[RFC PATCH 05/12] arm: mm: retry on QSD icache parity errors

[RFC PATCH 05/12] arm: mm: retry on QSD icache parity errors

[Daniel Walker]

From: Steve Muckle <smuckle@quicinc.com>

Parity errors in the icache on QSD can be worked around either by
retrying the access, or invalidating the icache. The whole icache
must be invalidated since the data abort is imprecise (the faulting
address is not known).

Signed-off-by: Steve Muckle <smuckle@quicinc.com>
Signed-off-by: Daniel Walker <dwalker@codeaurora.org>
---
 arch/arm/mm/fault.c |   36 +++++++++++++++++++++++++++++++++++-
 1 files changed, 35 insertions(+), 1 deletions(-)

diff --git a/arch/arm/mm/fault.c b/arch/arm/mm/fault.c
index 10e0680..bea3e75 100644
--- a/arch/arm/mm/fault.c
+++ b/arch/arm/mm/fault.c
@@ -2,6 +2,7 @@
  *  linux/arch/arm/mm/fault.c
  *
  *  Copyright (C) 1995  Linus Torvalds
+ *  Copyright (c) 2009, Code Aurora Forum. All rights reserved.
  *  Modifications for ARM processor (c) 1995-2004 Russell King
  *
  * This program is free software; you can redistribute it and/or modify
@@ -442,6 +443,39 @@ do_bad(unsigned long addr, unsigned int fsr, struct pt_regs *regs)
 	return 1;
 }
 
+static int
+do_imprecise_ext(unsigned long addr, unsigned int fsr, struct pt_regs *regs)
+{
+#ifdef CONFIG_ARCH_MSM_SCORPION
+	unsigned int regval;
+	static unsigned char flush_toggle;
+
+	asm("mrc p15, 0, %0, c5, c1, 0\n" /* read adfsr for fault status */
+	    : "=r" (regval));
+	if (regval == 0x2) {
+		/* Fault was caused by icache parity error. Alternate
+		 * simply retrying the access and flushing the icache. */
+		flush_toggle ^= 1;
+		if (flush_toggle)
+			asm("mcr p15, 0, %0, c7, c5, 0\n"
+			    :
+			    : "r" (regval)); /* input value is ignored */
+		/* Clear fault in EFSR. */
+		asm("mcr p15, 7, %0, c15, c0, 1\n"
+		    :
+		    : "r" (regval));
+		/* Clear fault in ADFSR. */
+		regval = 0;
+		asm("mcr p15, 0, %0, c5, c1, 0\n"
+		    :
+		    : "r" (regval));
+		return 0;
+	}
+#endif
+
+	return 1;
+}
+
 static struct fsr_info {
 	int	(*fn)(unsigned long addr, unsigned int fsr, struct pt_regs *regs);
 	int	sig;
@@ -479,7 +513,7 @@ static struct fsr_info {
 	{ do_bad,		SIGBUS,  0,		"unknown 19"			   },
 	{ do_bad,		SIGBUS,  0,		"lock abort"			   }, /* xscale */
 	{ do_bad,		SIGBUS,  0,		"unknown 21"			   },
-	{ do_bad,		SIGBUS,  BUS_OBJERR,	"imprecise external abort"	   }, /* xscale */
+	{ do_imprecise_ext,	SIGBUS,  BUS_OBJERR,	"imprecise external abort"	   }, /* xscale */
 	{ do_bad,		SIGBUS,  0,		"unknown 23"			   },
 	{ do_bad,		SIGBUS,  0,		"dcache parity error"		   }, /* xscale */
 	{ do_bad,		SIGBUS,  0,		"unknown 25"			   },
-- 
1.6.3.3

[RFC PATCH 05/12] arm: mm: retry on QSD icache parity errors

[Pavel Machek]

Hi!

