae0496fa4235..3c6c976eaf5c 100644
--- a/arch/arm64/kernel/smccc-call.S
+++ b/arch/arm64/kernel/smccc-call.S
@@ -12,6 +12,7 @@
  *
  */
 #include <linux/linkage.h>
+#include <linux/arm-smccc.h>
 #include <asm/asm-offsets.h>
 
 	.macro SMCCC instr
@@ -20,7 +21,12 @@
 	ldr	x4, [sp]
 	stp	x0, x1, [x4, #ARM_SMCCC_RES_X0_OFFS]
 	stp	x2, x3, [x4, #ARM_SMCCC_RES_X2_OFFS]
-	ret
+	ldr	x9, [x4, #ARM_SMCCC_RES_QUIRK_ID_OFFS]
+	cbz	x9, 1f /* ARM_SMCCC_QUIRK_NONE */
+	cmp	x9, #ARM_SMCCC_QUIRK_QCOM_A6
+	b.ne	1f
+	str	x6, [x4, ARM_SMCCC_RES_QUIRK_STATE_OFFS]
+1:	ret
 	.cfi_endproc
 	.endm
 
diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
index b5abfda80465..a3a6e291feb6 100644
--- a/include/linux/arm-smccc.h
+++ b/include/linux/arm-smccc.h
@@ -14,9 +14,6 @@
 #ifndef __LINUX_ARM_SMCCC_H
 #define __LINUX_ARM_SMCCC_H
 
-#include <linux/linkage.h>
-#include <linux/types.h>
-
 /*
  * This file provides common defines for ARM SMC Calling Convention as
  * specified in
@@ -60,6 +57,21 @@
 #define ARM_SMCCC_OWNER_TRUSTED_OS	50
 #define ARM_SMCCC_OWNER_TRUSTED_OS_END	63
 
+#define ARM_SMCCC_QUIRK_NONE	0
+#define ARM_SMCCC_QUIRK_QCOM_A6	1 /* Save/restore register a6 */
+
+#ifndef __ASSEMBLY__
+
+#include <linux/linkage.h>
+#include <linux/types.h>
+
+struct arm_smccc_quirk {
+	int	id;
+	union {
+		unsigned long a6;
+	} state;
+};
+
 /**
  * struct arm_smccc_res - Result from SMC/HVC call
  * @a0-a3 result values from registers 0 to 3
@@ -69,6 +81,7 @@ struct arm_smccc_res {
 	unsigned long a1;
 	unsigned long a2;
 	unsigned long a3;
+	struct arm_smccc_quirk quirk;
 };
 
 /**
@@ -101,4 +114,5 @@ asmlinkage void arm_smccc_hvc(unsigned long a0, unsigned long a1,
 			unsigned long a5, unsigned long a6, unsigned long a7,
 			struct arm_smccc_res *res);
 
+#endif /* !__ASSEMBLY__ */
 #endif /*__LINUX_ARM_SMCCC_H*/

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Andy Gross]

