[PATCH 1/4] powerpc/boot/fdt: Add little-endian support to libfdt wrappers

[PATCH 1/4] powerpc/boot/fdt: Add little-endian support to libfdt wrappers

[Jeremy Kerr]

For epapr-style boot, we may be little-endian. This change implements
the proper conversion for fdt*_to_cpu and cpu_to_fdt*. We also need the
full cpu_to_* and *_to_cpu macros for this.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

---
 arch/powerpc/boot/libfdt_env.h |   14 ++++++++------
 arch/powerpc/boot/of.h         |    8 ++++++++
 2 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/arch/powerpc/boot/libfdt_env.h b/arch/powerpc/boot/libfdt_env.h
index c89fdb1..8dcd744 100644
--- a/arch/powerpc/boot/libfdt_env.h
+++ b/arch/powerpc/boot/libfdt_env.h
@@ -4,15 +4,17 @@
 #include <types.h>
 #include <string.h>
 
+#include "of.h"
+
 typedef u32 uint32_t;
 typedef u64 uint64_t;
 typedef unsigned long uintptr_t;
 
-#define fdt16_to_cpu(x)		(x)
-#define cpu_to_fdt16(x)		(x)
-#define fdt32_to_cpu(x)		(x)
-#define cpu_to_fdt32(x)		(x)
-#define fdt64_to_cpu(x)		(x)
-#define cpu_to_fdt64(x)		(x)
+#define fdt16_to_cpu(x)		be16_to_cpu(x)
+#define cpu_to_fdt16(x)		cpu_to_be16(x)
+#define fdt32_to_cpu(x)		be32_to_cpu(x)
+#define cpu_to_fdt32(x)		cpu_to_be32(x)
+#define fdt64_to_cpu(x)		be64_to_cpu(x)
+#define cpu_to_fdt64(x)		cpu_to_be64(x)
 
 #endif /* _ARCH_POWERPC_BOOT_LIBFDT_ENV_H */
diff --git a/arch/powerpc/boot/of.h b/arch/powerpc/boot/of.h
index c8c1750..5603320 100644
--- a/arch/powerpc/boot/of.h
+++ b/arch/powerpc/boot/of.h
@@ -24,11 +24,19 @@ void of_console_init(void);
 typedef u32			__be32;
 
 #ifdef __LITTLE_ENDIAN__
+#define cpu_to_be16(x) swab16(x)
+#define be16_to_cpu(x) swab16(x)
 #define cpu_to_be32(x) swab32(x)
 #define be32_to_cpu(x) swab32(x)
+#define cpu_to_be64(x) swab64(x)
+#define be64_to_cpu(x) swab64(x)
 #else
+#define cpu_to_be16(x) (x)
+#define be16_to_cpu(x) (x)
 #define cpu_to_be32(x) (x)
 #define be32_to_cpu(x) (x)
+#define cpu_to_be64(x) (x)
+#define be64_to_cpu(x) (x)
 #endif
 
 #define PROM_ERROR (-1u)

[PATCH 2/4] powerpc/boot/wrapper: use the pseries wrapper for zImage.epapr

[Jeremy Kerr]

We'll likely be entering the zImage.epapr as BE, so include the pseries
implementation of _zimage_start, which adds the endian fixup magic.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

---
 arch/powerpc/boot/wrapper |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/boot/wrapper b/arch/powerpc/boot/wrapper
index ae0f88e..3f50c27 100755
--- a/arch/powerpc/boot/wrapper
+++ b/arch/powerpc/boot/wrapper
@@ -277,7 +277,7 @@ treeboot-iss4xx-mpic)
     platformo="$object/treeboot-iss4xx.o"
     ;;
 epapr)
-    platformo="$object/epapr.o $object/epapr-wrapper.o"
+    platformo="$object/pseries-head.o $object/epapr.o $object/epapr-wrapper.o"
     link_address='0x20000000'
     pie=-pie
     ;;

Re: [PATCH 2/4] powerpc/boot/wrapper: use the pseries wrapper for zImage.epapr

[Benjamin Herrenschmidt]

On Tue, 2015-02-10 at 20:15 +0800, Jeremy Kerr wrote:
> We'll likely be entering the zImage.epapr as BE, so include the pseries
> implementation of _zimage_start, which adds the endian fixup magic.