> Parity errors in the icache on QSD can be worked around either by
> retrying the access, or invalidating the icache. The whole icache
> must be invalidated since the data abort is imprecise (the faulting
> address is not known).
> 
> Signed-off-by: Steve Muckle <smuckle@quicinc.com>
> Signed-off-by: Daniel Walker <dwalker@codeaurora.org>


> @@ -442,6 +443,39 @@ do_bad(unsigned long addr, unsigned int fsr, struct pt_regs *regs)
>  	return 1;
>  }
>  
> +static int
> +do_imprecise_ext(unsigned long addr, unsigned int fsr, struct pt_regs *regs)
> +{
> +#ifdef CONFIG_ARCH_MSM_SCORPION
> +	unsigned int regval;
> +	static unsigned char flush_toggle;
> +
> +	asm("mrc p15, 0, %0, c5, c1, 0\n" /* read adfsr for fault status */
> +	    : "=r" (regval));
> +	if (regval == 0x2) {
> +		/* Fault was caused by icache parity error. Alternate
> +		 * simply retrying the access and flushing the icache. */
> +		flush_toggle ^= 1;
> +		if (flush_toggle)
> +			asm("mcr p15, 0, %0, c7, c5, 0\n"
> +			    :
> +			    : "r" (regval)); /* input value is ignored */
> +		/* Clear fault in EFSR. */
> +		asm("mcr p15, 7, %0, c15, c0, 1\n"
> +		    :
> +		    : "r" (regval));
> +		/* Clear fault in ADFSR. */
> +		regval = 0;
> +		asm("mcr p15, 0, %0, c5, c1, 0\n"
> +		    :
> +		    : "r" (regval));
> +		return 0;
> +	}
> +#endif

Tell the user or at least keep the count, so cache going bad can be
recognized?

Also you probably need asm volatile, and probably do not need those
\n-s.

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[RFC PATCH 05/12] arm: mm: retry on QSD icache parity errors

[Catalin Marinas]

On Thu, 2010-01-28 at 22:59 +0000, Daniel Walker wrote:
> diff --git a/arch/arm/mm/fault.c b/arch/arm/mm/fault.c
> index 10e0680..bea3e75 100644
> --- a/arch/arm/mm/fault.c
> +++ b/arch/arm/mm/fault.c
[...]
>  static struct fsr_info {
>         int     (*fn)(unsigned long addr, unsigned int fsr, struct pt_regs *regs);
>         int     sig;
> @@ -479,7 +513,7 @@ static struct fsr_info {
>         { do_bad,               SIGBUS,  0,             "unknown 19"                       },
>         { do_bad,               SIGBUS,  0,             "lock abort"                       }, /* xscale */
>         { do_bad,               SIGBUS,  0,             "unknown 21"                       },
> -       { do_bad,               SIGBUS,  BUS_OBJERR,    "imprecise external abort"         }, /* xscale */
> +       { do_imprecise_ext,     SIGBUS,  BUS_OBJERR,    "imprecise external abort"         }, /* xscale */

Can you not use hook_fault_code() and move the do_imprecise_ext() to a
platform-specific file (under arch/arm/mach-* maybe)?

-- 
Catalin

[RFC PATCH 05/12] arm: mm: retry on QSD icache parity errors

[Daniel Walker]

On Fri, 2010-01-29 at 11:11 +0000, Catalin Marinas wrote:
> On Thu, 2010-01-28 at 22:59 +0000, Daniel Walker wrote:
> > diff --git a/arch/arm/mm/fault.c b/arch/arm/mm/fault.c
> > index 10e0680..bea3e75 100644
> > --- a/arch/arm/mm/fault.c
> > +++ b/arch/arm/mm/fault.c
> [...]
> >  static struct fsr_info {
> >         int     (*fn)(unsigned long addr, unsigned int fsr, struct pt_regs *regs);
> >         int     sig;
> > @@ -479,7 +513,7 @@ static struct fsr_info {
> >         { do_bad,               SIGBUS,  0,             "unknown 19"                       },
> >         { do_bad,               SIGBUS,  0,             "lock abort"                       }, /* xscale */
> >         { do_bad,               SIGBUS,  0,             "unknown 21"                       },
> > -       { do_bad,               SIGBUS,  BUS_OBJERR,    "imprecise external abort"         }, /* xscale */
> > +       { do_imprecise_ext,     SIGBUS,  BUS_OBJERR,    "imprecise external abort"         }, /* xscale */
> 
> Can you not use hook_fault_code() and move the do_imprecise_ext() to a
> platform-specific file (under arch/arm/mach-* maybe)?
> 

How do you mean? I'm not sure how we would move this into
arch/arm/mach- ..

