[PATCH] [2/9] Add a kernel_address() that works for data too

[PATCH] [2/9] Add a kernel_address() that works for data too

[Andi Kleen]

Add a variant of kernel_text_address() that includes kernel data.

Assumes kernel is _text ... _end - init section. True everywhere?

Cc: linux-arch@vger.kernel.org
Signed-off-by: Andi Kleen <ak@linux.intel.com>

---
 include/linux/kernel.h |    1 +
 kernel/extable.c       |   18 ++++++++++++++++++
 2 files changed, 19 insertions(+)

Index: linux-2.6.33-rc2-ak/include/linux/kernel.h
===================================================================
--- linux-2.6.33-rc2-ak.orig/include/linux/kernel.h
+++ linux-2.6.33-rc2-ak/include/linux/kernel.h
@@ -205,6 +205,7 @@ extern unsigned long long memparse(const
 extern int core_kernel_text(unsigned long addr);
 extern int __kernel_text_address(unsigned long addr);
 extern int kernel_text_address(unsigned long addr);
+extern int kernel_address(unsigned long addr);
 extern int func_ptr_is_kernel_text(void *ptr);
 
 struct pid;
Index: linux-2.6.33-rc2-ak/kernel/extable.c
===================================================================
--- linux-2.6.33-rc2-ak.orig/kernel/extable.c
+++ linux-2.6.33-rc2-ak/kernel/extable.c
@@ -72,6 +72,18 @@ int core_kernel_text(unsigned long addr)
 	return 0;
 }
 
+static int core_kernel_address(unsigned long addr)
+{
+	if ((addr >= (unsigned long)_text &&
+	     addr <= (unsigned long)_end)) {
+		if (addr >= (unsigned long)__init_begin &&
+		    addr < (unsigned long)__init_end)
+			return system_state == SYSTEM_BOOTING;
+		return 1;
+	}
+	return 0;
+}
+
 int __kernel_text_address(unsigned long addr)
 {
 	if (core_kernel_text(addr))
@@ -98,6 +110,12 @@ int kernel_text_address(unsigned long ad
 	return is_module_text_address(addr);
 }
 
+/* text or data in core kernel or module */
+int kernel_address(unsigned long addr)
+{
+	return core_kernel_address(addr) || is_module_address(addr);
+}
+
 /*
  * On some architectures (PPC64, IA64) function pointers
  * are actually only tokens to some data that then holds the

Re: [PATCH] [2/9] Add a kernel_address() that works for data too

[Russell King]

On Tue, Jan 05, 2010 at 03:15:27AM +0100, Andi Kleen wrote:
> 
> Add a variant of kernel_text_address() that includes kernel data.
> 
> Assumes kernel is _text ... _end - init section. True everywhere?

No.  XIP kernels have the text and data/bss separated into two distinct
address regions.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:

Re: [PATCH] [2/9] Add a kernel_address() that works for data too

[Sam Ravnborg]

On Tue, Jan 05, 2010 at 03:15:27AM +0100, Andi Kleen wrote:
> 
> Add a variant of kernel_text_address() that includes kernel data.
> 
> Assumes kernel is _text ... _end - init section. True everywhere?

Tim Abbott has done a great job lately to unify the various
linker scripts used by the different architectures.

Architectures are supposed to follow the skeleton outlined
in include/asm-generic/vmlinux.lds.h

But I think the skeleton needs a small update.
It describes that we mark start of .text with _stext.
But reality is that we use _text for this.

But you can trust _etext an almost all architectures.
It is a bug if it is missing.

So [_text, _etext] is the text section.

The data section may be placed before or after - it depends on the architecture.
But again - only some architectures define _sdata.
But all? define _edata.