On Mon, Aug 22, 2016 at 03:53:26PM +0100, Will Deacon wrote:
> On Mon, Aug 22, 2016 at 09:02:46AM -0500, Andy Gross wrote:
> > On Mon, Aug 22, 2016 at 02:43:14PM +0100, Will Deacon wrote:
> > > On Sat, Aug 20, 2016 at 12:51:13AM -0500, Andy Gross wrote:
> > > > This patch adds the SMC Session ID to the results passed back from SMC
> > > > calls.  The Qualcomm SMC implementation provides for interrupted SMC
> > > > functions.  When this occurs, the SMC call will return a session ID that
> > > > is required to be used when resuming the interrupted SMC call.
> > > > 
> > > > Signed-off-by: Andy Gross <andy.gross@linaro.org>
> > > > ---
> > > >  arch/arm64/kernel/asm-offsets.c | 1 +
> > > >  arch/arm64/kernel/smccc-call.S  | 1 +
> > > >  include/linux/arm-smccc.h       | 4 +++-
> > > >  3 files changed, 5 insertions(+), 1 deletion(-)
> > > 
> > > [...]
> > > 
> > > > diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
> > > > index b5abfda..82d919f 100644
> > > > --- a/include/linux/arm-smccc.h
> > > > +++ b/include/linux/arm-smccc.h
> > > > @@ -63,18 +63,20 @@
> > > >  /**
> > > >   * struct arm_smccc_res - Result from SMC/HVC call
> > > >   * @a0-a3 result values from registers 0 to 3
> > > > + * @a6 Session ID register (optional)
> > > >   */
> > > >  struct arm_smccc_res {
> > > >  	unsigned long a0;
> > > >  	unsigned long a1;
> > > >  	unsigned long a2;
> > > >  	unsigned long a3;
> > > > +	unsigned long a6;
> > > >  };
> > > >  
> > > >  /**
> > > >   * arm_smccc_smc() - make SMC calls
> > > >   * @a0-a7: arguments passed in registers 0 to 7
> > > > - * @res: result values from registers 0 to 3
> > > > + * @res: result values from registers 0 to 3 and optional register 6
> > > 
> > > AFAICT from reading the SMCCC spec, parameter register 6 is "Unpredictable,
> > > Scratch registers" in return state, so I don't think this is correct.
> > > 
> > > What am I missing?
> > 
> > In the case of Qualcomm's implementation, they return a value in register 6 that
> > may or may not be used in subsequent calls.  If I want to leverage the arm_smccc
> > functions, then I need to extend them to include the optional return value.  The
> > downside to this is that everyone who uses this is exposed to it.
> 
> Yes, I'm not keen on forcing this behaviour for everybody, as you never
> know what other firmware might do with unexpected a6 values. Could we
> perhaps quirk it, along the lines of the completely untested patch below?

A quirk would work fine.  I'll try this out and get back to you.

Thanks,

Andy


> --->8
> 
> diff --git a/arch/arm64/kernel/asm-offsets.c b/arch/arm64/kernel/asm-offsets.c
> index 05070b72fc28..1895e87d0240 100644
> --- a/arch/arm64/kernel/asm-offsets.c
> +++ b/arch/arm64/kernel/asm-offsets.c
> @@ -141,6 +141,8 @@ int main(void)
>  #endif
>    DEFINE(ARM_SMCCC_RES_X0_OFFS,	offsetof(struct arm_smccc_res, a0));
>    DEFINE(ARM_SMCCC_RES_X2_OFFS,	offsetof(struct arm_smccc_res, a2));
> +  DEFINE(ARM_SMCCC_RES_QUIRK_ID_OFFS,		offsetof(struct arm_smccc_res, quirk.id));
> +  DEFINE(ARM_SMCCC_RES_QUIRK_STATE_OFFS,	offsetof(struct arm_smccc_res, quirk.state));
>    BLANK();
>    DEFINE(HIBERN_PBE_ORIG,	offsetof(struct pbe, orig_address));
>    DEFINE(HIBERN_PBE_ADDR,	offsetof(struct pbe, address));
> diff --git a/arch/arm64/kernel/smccc-call.S b/arch/arm64/kernel/smccc-call.S
> index ae0496fa4235..3c6c976eaf5c 100644
> --- a/arch/arm64/kernel/smccc-call.S
> +++ b/arch/arm64/kernel/smccc-call.S
> @@ -12,6 +12,7 @@
>   *
>   */
>  #include <linux/linkage.h>
> +#include <linux/arm-smccc.h>
>  #include <asm/asm-offsets.h>
>  
>  	.macro SMCCC instr
> @@ -20,7 +21,12 @@
>  	ldr	x4, [sp]
>  	stp	x0, x1, [x4, #ARM_SMCCC_RES_X0_OFFS]
>  	stp	x2, x3, [x4, #ARM_SMCCC_RES_X2_OFFS]
> -	ret
> +	ldr	x9, [x4, #ARM_SMCCC_RES_QUIRK_ID_OFFS]
> +	cbz	x9, 1f /* ARM_SMCCC_QUIRK_NONE */
> +	cmp	x9, #ARM_SMCCC_QUIRK_QCOM_A6
> +	b.ne	1f
> +	str	x6, [x4, ARM_SMCCC_RES_QUIRK_STATE_OFFS]
> +1:	ret
>  	.cfi_endproc
>  	.endm
>  
> diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
> index b5abfda80465..a3a6e291feb6 100644
> --- a/include/linux/arm-smccc.h
> +++ b/include/linux/arm-smccc.h
> @@ -14,9 +14,6 @@
>  #ifndef __LINUX_ARM_SMCCC_H
>  #define __LINUX_ARM_SMCCC_H
>  
> -#include <linux/linkage.h>
> -#include <linux/types.h>
> -
>  /*
>   * This file provides common defines for ARM SMC Calling Convention as
>   * specified in
> @@ -60,6 +57,21 @@
>  #define ARM_SMCCC_OWNER_TRUSTED_OS	50
>  #define ARM_SMCCC_OWNER_TRUSTED_OS_END	63
>  
> +#define ARM_SMCCC_QUIRK_NONE	0
> +#define ARM_SMCCC_QUIRK_QCOM_A6	1 /* Save/restore register a6 */
> +
> +#ifndef __ASSEMBLY__
> +
> +#include <linux/linkage.h>
> +#include <linux/types.h>
> +
> +struct arm_smccc_quirk {
> +	int	id;
> +	union {
> +		unsigned long a6;
> +	} state;
> +};
> +
>  /**
>   * struct arm_smccc_res - Result from SMC/HVC call
>   * @a0-a3 result values from registers 0 to 3
> @@ -69,6 +81,7 @@ struct arm_smccc_res {
>  	unsigned long a1;
>  	unsigned long a2;
>  	unsigned long a3;
> +	struct arm_smccc_quirk quirk;
>  };
>  
>  /**
> @@ -101,4 +114,5 @@ asmlinkage void arm_smccc_hvc(unsigned long a0, unsigned long a1,
>  			unsigned long a5, unsigned long a6, unsigned long a7,
>  			struct arm_smccc_res *res);
>  
> +#endif /* !__ASSEMBLY__ */
>  #endif /*__LINUX_ARM_SMCCC_H*/
> --
> To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Stephen Boyd]