Daniel

[RFC PATCH 05/12] arm: mm: retry on QSD icache parity errors

[Russell King - ARM Linux]

On Tue, Feb 23, 2010 at 10:36:08AM -0800, Daniel Walker wrote:
> On Fri, 2010-01-29 at 11:11 +0000, Catalin Marinas wrote:
> > On Thu, 2010-01-28 at 22:59 +0000, Daniel Walker wrote:
> > > diff --git a/arch/arm/mm/fault.c b/arch/arm/mm/fault.c
> > > index 10e0680..bea3e75 100644
> > > --- a/arch/arm/mm/fault.c
> > > +++ b/arch/arm/mm/fault.c
> > [...]
> > >  static struct fsr_info {
> > >         int     (*fn)(unsigned long addr, unsigned int fsr, struct pt_regs *regs);
> > >         int     sig;
> > > @@ -479,7 +513,7 @@ static struct fsr_info {
> > >         { do_bad,               SIGBUS,  0,             "unknown 19"                       },
> > >         { do_bad,               SIGBUS,  0,             "lock abort"                       }, /* xscale */
> > >         { do_bad,               SIGBUS,  0,             "unknown 21"                       },
> > > -       { do_bad,               SIGBUS,  BUS_OBJERR,    "imprecise external abort"         }, /* xscale */
> > > +       { do_imprecise_ext,     SIGBUS,  BUS_OBJERR,    "imprecise external abort"         }, /* xscale */
> > 
> > Can you not use hook_fault_code() and move the do_imprecise_ext() to a
> > platform-specific file (under arch/arm/mach-* maybe)?
> > 
> 
> How do you mean? I'm not sure how we would move this into
> arch/arm/mach- ..

Err, exactly as Catalin says.  Use hook_fault_code() to hook into the
fault table at run time, just like platforms which have to handle
imprecise exceptions on PCI and the alignment fault handler do.

[RFC PATCH 05/12] arm: mm: retry on QSD icache parity errors

[Daniel Walker]

On Tue, 2010-02-23 at 19:02 +0000, Russell King - ARM Linux wrote:

> > > Can you not use hook_fault_code() and move the do_imprecise_ext() to a
> > > platform-specific file (under arch/arm/mach-* maybe)?
> > > 
> > 
> > How do you mean? I'm not sure how we would move this into
> > arch/arm/mach- ..
> 
> Err, exactly as Catalin says.  Use hook_fault_code() to hook into the
> fault table at run time, just like platforms which have to handle
> imprecise exceptions on PCI and the alignment fault handler do.


"Can you not use hook_fault_code()"

Typo maybe ?

Daniel

[RFC PATCH 05/12] arm: mm: retry on QSD icache parity errors

[Russell King - ARM Linux]

On Tue, Feb 23, 2010 at 11:06:58AM -0800, Daniel Walker wrote:
> On Tue, 2010-02-23 at 19:02 +0000, Russell King - ARM Linux wrote:
> 
> > > > Can you not use hook_fault_code() and move the do_imprecise_ext() to a
> > > > platform-specific file (under arch/arm/mach-* maybe)?
> > > > 
> > > 
> > > How do you mean? I'm not sure how we would move this into
> > > arch/arm/mach- ..
> > 
> > Err, exactly as Catalin says.  Use hook_fault_code() to hook into the
> > fault table at run time, just like platforms which have to handle
> > imprecise exceptions on PCI and the alignment fault handler do.
> 
> 
> "Can you not use hook_fault_code()"
> 
> Typo maybe ?

I don't think so.  It's a request for you to use hook_fault_code().