Wont that break embedded epapr stuff ?

> Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
> 
> ---
>  arch/powerpc/boot/wrapper |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/powerpc/boot/wrapper b/arch/powerpc/boot/wrapper
> index ae0f88e..3f50c27 100755
> --- a/arch/powerpc/boot/wrapper
> +++ b/arch/powerpc/boot/wrapper
> @@ -277,7 +277,7 @@ treeboot-iss4xx-mpic)
>      platformo="$object/treeboot-iss4xx.o"
>      ;;
>  epapr)
> -    platformo="$object/epapr.o $object/epapr-wrapper.o"
> +    platformo="$object/pseries-head.o $object/epapr.o $object/epapr-wrapper.o"
>      link_address='0x20000000'
>      pie=-pie
>      ;;
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev

Re: [PATCH 2/4] powerpc/boot/wrapper: use the pseries wrapper for zImage.epapr

[Jeremy Kerr]

Hi Ben,

>> We'll likely be entering the zImage.epapr as BE, so include the pseries
>> implementation of _zimage_start, which adds the endian fixup magic.
> 
> Wont that break embedded epapr stuff ?

How so? This will just give us the endian fixup trampoline (followed by
a branch to _zimage_start_lib) as the entry point, rather than directly
entering _zimage_start_lib. Will that not work on embedded?

Cheers,


Jeremy

Re: [PATCH 2/4] powerpc/boot/wrapper: use the pseries wrapper for zImage.epapr

[Benjamin Herrenschmidt]

On Wed, 2015-02-11 at 10:16 +0800, Jeremy Kerr wrote:
> Hi Ben,
> 
> >> We'll likely be entering the zImage.epapr as BE, so include the pseries
> >> implementation of _zimage_start, which adds the endian fixup magic.
> > 
> > Wont that break embedded epapr stuff ?
> 
> How so? This will just give us the endian fixup trampoline (followed by
> a branch to _zimage_start_lib) as the entry point, rather than directly
> entering _zimage_start_lib. Will that not work on embedded?

Hrm, the trampoline only works on server, so it would have a weird
effect if booting embedded with the wrong endian, but ok.

Ben.

[PATCH 3/4] powerpc/boot: Fix stack corruption in epapr entry point

[Jeremy Kerr]

Currently, a 64-bit little-endian zImage.epapr won't boot in epapr mode,
as as we never return from platform_init.

Before entering C, we initialise our stack by setting r1 16 bytes below
the end of the _bss_stack:

  stwu	r0,-16(r1)	/* establish a stack frame */

However, the called function will save the caller's lr in the caller's
frame's lr save area, at -16(r1) to -32(r1).

This means that writes to the fdt variable corrupt the saved link
register:

 0000000020c06018 l     O .bss   0000000000001000 _bss_stack
 0000000020c07018 l     O .bss   0000000000000008 fdt

This change allocates the minimum of 32 bytes for the base of the stack
instead.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

---
 arch/powerpc/boot/crt0.S |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/boot/crt0.S b/arch/powerpc/boot/crt0.S
index 14de4f8..bf8eeee 100644
--- a/arch/powerpc/boot/crt0.S
+++ b/arch/powerpc/boot/crt0.S
@@ -218,7 +218,7 @@ p_base:	mflr	r10		/* r10 now points to runtime addr of p_base */
 	beq	6f
 	ld	r1,0(r8)
 	li	r0,0
-	stdu	r0,-16(r1)	/* establish a stack frame */
+	stdu	r0,-32(r1)	/* establish a stack frame */
 6:
 #endif  /* __powerpc64__ */
 	/* Call platform_init() */

Re: [PATCH 3/4] powerpc/boot: Fix stack corruption in epapr entry point

[Benjamin Herrenschmidt]

On Tue, 2015-02-10 at 20:15 +0800, Jeremy Kerr wrote:
> Currently, a 64-bit little-endian zImage.epapr won't boot in epapr mode,
> as as we never return from platform_init.
> 
> Before entering C, we initialise our stack by setting r1 16 bytes below
> the end of the _bss_stack:
> 
>   stwu	r0,-16(r1)	/* establish a stack frame */
> 
> However, the called function will save the caller's lr in the caller's
> frame's lr save area, at -16(r1) to -32(r1).
> 
> This means that writes to the fdt variable corrupt the saved link
> register:
> 
>  0000000020c06018 l     O .bss   0000000000001000 _bss_stack
>  0000000020c07018 l     O .bss   0000000000000008 fdt
> 
> This change allocates the minimum of 32 bytes for the base of the stack
> instead.