On 08/22, Will Deacon wrote:
> On Mon, Aug 22, 2016 at 09:02:46AM -0500, Andy Gross wrote:
> > On Mon, Aug 22, 2016 at 02:43:14PM +0100, Will Deacon wrote:
> > > On Sat, Aug 20, 2016 at 12:51:13AM -0500, Andy Gross wrote:
> > > >  
> > > >  /**
> > > >   * arm_smccc_smc() - make SMC calls
> > > >   * @a0-a7: arguments passed in registers 0 to 7
> > > > - * @res: result values from registers 0 to 3
> > > > + * @res: result values from registers 0 to 3 and optional register 6
> > > 
> > > AFAICT from reading the SMCCC spec, parameter register 6 is "Unpredictable,
> > > Scratch registers" in return state, so I don't think this is correct.
> > > 
> > > What am I missing?
> > 
> > In the case of Qualcomm's implementation, they return a value in register 6 that
> > may or may not be used in subsequent calls.  If I want to leverage the arm_smccc
> > functions, then I need to extend them to include the optional return value.  The
> > downside to this is that everyone who uses this is exposed to it.
> 
> Yes, I'm not keen on forcing this behaviour for everybody, as you never
> know what other firmware might do with unexpected a6 values. Could we
> perhaps quirk it, along the lines of the completely untested patch below?
> 

How would the firmware know about the a6 values? It isn't passed
to the firmware unless the caller of arm_smccc_smc() uses it. Is
the concern that we would be exposing x6 as a return register for
all users of arm_smccc_smc()? Presumably callers of
arm_smccc_smc() would know if they want to use that extra
register or not, so doing all the quirk stuff seems like more
instructions over always saving away x6 on the return path, but
maybe there's some reasoning I missed.

This all comes about because the firmware generates a session id
for the SMC call and jams it in x6. The assembly on the
non-secure side is written with a tight loop around the smc
instruction so that when the return value indicates
"interrupted", x6 is kept intact and the non-secure OS can jump
back to the secure OS without register reloading. Perhaps
referring to x6 as result value is not correct because it's
really a session id that's irrelevant once the smc call
completes.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Lorenzo Pieralisi]

