[PATCH] Remove ARM errata Workarounds 458693 and 460075

[PATCH] Remove ARM errata Workarounds 458693 and 460075

[Nisal Menuka]

According to ARM, these errata exist only in a version of Cortex-A8
(r2p0) which was never built. Therefore, I believe there are no platforms
where this workaround should be enabled.
link :http://infocenter.arm.com/help/index.jsp?topic=
/com.arm.doc.faqs/ka15634.html

Signed-off-by: Nisal Menuka <nisalmenuka23@gmail.com>
---
 arch/arm/Kconfig      | 27 ---------------------------
 arch/arm/mm/proc-v7.S | 14 --------------
 2 files changed, 41 deletions(-)

diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index 0d4e71b..d527963 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -1005,33 +1005,6 @@ config ARM_ERRATA_430973
 	  Note that setting specific bits in the ACTLR register may not be
 	  available in non-secure mode.
 
-config ARM_ERRATA_458693
-	bool "ARM errata: Processor deadlock when a false hazard is created"
-	depends on CPU_V7
-	depends on !ARCH_MULTIPLATFORM
-	help
-	  This option enables the workaround for the 458693 Cortex-A8 (r2p0)
-	  erratum. For very specific sequences of memory operations, it is
-	  possible for a hazard condition intended for a cache line to instead
-	  be incorrectly associated with a different cache line. This false
-	  hazard might then cause a processor deadlock. The workaround enables
-	  the L1 caching of the NEON accesses and disables the PLD instruction
-	  in the ACTLR register. Note that setting specific bits in the ACTLR
-	  register may not be available in non-secure mode.
-
-config ARM_ERRATA_460075
-	bool "ARM errata: Data written to the L2 cache can be overwritten with stale data"
-	depends on CPU_V7
-	depends on !ARCH_MULTIPLATFORM
-	help
-	  This option enables the workaround for the 460075 Cortex-A8 (r2p0)
-	  erratum. Any asynchronous access to the L2 cache may encounter a
-	  situation in which recent store transactions to the L2 cache are lost
-	  and overwritten with stale memory contents from external memory. The
-	  workaround disables the write-allocate mode for the L2 cache via the
-	  ACTLR register. Note that setting specific bits in the ACTLR register
-	  may not be available in non-secure mode.
-
 config ARM_ERRATA_742230
 	bool "ARM errata: DMB operation may be faulty"
 	depends on CPU_V7 && SMP
diff --git a/arch/arm/mm/proc-v7.S b/arch/arm/mm/proc-v7.S
index d00d52c..43a4a12 100644
--- a/arch/arm/mm/proc-v7.S
+++ b/arch/arm/mm/proc-v7.S
@@ -306,20 +306,6 @@ __ca8_errata:
 	orreq	r0, r0, #(1 << 6)		@ set IBE to 1
 	mcreq	p15, 0, r0, c1, c0, 1		@ write aux control register
 #endif
-#ifdef CONFIG_ARM_ERRATA_458693
-	teq	r6, #0x20			@ only present in r2p0
-	mrceq	p15, 0, r0, c1, c0, 1		@ read aux control register
-	orreq	r0, r0, #(1 << 5)		@ set L1NEON to 1
-	orreq	r0, r0, #(1 << 9)		@ set PLDNOP to 1
-	mcreq	p15, 0, r0, c1, c0, 1		@ write aux control register
-#endif
-#ifdef CONFIG_ARM_ERRATA_460075
-	teq	r6, #0x20			@ only present in r2p0
-	mrceq	p15, 1, r0, c9, c0, 2		@ read L2 cache aux ctrl register
-	tsteq	r0, #1 << 22
-	orreq	r0, r0, #(1 << 22)		@ set the Write Allocate disable bit
-	mcreq	p15, 1, r0, c9, c0, 2		@ write the L2 cache aux ctrl register
-#endif
 	b	__errata_finish
 
 __ca9_errata:
-- 
2.7.4

Re: [PATCH] Remove ARM errata Workarounds 458693 and 460075

[Russell King - ARM Linux]

On Sat, Apr 15, 2017 at 07:06:06PM -0500, Nisal Menuka wrote:
> According to ARM, these errata exist only in a version of Cortex-A8
> (r2p0) which was never built. Therefore, I believe there are no platforms
> where this workaround should be enabled.
> link :http://infocenter.arm.com/help/index.jsp?topic=
> /com.arm.doc.faqs/ka15634.html

These were submitted by ARM Ltd back in 2009 - if the silicon was never
built, there would've been no reason to submit them.  Maybe Catalin can
shed some light on this, being the commit author who introduced these?