Is that sufficient for ABI v1 ? It feels like it's broken for BE as
well...

Ben.

> Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
> 
> ---
>  arch/powerpc/boot/crt0.S |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/powerpc/boot/crt0.S b/arch/powerpc/boot/crt0.S
> index 14de4f8..bf8eeee 100644
> --- a/arch/powerpc/boot/crt0.S
> +++ b/arch/powerpc/boot/crt0.S
> @@ -218,7 +218,7 @@ p_base:	mflr	r10		/* r10 now points to runtime addr of p_base */
>  	beq	6f
>  	ld	r1,0(r8)
>  	li	r0,0
> -	stdu	r0,-16(r1)	/* establish a stack frame */
> +	stdu	r0,-32(r1)	/* establish a stack frame */
>  6:
>  #endif  /* __powerpc64__ */
>  	/* Call platform_init() */
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev

[PATCH 4/4] powerpc/boot: don't clobber r6 and r7 in epapr boot

[Jeremy Kerr]

We use r6 and r7 for epapr boot, but the current pre-C init will clobber
both of these.

This change does a simple replacement, of r6 -> r12 and r7 -> r13, so
that we hit platform init with these registers intact.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

---
 arch/powerpc/boot/crt0.S |   24 ++++++++++++------------
 1 file changed, 12 insertions(+), 12 deletions(-)

diff --git a/arch/powerpc/boot/crt0.S b/arch/powerpc/boot/crt0.S
index bf8eeee..7e50211 100644
--- a/arch/powerpc/boot/crt0.S
+++ b/arch/powerpc/boot/crt0.S
@@ -155,29 +155,29 @@ p_base:	mflr	r10		/* r10 now points to runtime addr of p_base */
 	ld	r9,(p_rela-p_base)(r10)
 	add	r9,r9,r10
 
-	li	r7,0
+	li	r13,0
 	li	r8,0
-9:	ld	r6,0(r11)       /* get tag */
-	cmpdi	r6,0
+9:	ld	r12,0(r11)       /* get tag */
+	cmpdi	r12,0
 	beq	12f              /* end of list */
-	cmpdi	r6,RELA
+	cmpdi	r12,RELA
 	bne	10f
-	ld	r7,8(r11)       /* get RELA pointer in r7 */
+	ld	r13,8(r11)       /* get RELA pointer in r13 */
 	b	11f
-10:	addis	r6,r6,(-RELACOUNT)@ha
-	cmpdi	r6,RELACOUNT@l
+10:	addis	r12,r12,(-RELACOUNT)@ha
+	cmpdi	r12,RELACOUNT@l
 	bne	11f
 	ld	r8,8(r11)       /* get RELACOUNT value in r8 */
 11:	addi	r11,r11,16
 	b	9b
 12:
-	cmpdi	r7,0            /* check we have both RELA and RELACOUNT */
+	cmpdi	r13,0            /* check we have both RELA and RELACOUNT */
 	cmpdi	cr1,r8,0
 	beq	3f
 	beq	cr1,3f
 
 	/* Calcuate the runtime offset. */
-	subf	r7,r7,r9
+	subf	r13,r13,r9
 
 	/* Run through the list of relocations and process the
 	 * R_PPC64_RELATIVE ones. */
@@ -185,10 +185,10 @@ p_base:	mflr	r10		/* r10 now points to runtime addr of p_base */
 13:	ld	r0,8(r9)        /* ELF64_R_TYPE(reloc->r_info) */
 	cmpdi	r0,22           /* R_PPC64_RELATIVE */
 	bne	3f
-	ld	r6,0(r9)        /* reloc->r_offset */
+	ld	r12,0(r9)        /* reloc->r_offset */
 	ld	r0,16(r9)       /* reloc->r_addend */
-	add	r0,r0,r7
-	stdx	r0,r7,r6
+	add	r0,r0,r13
+	stdx	r0,r13,r12
 	addi	r9,r9,24
 	bdnz	13b
 