On Mon, Aug 22, 2016 at 05:38:31PM -0700, Stephen Boyd wrote:
> On 08/22, Will Deacon wrote:
> > On Mon, Aug 22, 2016 at 09:02:46AM -0500, Andy Gross wrote:
> > > On Mon, Aug 22, 2016 at 02:43:14PM +0100, Will Deacon wrote:
> > > > On Sat, Aug 20, 2016 at 12:51:13AM -0500, Andy Gross wrote:
> > > > >  
> > > > >  /**
> > > > >   * arm_smccc_smc() - make SMC calls
> > > > >   * @a0-a7: arguments passed in registers 0 to 7
> > > > > - * @res: result values from registers 0 to 3
> > > > > + * @res: result values from registers 0 to 3 and optional register 6
> > > > 
> > > > AFAICT from reading the SMCCC spec, parameter register 6 is "Unpredictable,
> > > > Scratch registers" in return state, so I don't think this is correct.
> > > > 
> > > > What am I missing?
> > > 
> > > In the case of Qualcomm's implementation, they return a value in register 6 that
> > > may or may not be used in subsequent calls.  If I want to leverage the arm_smccc
> > > functions, then I need to extend them to include the optional return value.  The
> > > downside to this is that everyone who uses this is exposed to it.
> > 
> > Yes, I'm not keen on forcing this behaviour for everybody, as you never
> > know what other firmware might do with unexpected a6 values. Could we
> > perhaps quirk it, along the lines of the completely untested patch below?
> > 
> 
> How would the firmware know about the a6 values? It isn't passed
> to the firmware unless the caller of arm_smccc_smc() uses it. Is
> the concern that we would be exposing x6 as a return register for
> all users of arm_smccc_smc()? Presumably callers of
> arm_smccc_smc() would know if they want to use that extra
> register or not, so doing all the quirk stuff seems like more
> instructions over always saving away x6 on the return path, but
> maybe there's some reasoning I missed.

It is a concern of exposing a return register that is not a
return register according to the SMCCC.

> This all comes about because the firmware generates a session id
> for the SMC call and jams it in x6. The assembly on the
> non-secure side is written with a tight loop around the smc
> instruction so that when the return value indicates
> "interrupted", x6 is kept intact and the non-secure OS can jump
> back to the secure OS without register reloading. Perhaps
> referring to x6 as result value is not correct because it's
> really a session id that's irrelevant once the smc call
> completes.

And by using the quirk we need to reload x6 from the stack anyway
which defeats the purpose of what you want to achieve (if we have
to stash it we can do it in the caller stack and reload it before
re-issuing the SMC without touching the SMCC code at all, am I
missing something here ?).

Lorenzo

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Lorenzo Pieralisi]

On Mon, Aug 22, 2016 at 05:38:31PM -0700, Stephen Boyd wrote:

[...]

> This all comes about because the firmware generates a session id
> for the SMC call and jams it in x6. The assembly on the
> non-secure side is written with a tight loop around the smc
> instruction so that when the return value indicates
> "interrupted", x6 is kept intact and the non-secure OS can jump
> back to the secure OS without register reloading. Perhaps
> referring to x6 as result value is not correct because it's
> really a session id that's irrelevant once the smc call
> completes.

Sorry I missed this bit. The session id is _generated_ by secure
firmware (probably only when the value passed in x6 == 0 (?))
and actually returned to the caller so that subsequent (interrupted)
calls can re-issue the same value, is that correct ?

If that's the case the value in x6 is a result value from an SMCCC
perspective and your current FW is not SMCCC compliant.

Lorenzo

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Andy Gross]

On Tue, Aug 23, 2016 at 11:38:41AM +0100, Lorenzo Pieralisi wrote:
> On Mon, Aug 22, 2016 at 05:38:31PM -0700, Stephen Boyd wrote:
> 
> [...]
> 
> > This all comes about because the firmware generates a session id
> > for the SMC call and jams it in x6. The assembly on the
> > non-secure side is written with a tight loop around the smc
> > instruction so that when the return value indicates
> > "interrupted", x6 is kept intact and the non-secure OS can jump
> > back to the secure OS without register reloading. Perhaps
> > referring to x6 as result value is not correct because it's
> > really a session id that's irrelevant once the smc call
> > completes.
> 
> Sorry I missed this bit. The session id is _generated_ by secure
> firmware (probably only when the value passed in x6 == 0 (?))
> and actually returned to the caller so that subsequent (interrupted)
> calls can re-issue the same value, is that correct ?

Yes, that is exactly what is going on.  You always pass in 0 for the first call.
If the call is interrupted and needs to be re-executed, you will get a specific
result in a0 that tells you to redo the call using x6 as your session ID.

> 
> If that's the case the value in x6 is a result value from an SMCCC
> perspective and your current FW is not SMCCC compliant.

