diff mbox series

arm64: set "rodata=on" as default

Message ID 20241021053948.5848-1-shijie@os.amperecomputing.com (mailing list archive)
State New, archived
Headers show
Series arm64: set "rodata=on" as default | expand

Commit Message

Huang Shijie Oct. 21, 2024, 5:39 a.m. UTC
From Documentation/admin-guide/kernel-parameters.txt, we know that:
 rodata=	[KNL,EARLY]
	on	Mark read-only kernel memory as read-only (default).
	off	Leave read-only kernel memory writable for debugging.
	full	Mark read-only kernel memory and aliases as read-only
		[arm64]

So the "rodata=on" is the default.

But the current code does not follow the document, it makes "rodata=full"
as the default.

After patch
  commit acfa60dbe038 ("arm64: mm: Fix "rodata=on" when CONFIG_RODATA_FULL_DEFAULT_ENABLED=y")
the "rodata=on" can works fine now.

The "rodata=on" can provide us more block mappings and contiguous hits
to map the linear region which minimize the TLB footprint. And the
linear aliases of pages belonging to read-only mappings in vmalloc
region are also marked as read-only now.

This patch disables RODATA_FULL_DEFAULT_ENABLED by default,
so the default value:
     rodata_full=false, rodata_enabled=true
then the default behavior follows the "rodata=on".
And we can get better performance with the "rodata=on" as the default
too.

Signed-off-by: Huang Shijie <shijie@os.amperecomputing.com>
---
 arch/arm64/Kconfig | 1 -
 1 file changed, 1 deletion(-)

Comments

Will Deacon Oct. 23, 2024, 4:35 p.m. UTC | #1
On Mon, Oct 21, 2024 at 01:39:48PM +0800, Huang Shijie wrote:
> From Documentation/admin-guide/kernel-parameters.txt, we know that:
>  rodata=	[KNL,EARLY]
> 	on	Mark read-only kernel memory as read-only (default).
> 	off	Leave read-only kernel memory writable for debugging.
> 	full	Mark read-only kernel memory and aliases as read-only
> 		[arm64]
> 
> So the "rodata=on" is the default.
> 
> But the current code does not follow the document, it makes "rodata=full"
> as the default.
> 
> After patch
>   commit acfa60dbe038 ("arm64: mm: Fix "rodata=on" when CONFIG_RODATA_FULL_DEFAULT_ENABLED=y")
> the "rodata=on" can works fine now.
> 
> The "rodata=on" can provide us more block mappings and contiguous hits
> to map the linear region which minimize the TLB footprint. And the
> linear aliases of pages belonging to read-only mappings in vmalloc
> region are also marked as read-only now.
> 
> This patch disables RODATA_FULL_DEFAULT_ENABLED by default,
> so the default value:
>      rodata_full=false, rodata_enabled=true
> then the default behavior follows the "rodata=on".
> And we can get better performance with the "rodata=on" as the default
> too.
> 
> Signed-off-by: Huang Shijie <shijie@os.amperecomputing.com>
> ---
>  arch/arm64/Kconfig | 1 -
>  1 file changed, 1 deletion(-)
> 
> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> index fd9df6dcc593..6f30f749156e 100644
> --- a/arch/arm64/Kconfig
> +++ b/arch/arm64/Kconfig
> @@ -1649,7 +1649,6 @@ config MITIGATE_SPECTRE_BRANCH_HISTORY
>  
>  config RODATA_FULL_DEFAULT_ENABLED
>  	bool "Apply r/o permissions of VM areas also to their linear aliases"
> -	default y
>  	help
>  	  Apply read-only attributes of VM areas to the linear alias of
>  	  the backing pages as well. This prevents code or read-only data

I think this is the wrong approach.

The "full" behaviour on arm64 is equivalent to the "on" behaviour on x86
so the more consistent change to make would be:

  - Make our "on" behaviour be what is currently done by "full"
  - Remove RODATA_FULL_DEFAULT_ENABLED
  - Introduce a new option (e.g. "rodata=noalias") which would match the
    current "on" behaviour
  - Update (simplify) the documentation

That way, the default behaviour and the "on"/"off" options follow the
x86 behaviour and expert users can fine-tune the security/performance
trade-off using the "noalias" option.

Will
Shijie Huang Oct. 24, 2024, 3:57 a.m. UTC | #2
On 2024/10/24 0:35, Will Deacon wrote:
> The "full" behaviour on arm64 is equivalent to the "on" behaviour on x86
> so the more consistent change to make would be:
>
>    - Make our "on" behaviour be what is currently done by "full"
>    - Remove RODATA_FULL_DEFAULT_ENABLED
>    - Introduce a new option (e.g. "rodata=noalias") which would match the
>      current "on" behaviour
>    - Update (simplify) the documentation
>
> That way, the default behaviour and the "on"/"off" options follow the
> x86 behaviour and expert users can fine-tune the security/performance
> trade-off using the "noalias" option.

Okay. I will do it in the next version.


Thanks

Huang Shijie
diff mbox series

Patch

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index fd9df6dcc593..6f30f749156e 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -1649,7 +1649,6 @@  config MITIGATE_SPECTRE_BRANCH_HISTORY
 
 config RODATA_FULL_DEFAULT_ENABLED
 	bool "Apply r/o permissions of VM areas also to their linear aliases"
-	default y
 	help
 	  Apply read-only attributes of VM areas to the linear alias of
 	  the backing pages as well. This prevents code or read-only data