Re: [PATCH 4/4] powerpc/boot: don't clobber r6 and r7 in epapr boot

[Benjamin Herrenschmidt]

On Tue, 2015-02-10 at 20:15 +0800, Jeremy Kerr wrote:
> We use r6 and r7 for epapr boot, but the current pre-C init will clobber
> both of these.
> 
> This change does a simple replacement, of r6 -> r12 and r7 -> r13, so
> that we hit platform init with these registers intact.
> 
> Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

You might want to also remove the use of r8 and r9 which are supposed
to contain the OPAL entry and base for debugging purposes. We would also
have to restore them before booting the kernel.

[ Note: kexec is also missing that ]

Ben.

> ---
>  arch/powerpc/boot/crt0.S |   24 ++++++++++++------------
>  1 file changed, 12 insertions(+), 12 deletions(-)
> 
> diff --git a/arch/powerpc/boot/crt0.S b/arch/powerpc/boot/crt0.S
> index bf8eeee..7e50211 100644
> --- a/arch/powerpc/boot/crt0.S
> +++ b/arch/powerpc/boot/crt0.S
> @@ -155,29 +155,29 @@ p_base:	mflr	r10		/* r10 now points to runtime addr of p_base */
>  	ld	r9,(p_rela-p_base)(r10)
>  	add	r9,r9,r10
>  
> -	li	r7,0
> +	li	r13,0
>  	li	r8,0
> -9:	ld	r6,0(r11)       /* get tag */
> -	cmpdi	r6,0
> +9:	ld	r12,0(r11)       /* get tag */
> +	cmpdi	r12,0
>  	beq	12f              /* end of list */
> -	cmpdi	r6,RELA
> +	cmpdi	r12,RELA
>  	bne	10f
> -	ld	r7,8(r11)       /* get RELA pointer in r7 */
> +	ld	r13,8(r11)       /* get RELA pointer in r13 */
>  	b	11f
> -10:	addis	r6,r6,(-RELACOUNT)@ha
> -	cmpdi	r6,RELACOUNT@l
> +10:	addis	r12,r12,(-RELACOUNT)@ha
> +	cmpdi	r12,RELACOUNT@l
>  	bne	11f
>  	ld	r8,8(r11)       /* get RELACOUNT value in r8 */
>  11:	addi	r11,r11,16
>  	b	9b
>  12:
> -	cmpdi	r7,0            /* check we have both RELA and RELACOUNT */
> +	cmpdi	r13,0            /* check we have both RELA and RELACOUNT */
>  	cmpdi	cr1,r8,0
>  	beq	3f
>  	beq	cr1,3f
>  
>  	/* Calcuate the runtime offset. */
> -	subf	r7,r7,r9
> +	subf	r13,r13,r9
>  
>  	/* Run through the list of relocations and process the
>  	 * R_PPC64_RELATIVE ones. */
> @@ -185,10 +185,10 @@ p_base:	mflr	r10		/* r10 now points to runtime addr of p_base */
>  13:	ld	r0,8(r9)        /* ELF64_R_TYPE(reloc->r_info) */
>  	cmpdi	r0,22           /* R_PPC64_RELATIVE */
>  	bne	3f
> -	ld	r6,0(r9)        /* reloc->r_offset */
> +	ld	r12,0(r9)        /* reloc->r_offset */
>  	ld	r0,16(r9)       /* reloc->r_addend */
> -	add	r0,r0,r7
> -	stdx	r0,r7,r6
> +	add	r0,r0,r13
> +	stdx	r0,r13,r12
>  	addi	r9,r9,24
>  	bdnz	13b
>  
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev

Re: [PATCH 4/4] powerpc/boot: don't clobber r6 and r7 in epapr boot

[Jeremy Kerr]

Hi Ben,

> You might want to also remove the use of r8 and r9 which are supposed
> to contain the OPAL entry and base for debugging purposes. We would also
> have to restore them before booting the kernel.