Should we then write our own ASM snippet to do exactly what we want?  It'd be
the same as the arm_smccc except with the extra str.  I'm ok with that, I was
just hoping to leverage the existing smccc code.  The quirk also works well,
except it costs everyone else 1 load and compare.

Regards,

Andy

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Andy Gross]

On Mon, Aug 22, 2016 at 10:16:40AM -0500, Andy Gross wrote:
> On Mon, Aug 22, 2016 at 03:53:26PM +0100, Will Deacon wrote:
> > On Mon, Aug 22, 2016 at 09:02:46AM -0500, Andy Gross wrote:
> > > On Mon, Aug 22, 2016 at 02:43:14PM +0100, Will Deacon wrote:
> > > > On Sat, Aug 20, 2016 at 12:51:13AM -0500, Andy Gross wrote:

<snip>

> > > 
> > > In the case of Qualcomm's implementation, they return a value in register 6 that
> > > may or may not be used in subsequent calls.  If I want to leverage the arm_smccc
> > > functions, then I need to extend them to include the optional return value.  The
> > > downside to this is that everyone who uses this is exposed to it.
> > 
> > Yes, I'm not keen on forcing this behaviour for everybody, as you never
> > know what other firmware might do with unexpected a6 values. Could we
> > perhaps quirk it, along the lines of the completely untested patch below?
> 
> A quirk would work fine.  I'll try this out and get back to you.

Update:  Aside from a minor change with either the id type or using w9 register
for the id load, this works just fine.

> > --->8
> > 
> > diff --git a/arch/arm64/kernel/asm-offsets.c b/arch/arm64/kernel/asm-offsets.c
> > index 05070b72fc28..1895e87d0240 100644
> > --- a/arch/arm64/kernel/asm-offsets.c
> > +++ b/arch/arm64/kernel/asm-offsets.c
> > @@ -141,6 +141,8 @@ int main(void)
> >  #endif
> >    DEFINE(ARM_SMCCC_RES_X0_OFFS,	offsetof(struct arm_smccc_res, a0));
> >    DEFINE(ARM_SMCCC_RES_X2_OFFS,	offsetof(struct arm_smccc_res, a2));
> > +  DEFINE(ARM_SMCCC_RES_QUIRK_ID_OFFS,		offsetof(struct arm_smccc_res, quirk.id));
> > +  DEFINE(ARM_SMCCC_RES_QUIRK_STATE_OFFS,	offsetof(struct arm_smccc_res, quirk.state));
> >    BLANK();
> >    DEFINE(HIBERN_PBE_ORIG,	offsetof(struct pbe, orig_address));
> >    DEFINE(HIBERN_PBE_ADDR,	offsetof(struct pbe, address));
> > diff --git a/arch/arm64/kernel/smccc-call.S b/arch/arm64/kernel/smccc-call.S
> > index ae0496fa4235..3c6c976eaf5c 100644
> > --- a/arch/arm64/kernel/smccc-call.S
> > +++ b/arch/arm64/kernel/smccc-call.S
> > @@ -12,6 +12,7 @@
> >   *
> >   */
> >  #include <linux/linkage.h>
> > +#include <linux/arm-smccc.h>
> >  #include <asm/asm-offsets.h>
> >  
> >  	.macro SMCCC instr
> > @@ -20,7 +21,12 @@
> >  	ldr	x4, [sp]
> >  	stp	x0, x1, [x4, #ARM_SMCCC_RES_X0_OFFS]
> >  	stp	x2, x3, [x4, #ARM_SMCCC_RES_X2_OFFS]
> > -	ret
> > +	ldr	x9, [x4, #ARM_SMCCC_RES_QUIRK_ID_OFFS]
> > +	cbz	x9, 1f /* ARM_SMCCC_QUIRK_NONE */
> > +	cmp	x9, #ARM_SMCCC_QUIRK_QCOM_A6
> > +	b.ne	1f
> > +	str	x6, [x4, ARM_SMCCC_RES_QUIRK_STATE_OFFS]
> > +1:	ret
> >  	.cfi_endproc
> >  	.endm
> >  
> > diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
> > index b5abfda80465..a3a6e291feb6 100644
> > --- a/include/linux/arm-smccc.h
> > +++ b/include/linux/arm-smccc.h
> > @@ -14,9 +14,6 @@
> >  #ifndef __LINUX_ARM_SMCCC_H
> >  #define __LINUX_ARM_SMCCC_H
> >  
> > -#include <linux/linkage.h>
> > -#include <linux/types.h>
> > -
> >  /*
> >   * This file provides common defines for ARM SMC Calling Convention as
> >   * specified in
> > @@ -60,6 +57,21 @@
> >  #define ARM_SMCCC_OWNER_TRUSTED_OS	50
> >  #define ARM_SMCCC_OWNER_TRUSTED_OS_END	63
> >  
> > +#define ARM_SMCCC_QUIRK_NONE	0
> > +#define ARM_SMCCC_QUIRK_QCOM_A6	1 /* Save/restore register a6 */
> > +
> > +#ifndef __ASSEMBLY__
> > +
> > +#include <linux/linkage.h>
> > +#include <linux/types.h>
> > +
> > +struct arm_smccc_quirk {
> > +	int	id;
> > +	union {
> > +		unsigned long a6;
> > +	} state;
> > +};
> > +
> >  /**
> >   * struct arm_smccc_res - Result from SMC/HVC call
> >   * @a0-a3 result values from registers 0 to 3
> > @@ -69,6 +81,7 @@ struct arm_smccc_res {
> >  	unsigned long a1;
> >  	unsigned long a2;
> >  	unsigned long a3;
> > +	struct arm_smccc_quirk quirk;
> >  };
> >  
> >  /**
> > @@ -101,4 +114,5 @@ asmlinkage void arm_smccc_hvc(unsigned long a0, unsigned long a1,
> >  			unsigned long a5, unsigned long a6, unsigned long a7,
> >  			struct arm_smccc_res *res);
> >  
> > +#endif /* !__ASSEMBLY__ */
> >  #endif /*__LINUX_ARM_SMCCC_H*/
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Bjorn Andersson]