	Sam


> 
> Cc: linux-arch@vger.kernel.org
> Signed-off-by: Andi Kleen <ak@linux.intel.com>
> 
> ---
>  include/linux/kernel.h |    1 +
>  kernel/extable.c       |   18 ++++++++++++++++++
>  2 files changed, 19 insertions(+)
> 
> Index: linux-2.6.33-rc2-ak/include/linux/kernel.h
> ===================================================================
> --- linux-2.6.33-rc2-ak.orig/include/linux/kernel.h
> +++ linux-2.6.33-rc2-ak/include/linux/kernel.h
> @@ -205,6 +205,7 @@ extern unsigned long long memparse(const
>  extern int core_kernel_text(unsigned long addr);
>  extern int __kernel_text_address(unsigned long addr);
>  extern int kernel_text_address(unsigned long addr);
> +extern int kernel_address(unsigned long addr);
>  extern int func_ptr_is_kernel_text(void *ptr);
>  
>  struct pid;
> Index: linux-2.6.33-rc2-ak/kernel/extable.c
> ===================================================================
> --- linux-2.6.33-rc2-ak.orig/kernel/extable.c
> +++ linux-2.6.33-rc2-ak/kernel/extable.c
> @@ -72,6 +72,18 @@ int core_kernel_text(unsigned long addr)
>  	return 0;
>  }
>  
> +static int core_kernel_address(unsigned long addr)
> +{
> +	if ((addr >= (unsigned long)_text &&
> +	     addr <= (unsigned long)_end)) {
> +		if (addr >= (unsigned long)__init_begin &&
> +		    addr < (unsigned long)__init_end)
> +			return system_state == SYSTEM_BOOTING;
> +		return 1;
> +	}
> +	return 0;
> +}
> +
>  int __kernel_text_address(unsigned long addr)
>  {
>  	if (core_kernel_text(addr))
> @@ -98,6 +110,12 @@ int kernel_text_address(unsigned long ad
>  	return is_module_text_address(addr);
>  }
>  
> +/* text or data in core kernel or module */
> +int kernel_address(unsigned long addr)
> +{
> +	return core_kernel_address(addr) || is_module_address(addr);
> +}
> +
>  /*
>   * On some architectures (PPC64, IA64) function pointers
>   * are actually only tokens to some data that then holds the
> --
> To unsubscribe from this list: send the line "unsubscribe linux-arch" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

Re: [PATCH] [2/9] Add a kernel_address() that works for data too

[Russell King]

On Tue, Jan 05, 2010 at 09:58:53AM +0100, Sam Ravnborg wrote:
> But you can trust _etext an almost all architectures.
> It is a bug if it is missing.
> 
> So [_text, _etext] is the text section.
> 
> The data section may be placed before or after - it depends on the architecture.
> But again - only some architectures define _sdata.
> But all? define _edata.

You can not guarantee that the data segment is after the text segment,
unless you want to outlaw XIP kernels.  XIP kernels have the text
segment mapped at a completely different address to the data segment.

I'd suggest the only way to identify the data segment in a generic way
is to have everyone define _sdata, or more preferably _data (to be
consistent with _text), to be the start of the data segment.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:

Re: [PATCH] [2/9] Add a kernel_address() that works for data too

[Andi Kleen]

> I'd suggest the only way to identify the data segment in a generic way
> is to have everyone define _sdata, or more preferably _data (to be
> consistent with _text), to be the start of the data segment.

I agree. I'll take a look at that.

-Andi
-- 
ak@linux.intel.com -- Speaking for myself only.

Re: [PATCH] [2/9] Add a kernel_address() that works for data too

[Andi Kleen]

> I'd suggest the only way to identify the data segment in a generic way
> is to have everyone define _sdata, or more preferably _data (to be
> consistent with _text), to be the start of the data segment.

I agree. I'll take a look at that.

-Andi
-- 
ak@linux.intel.com -- Speaking for myself only.

Re: [PATCH] [2/9] Add a kernel_address() that works for data too

[Andrew Morton]

On Tue, 5 Jan 2010 20:15:03 +0100
Andi Kleen <andi@firstfloor.org> wrote:

> > I'd suggest the only way to identify the data segment in a generic way
> > is to have everyone define _sdata, or more preferably _data (to be
> > consistent with _text), to be the start of the data segment.
> 
> I agree. I'll take a look at that.
> 

I'll merge this as-is for now.  Please send updates when convenient.