> Signed-off-by: Nisal Menuka <nisalmenuka23@gmail.com>
> ---
>  arch/arm/Kconfig      | 27 ---------------------------
>  arch/arm/mm/proc-v7.S | 14 --------------
>  2 files changed, 41 deletions(-)
> 
> diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
> index 0d4e71b..d527963 100644
> --- a/arch/arm/Kconfig
> +++ b/arch/arm/Kconfig
> @@ -1005,33 +1005,6 @@ config ARM_ERRATA_430973
>  	  Note that setting specific bits in the ACTLR register may not be
>  	  available in non-secure mode.
>  
> -config ARM_ERRATA_458693
> -	bool "ARM errata: Processor deadlock when a false hazard is created"
> -	depends on CPU_V7
> -	depends on !ARCH_MULTIPLATFORM
> -	help
> -	  This option enables the workaround for the 458693 Cortex-A8 (r2p0)
> -	  erratum. For very specific sequences of memory operations, it is
> -	  possible for a hazard condition intended for a cache line to instead
> -	  be incorrectly associated with a different cache line. This false
> -	  hazard might then cause a processor deadlock. The workaround enables
> -	  the L1 caching of the NEON accesses and disables the PLD instruction
> -	  in the ACTLR register. Note that setting specific bits in the ACTLR
> -	  register may not be available in non-secure mode.
> -
> -config ARM_ERRATA_460075
> -	bool "ARM errata: Data written to the L2 cache can be overwritten with stale data"
> -	depends on CPU_V7
> -	depends on !ARCH_MULTIPLATFORM
> -	help
> -	  This option enables the workaround for the 460075 Cortex-A8 (r2p0)
> -	  erratum. Any asynchronous access to the L2 cache may encounter a
> -	  situation in which recent store transactions to the L2 cache are lost
> -	  and overwritten with stale memory contents from external memory. The
> -	  workaround disables the write-allocate mode for the L2 cache via the
> -	  ACTLR register. Note that setting specific bits in the ACTLR register
> -	  may not be available in non-secure mode.
> -
>  config ARM_ERRATA_742230
>  	bool "ARM errata: DMB operation may be faulty"
>  	depends on CPU_V7 && SMP
> diff --git a/arch/arm/mm/proc-v7.S b/arch/arm/mm/proc-v7.S
> index d00d52c..43a4a12 100644
> --- a/arch/arm/mm/proc-v7.S
> +++ b/arch/arm/mm/proc-v7.S
> @@ -306,20 +306,6 @@ __ca8_errata:
>  	orreq	r0, r0, #(1 << 6)		@ set IBE to 1
>  	mcreq	p15, 0, r0, c1, c0, 1		@ write aux control register
>  #endif
> -#ifdef CONFIG_ARM_ERRATA_458693
> -	teq	r6, #0x20			@ only present in r2p0
> -	mrceq	p15, 0, r0, c1, c0, 1		@ read aux control register
> -	orreq	r0, r0, #(1 << 5)		@ set L1NEON to 1
> -	orreq	r0, r0, #(1 << 9)		@ set PLDNOP to 1
> -	mcreq	p15, 0, r0, c1, c0, 1		@ write aux control register
> -#endif
> -#ifdef CONFIG_ARM_ERRATA_460075
> -	teq	r6, #0x20			@ only present in r2p0
> -	mrceq	p15, 1, r0, c9, c0, 2		@ read L2 cache aux ctrl register
> -	tsteq	r0, #1 << 22
> -	orreq	r0, r0, #(1 << 22)		@ set the Write Allocate disable bit
> -	mcreq	p15, 1, r0, c9, c0, 2		@ write the L2 cache aux ctrl register
> -#endif
>  	b	__errata_finish
>  
>  __ca9_errata:
> -- 
> 2.7.4
> 

-- 
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line: currently at 9.6Mbps down 400kbps up
according to speedtest.net.

Re: [PATCH] Remove ARM errata Workarounds 458693 and 460075

[Catalin Marinas]

On Sun, Apr 16, 2017 at 09:04:46AM +0100, Russell King - ARM Linux wrote:
> On Sat, Apr 15, 2017 at 07:06:06PM -0500, Nisal Menuka wrote:
> > According to ARM, these errata exist only in a version of Cortex-A8
> > (r2p0) which was never built. Therefore, I believe there are no platforms
> > where this workaround should be enabled.
> > link :http://infocenter.arm.com/help/index.jsp?topic=
> > /com.arm.doc.faqs/ka15634.html
> 
> These were submitted by ARM Ltd back in 2009 - if the silicon was never
> built, there would've been no reason to submit them.  Maybe Catalin can
> shed some light on this, being the commit author who introduced these?