On Fri 19 Aug 22:51 PDT 2016, Andy Gross wrote:

> This patch adds the SMC Session ID to the results passed back from SMC
> calls.  The Qualcomm SMC implementation provides for interrupted SMC
> functions.  When this occurs, the SMC call will return a session ID that
> is required to be used when resuming the interrupted SMC call.
> 
> Signed-off-by: Andy Gross <andy.gross@linaro.org>

This fixes the spurious remoteproc firmware authentication failures I've
seen lately.

Tested-by: Bjorn Andersson <bjorn.andersson@linaro.org>

Regards,
Bjorn

> ---
>  arch/arm64/kernel/asm-offsets.c | 1 +
>  arch/arm64/kernel/smccc-call.S  | 1 +
>  include/linux/arm-smccc.h       | 4 +++-
>  3 files changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/kernel/asm-offsets.c b/arch/arm64/kernel/asm-offsets.c
> index 05070b7..ff38c58 100644
> --- a/arch/arm64/kernel/asm-offsets.c
> +++ b/arch/arm64/kernel/asm-offsets.c
> @@ -141,6 +141,7 @@ int main(void)
>  #endif
>    DEFINE(ARM_SMCCC_RES_X0_OFFS,	offsetof(struct arm_smccc_res, a0));
>    DEFINE(ARM_SMCCC_RES_X2_OFFS,	offsetof(struct arm_smccc_res, a2));
> +  DEFINE(ARM_SMCCC_RES_X6_OFFS,	offsetof(struct arm_smccc_res, a6));
>    BLANK();
>    DEFINE(HIBERN_PBE_ORIG,	offsetof(struct pbe, orig_address));
>    DEFINE(HIBERN_PBE_ADDR,	offsetof(struct pbe, address));
> diff --git a/arch/arm64/kernel/smccc-call.S b/arch/arm64/kernel/smccc-call.S
> index ae0496f..278dc9a 100644
> --- a/arch/arm64/kernel/smccc-call.S
> +++ b/arch/arm64/kernel/smccc-call.S
> @@ -20,6 +20,7 @@
>  	ldr	x4, [sp]
>  	stp	x0, x1, [x4, #ARM_SMCCC_RES_X0_OFFS]
>  	stp	x2, x3, [x4, #ARM_SMCCC_RES_X2_OFFS]
> +	str	x6, [x4, #ARM_SMCCC_RES_X6_OFFS]
>  	ret
>  	.cfi_endproc
>  	.endm
> diff --git a/include/linux/arm-smccc.h b/include/linux/arm-smccc.h
> index b5abfda..82d919f 100644
> --- a/include/linux/arm-smccc.h
> +++ b/include/linux/arm-smccc.h
> @@ -63,18 +63,20 @@
>  /**
>   * struct arm_smccc_res - Result from SMC/HVC call
>   * @a0-a3 result values from registers 0 to 3
> + * @a6 Session ID register (optional)
>   */
>  struct arm_smccc_res {
>  	unsigned long a0;
>  	unsigned long a1;
>  	unsigned long a2;
>  	unsigned long a3;
> +	unsigned long a6;
>  };
>  
>  /**
>   * arm_smccc_smc() - make SMC calls
>   * @a0-a7: arguments passed in registers 0 to 7
> - * @res: result values from registers 0 to 3
> + * @res: result values from registers 0 to 3 and optional register 6
>   *
>   * This function is used to make SMC calls following SMC Calling Convention.
>   * The content of the supplied param are copied to registers 0 to 7 prior
> -- 
> 1.9.1
> 

