ARM: decompressor: Enable unaligned memory access for v6 and above

Modern GCC can generate code which makes use of the CPU's native
unaligned memory access capabilities.  This is useful for the C
decompressor implementations used for unpacking compressed kernels.

This patch disables alignment faults and enables the v6 unaligned
access model on CPUs which support these features (i.e., v6 and
later), allowing full unaligned access support for C code in the
decompressor.

The decompressor C code must not be built to assume that unaligned
access works if support for v5 or older platforms is included in
the kernel.

For correct code generation, C decompressor code must always use
the get_unaligned and put_unaligned accessors when dealing with
unaligned pointers, regardless of this patch.

Signed-off-by: Dave Martin <dave.martin@linaro.org>
Acked-by: Nicolas Pitre <nico@linaro.org>
---
This is the same as the previous post, with an additional comment
in the commit message regarding the use of {get,put}_unaligned,
as suggested by Nico.

Tested on ARM1136JF-S (Integrator/CP) and ARM1176JZF-S (RealView
PB1176JZF-S).  ARM1176 is like v7 regarding the MIDR and SCTLR
alignment control bits, so this tests the v7 code path.

 arch/arm/boot/compressed/head.S |   14 +++++++++++++-
 1 files changed, 13 insertions(+), 1 deletions(-)

On Tue, Nov 06, 2012 at 12:41:51PM +0000, Dave Martin wrote:
> Modern GCC can generate code which makes use of the CPU's native
> unaligned memory access capabilities.  This is useful for the C
> decompressor implementations used for unpacking compressed kernels.
> 
> This patch disables alignment faults and enables the v6 unaligned
> access model on CPUs which support these features (i.e., v6 and
> later), allowing full unaligned access support for C code in the
> decompressor.
> 
> The decompressor C code must not be built to assume that unaligned
> access works if support for v5 or older platforms is included in
> the kernel.
> 
> For correct code generation, C decompressor code must always use
> the get_unaligned and put_unaligned accessors when dealing with
> unaligned pointers, regardless of this patch.
> 
> Signed-off-by: Dave Martin <dave.martin@linaro.org>
> Acked-by: Nicolas Pitre <nico@linaro.org>

Does anyone have any further comments on this before I send it to the
patch system?

Cheers
---Dave

> ---
> This is the same as the previous post, with an additional comment
> in the commit message regarding the use of {get,put}_unaligned,
> as suggested by Nico.
> 
> Tested on ARM1136JF-S (Integrator/CP) and ARM1176JZF-S (RealView
> PB1176JZF-S).  ARM1176 is like v7 regarding the MIDR and SCTLR
> alignment control bits, so this tests the v7 code path.
> 
>  arch/arm/boot/compressed/head.S |   14 +++++++++++++-
>  1 files changed, 13 insertions(+), 1 deletions(-)
> 
> diff --git a/arch/arm/boot/compressed/head.S b/arch/arm/boot/compressed/head.S
> index 90275f0..49ca86e 100644
> --- a/arch/arm/boot/compressed/head.S
> +++ b/arch/arm/boot/compressed/head.S
> @@ -652,6 +652,15 @@ __setup_mmu:	sub	r3, r4, #16384		@ Page directory size
>  		mov	pc, lr
>  ENDPROC(__setup_mmu)
>  
> +@ Enable unaligned access on v6, to allow better code generation
> +@ for the decompressor C code:
> +__armv6_mmu_cache_on:
> +		mrc	p15, 0, r0, c1, c0, 0	@ read SCTLR
> +		bic	r0, r0, #2		@ A (no unaligned access fault)
> +		orr	r0, r0, #1 << 22	@ U (v6 unaligned access model)
> +		mcr	p15, 0, r0, c1, c0, 0	@ write SCTLR
> +		b	__armv4_mmu_cache_on
> +
>  __arm926ejs_mmu_cache_on:
>  #ifdef CONFIG_CPU_DCACHE_WRITETHROUGH
>  		mov	r0, #4			@ put dcache in WT mode
> @@ -694,6 +703,9 @@ __armv7_mmu_cache_on:
>  		bic	r0, r0, #1 << 28	@ clear SCTLR.TRE
>  		orr	r0, r0, #0x5000		@ I-cache enable, RR cache replacement
>  		orr	r0, r0, #0x003c		@ write buffer
> +		bic	r0, r0, #2		@ A (no unaligned access fault)
> +		orr	r0, r0, #1 << 22	@ U (v6 unaligned access model)
> +						@ (needed for ARM1176)
>  #ifdef CONFIG_MMU
>  #ifdef CONFIG_CPU_ENDIAN_BE8
>  		orr	r0, r0, #1 << 25	@ big-endian page tables
> @@ -914,7 +926,7 @@ proc_types:
>  
>  		.word	0x0007b000		@ ARMv6
>  		.word	0x000ff000
> -		W(b)	__armv4_mmu_cache_on
> +		W(b)	__armv6_mmu_cache_on
>  		W(b)	__armv4_mmu_cache_off
>  		W(b)	__armv6_mmu_cache_flush
>  
> -- 
> 1.7.4.1
>

On Tue, Nov 20, 2012 at 02:49:34PM +0000, Dave Martin wrote:
> On Tue, Nov 06, 2012 at 12:41:51PM +0000, Dave Martin wrote:
> > Modern GCC can generate code which makes use of the CPU's native
> > unaligned memory access capabilities.  This is useful for the C
> > decompressor implementations used for unpacking compressed kernels.
> > 
> > This patch disables alignment faults and enables the v6 unaligned
> > access model on CPUs which support these features (i.e., v6 and
> > later), allowing full unaligned access support for C code in the
> > decompressor.
> > 
> > The decompressor C code must not be built to assume that unaligned
> > access works if support for v5 or older platforms is included in
> > the kernel.
> > 
> > For correct code generation, C decompressor code must always use
> > the get_unaligned and put_unaligned accessors when dealing with
> > unaligned pointers, regardless of this patch.
> > 
> > Signed-off-by: Dave Martin <dave.martin@linaro.org>
> > Acked-by: Nicolas Pitre <nico@linaro.org>
> 
> Does anyone have any further comments on this before I send it to the
> patch system?

I don't, I think it's the right solution.  We keep ARMv6+ all working
the same, and this seems to be what GCC expects.