We normally try not to submit errata workarounds for revisions that are
not going to be built/deployed. It's possible that at the time there
were plans for r2p0 to be licensed and built (not just FPGA) but I don't
really remember the details. The A8 errata document indeed states that
r1p0 and r2p0 are obsolete but this can mean many things (like not
available to license).

I'll try to see if any of the A8 past product managers know anything
about this. In the meantime, I would leave them in (no run-time
overhead).

-- 
Catalin

Re: [PATCH] Remove ARM errata Workarounds 458693 and 460075

[Robin Murphy]

On 18/04/17 16:57, Catalin Marinas wrote:
> On Sun, Apr 16, 2017 at 09:04:46AM +0100, Russell King - ARM Linux wrote:
>> On Sat, Apr 15, 2017 at 07:06:06PM -0500, Nisal Menuka wrote:
>>> According to ARM, these errata exist only in a version of Cortex-A8
>>> (r2p0) which was never built. Therefore, I believe there are no platforms
>>> where this workaround should be enabled.
>>> link :http://infocenter.arm.com/help/index.jsp?topic=
>>> /com.arm.doc.faqs/ka15634.html
>>
>> These were submitted by ARM Ltd back in 2009 - if the silicon was never
>> built, there would've been no reason to submit them.  Maybe Catalin can
>> shed some light on this, being the commit author who introduced these?
> 
> We normally try not to submit errata workarounds for revisions that are
> not going to be built/deployed. It's possible that at the time there
> were plans for r2p0 to be licensed and built (not just FPGA) but I don't
> really remember the details. The A8 errata document indeed states that
> r1p0 and r2p0 are obsolete but this can mean many things (like not
> available to license).
> 
> I'll try to see if any of the A8 past product managers know anything
> about this. In the meantime, I would leave them in (no run-time
> overhead).

FWIW, I just fired up a RealView PB-A8 board to check, and that reports
r1p1. True, it's not strictly a real silicon implementation (I think
it's one of the structured ASIC test chips), but since it was, as far as
I'm aware, a commercially-available development system, it's not
impossible that someone may still own and use one of these beasts.

Robin.

Re: [PATCH] Remove ARM errata Workarounds 458693 and 460075

[Catalin Marinas]

On Tue, May 02, 2017 at 01:27:54PM +0100, Robin Murphy wrote:
> On 18/04/17 16:57, Catalin Marinas wrote:
> > On Sun, Apr 16, 2017 at 09:04:46AM +0100, Russell King - ARM Linux wrote:
> >> On Sat, Apr 15, 2017 at 07:06:06PM -0500, Nisal Menuka wrote:
> >>> According to ARM, these errata exist only in a version of Cortex-A8
> >>> (r2p0) which was never built. Therefore, I believe there are no platforms
> >>> where this workaround should be enabled.
> >>> link :http://infocenter.arm.com/help/index.jsp?topic=
> >>> /com.arm.doc.faqs/ka15634.html
> >>
> >> These were submitted by ARM Ltd back in 2009 - if the silicon was never
> >> built, there would've been no reason to submit them.  Maybe Catalin can
> >> shed some light on this, being the commit author who introduced these?
> > 
> > We normally try not to submit errata workarounds for revisions that are
> > not going to be built/deployed. It's possible that at the time there
> > were plans for r2p0 to be licensed and built (not just FPGA) but I don't
> > really remember the details. The A8 errata document indeed states that
> > r1p0 and r2p0 are obsolete but this can mean many things (like not
> > available to license).
> > 
> > I'll try to see if any of the A8 past product managers know anything
> > about this. In the meantime, I would leave them in (no run-time
> > overhead).
> 
> FWIW, I just fired up a RealView PB-A8 board to check, and that reports
> r1p1. True, it's not strictly a real silicon implementation (I think
> it's one of the structured ASIC test chips), but since it was, as far as
> I'm aware, a commercially-available development system, it's not
> impossible that someone may still own and use one of these beasts.

The above errata were specific to r1p0 (not r1p1) and r2p0. Since ARM
Ltd claims there are no products built around these revisions, I'm fine
with removing the workarounds from the kernel (internal testchips/FPGA
don't count as products but I doubt they are still relevant 8 years
later).

-- 
Catalin