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Andy Gross]

On Tue, Aug 23, 2016 at 11:38:41AM +0100, Lorenzo Pieralisi wrote:
> On Mon, Aug 22, 2016 at 05:38:31PM -0700, Stephen Boyd wrote:
> 
> [...]
> 
> > This all comes about because the firmware generates a session id
> > for the SMC call and jams it in x6. The assembly on the
> > non-secure side is written with a tight loop around the smc
> > instruction so that when the return value indicates
> > "interrupted", x6 is kept intact and the non-secure OS can jump
> > back to the secure OS without register reloading. Perhaps
> > referring to x6 as result value is not correct because it's
> > really a session id that's irrelevant once the smc call
> > completes.
> 
> Sorry I missed this bit. The session id is _generated_ by secure
> firmware (probably only when the value passed in x6 == 0 (?))
> and actually returned to the caller so that subsequent (interrupted)
> calls can re-issue the same value, is that correct ?
> 
> If that's the case the value in x6 is a result value from an SMCCC
> perspective and your current FW is not SMCCC compliant.
> 

So is Will's solution to this ok?  If so I will respin with the minor change to
get it working and resend.  If not, do I roll my own smccc wrapper?

Regards,

Andy

Re: [PATCH 1/2] arm64: kernel: Add SMC Session ID to results

[Will Deacon]

On Tue, Aug 30, 2016 at 03:16:42PM -0500, Andy Gross wrote:
> On Tue, Aug 23, 2016 at 11:38:41AM +0100, Lorenzo Pieralisi wrote:
> > On Mon, Aug 22, 2016 at 05:38:31PM -0700, Stephen Boyd wrote:
> > 
> > [...]
> > 
> > > This all comes about because the firmware generates a session id
> > > for the SMC call and jams it in x6. The assembly on the
> > > non-secure side is written with a tight loop around the smc
> > > instruction so that when the return value indicates
> > > "interrupted", x6 is kept intact and the non-secure OS can jump
> > > back to the secure OS without register reloading. Perhaps
> > > referring to x6 as result value is not correct because it's
> > > really a session id that's irrelevant once the smc call
> > > completes.
> > 
> > Sorry I missed this bit. The session id is _generated_ by secure
> > firmware (probably only when the value passed in x6 == 0 (?))
> > and actually returned to the caller so that subsequent (interrupted)
> > calls can re-issue the same value, is that correct ?
> > 
> > If that's the case the value in x6 is a result value from an SMCCC
> > perspective and your current FW is not SMCCC compliant.
> > 
> 
> So is Will's solution to this ok?  If so I will respin with the minor change to
> get it working and resend.  If not, do I roll my own smccc wrapper?

Obviously I'm biased, but I prefer to handle this as a quirk to make it
clear that it's a vendor-specific extension to the SMCCC, so if you
could post a patch based on the diff I sent, that would be great.

You'll also need to:

  (1) Make sure you don't break 32-bit ARM
  (2) Make sure that struct arm_smccc_res is always zero-initialised by
      its other users (to ensure that QUIRK_NONE is set). In fact, it
      might be nicer to pass the quirk structure as a separate argument,
      rather than embed it in arm_smccc_res.

Will