I'd prefer that to be a separate change, if that's okay - it'd require
us to change the kentry prototype, and would percolate to epapr_platform
init too.

Cheers,


Jeremy

Re: [PATCH 4/4] powerpc/boot: don't clobber r6 and r7 in epapr boot

[Benjamin Herrenschmidt]

On Wed, 2015-02-11 at 12:27 +0800, Jeremy Kerr wrote:
> Hi Ben,
> 
> > You might want to also remove the use of r8 and r9 which are supposed
> > to contain the OPAL entry and base for debugging purposes. We would also
> > have to restore them before booting the kernel.
> 
> I'd prefer that to be a separate change, if that's okay - it'd require
> us to change the kentry prototype, and would percolate to epapr_platform
> init too.

Ok.

Ben.

Re: [PATCH 1/4] powerpc/boot/fdt: Add little-endian support to libfdt wrappers

[Benjamin Herrenschmidt]

On Tue, 2015-02-10 at 20:15 +0800, Jeremy Kerr wrote:
> For epapr-style boot, we may be little-endian. This change implements
> the proper conversion for fdt*_to_cpu and cpu_to_fdt*. We also need the
> full cpu_to_* and *_to_cpu macros for this.
> 
> Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

> ---
>  arch/powerpc/boot/libfdt_env.h |   14 ++++++++------
>  arch/powerpc/boot/of.h         |    8 ++++++++
>  2 files changed, 16 insertions(+), 6 deletions(-)
> 
> diff --git a/arch/powerpc/boot/libfdt_env.h b/arch/powerpc/boot/libfdt_env.h
> index c89fdb1..8dcd744 100644
> --- a/arch/powerpc/boot/libfdt_env.h
> +++ b/arch/powerpc/boot/libfdt_env.h
> @@ -4,15 +4,17 @@
>  #include <types.h>
>  #include <string.h>
>  
> +#include "of.h"
> +
>  typedef u32 uint32_t;
>  typedef u64 uint64_t;
>  typedef unsigned long uintptr_t;
>  
> -#define fdt16_to_cpu(x)		(x)
> -#define cpu_to_fdt16(x)		(x)
> -#define fdt32_to_cpu(x)		(x)
> -#define cpu_to_fdt32(x)		(x)
> -#define fdt64_to_cpu(x)		(x)
> -#define cpu_to_fdt64(x)		(x)
> +#define fdt16_to_cpu(x)		be16_to_cpu(x)
> +#define cpu_to_fdt16(x)		cpu_to_be16(x)
> +#define fdt32_to_cpu(x)		be32_to_cpu(x)
> +#define cpu_to_fdt32(x)		cpu_to_be32(x)
> +#define fdt64_to_cpu(x)		be64_to_cpu(x)
> +#define cpu_to_fdt64(x)		cpu_to_be64(x)
>  
>  #endif /* _ARCH_POWERPC_BOOT_LIBFDT_ENV_H */
> diff --git a/arch/powerpc/boot/of.h b/arch/powerpc/boot/of.h
> index c8c1750..5603320 100644
> --- a/arch/powerpc/boot/of.h
> +++ b/arch/powerpc/boot/of.h
> @@ -24,11 +24,19 @@ void of_console_init(void);
>  typedef u32			__be32;
>  
>  #ifdef __LITTLE_ENDIAN__
> +#define cpu_to_be16(x) swab16(x)
> +#define be16_to_cpu(x) swab16(x)
>  #define cpu_to_be32(x) swab32(x)
>  #define be32_to_cpu(x) swab32(x)
> +#define cpu_to_be64(x) swab64(x)
> +#define be64_to_cpu(x) swab64(x)
>  #else
> +#define cpu_to_be16(x) (x)
> +#define be16_to_cpu(x) (x)
>  #define cpu_to_be32(x) (x)
>  #define be32_to_cpu(x) (x)
> +#define cpu_to_be64(x) (x)
> +#define be64_to_cpu(x) (x)
>  #endif
>  
>  #define PROM_ERROR (-1u)
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev