[PATCH] Documentation: Add powerpc options for spec_store_bypass_disable

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

[PATCH] Documentation: Add powerpc options for spec_store_bypass_disable

Michael Ellerman-2
Document the support for spec_store_bypass_disable that was added for
powerpc in commit a048a07d7f45 ("powerpc/64s: Add support for a store
forwarding barrier at kernel entry/exit").

Signed-off-by: Michael Ellerman <[hidden email]>
---
 Documentation/admin-guide/kernel-parameters.txt | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

I tried documenting the differences between the PPC options and X86 ones in one
section, but it got quite messy, so I went with this instead. Happy to take
advice on how better to structure it if anyone has opinions.

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index efc7aa7a0670..f320c7168b04 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -4060,6 +4060,8 @@
  This parameter controls whether the Speculative Store
  Bypass optimization is used.
 
+ On x86 the options are:
+
  on      - Unconditionally disable Speculative Store Bypass
  off     - Unconditionally enable Speculative Store Bypass
  auto    - Kernel detects whether the CPU model contains an
@@ -4075,12 +4077,20 @@
  seccomp - Same as "prctl" above, but all seccomp threads
   will disable SSB unless they explicitly opt out.
 
- Not specifying this option is equivalent to
- spec_store_bypass_disable=auto.
-
  Default mitigations:
  X86: If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
 
+ On powerpc the options are:
+
+ on,auto - On Power8 and Power9 insert a store-forwarding
+  barrier on kernel entry and exit. On Power7
+  perform a software flush on kernel entry and
+  exit.
+ off - No action.
+
+ Not specifying this option is equivalent to
+ spec_store_bypass_disable=auto.
+
  spia_io_base= [HW,MTD]
  spia_fio_base=
  spia_pedr=
--
2.14.1

Reply | Threaded
Open this post in threaded view
|

Re: [PATCH] Documentation: Add powerpc options for spec_store_bypass_disable

Kees Cook
On Mon, Jul 9, 2018 at 7:08 PM, Michael Ellerman <[hidden email]> wrote:
> Document the support for spec_store_bypass_disable that was added for
> powerpc in commit a048a07d7f45 ("powerpc/64s: Add support for a store
> forwarding barrier at kernel entry/exit").
>
> Signed-off-by: Michael Ellerman <[hidden email]>

Reviewed-by: Kees Cook <[hidden email]>

-Kees

> ---
>  Documentation/admin-guide/kernel-parameters.txt | 16 +++++++++++++---
>  1 file changed, 13 insertions(+), 3 deletions(-)
>
> I tried documenting the differences between the PPC options and X86 ones in one
> section, but it got quite messy, so I went with this instead. Happy to take
> advice on how better to structure it if anyone has opinions.
>
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index efc7aa7a0670..f320c7168b04 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -4060,6 +4060,8 @@
>                         This parameter controls whether the Speculative Store
>                         Bypass optimization is used.
>
> +                       On x86 the options are:
> +
>                         on      - Unconditionally disable Speculative Store Bypass
>                         off     - Unconditionally enable Speculative Store Bypass
>                         auto    - Kernel detects whether the CPU model contains an
> @@ -4075,12 +4077,20 @@
>                         seccomp - Same as "prctl" above, but all seccomp threads
>                                   will disable SSB unless they explicitly opt out.
>
> -                       Not specifying this option is equivalent to
> -                       spec_store_bypass_disable=auto.
> -
>                         Default mitigations:
>                         X86:    If CONFIG_SECCOMP=y "seccomp", otherwise "prctl"
>
> +                       On powerpc the options are:
> +
> +                       on,auto - On Power8 and Power9 insert a store-forwarding
> +                                 barrier on kernel entry and exit. On Power7
> +                                 perform a software flush on kernel entry and
> +                                 exit.
> +                       off     - No action.
> +
> +                       Not specifying this option is equivalent to
> +                       spec_store_bypass_disable=auto.
> +
>         spia_io_base=   [HW,MTD]
>         spia_fio_base=
>         spia_pedr=
> --
> 2.14.1
>



--
Kees Cook
Pixel Security
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH] Documentation: Add powerpc options for spec_store_bypass_disable

Thomas Gleixner
In reply to this post by Michael Ellerman-2
On Tue, 10 Jul 2018, Michael Ellerman wrote:

> Document the support for spec_store_bypass_disable that was added for
> powerpc in commit a048a07d7f45 ("powerpc/64s: Add support for a store
> forwarding barrier at kernel entry/exit").
>
> Signed-off-by: Michael Ellerman <[hidden email]>
> ---
>  Documentation/admin-guide/kernel-parameters.txt | 16 +++++++++++++---
>  1 file changed, 13 insertions(+), 3 deletions(-)
>
> I tried documenting the differences between the PPC options and X86 ones in one
> section, but it got quite messy, so I went with this instead. Happy to take
> advice on how better to structure it if anyone has opinions.

Looks good to me.

Acked-by: Thomas Gleixner <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH] Documentation: Add powerpc options for spec_store_bypass_disable

Jonathan Corbet
In reply to this post by Michael Ellerman-2
On Tue, 10 Jul 2018 12:08:36 +1000
Michael Ellerman <[hidden email]> wrote:

> Document the support for spec_store_bypass_disable that was added for
> powerpc in commit a048a07d7f45 ("powerpc/64s: Add support for a store
> forwarding barrier at kernel entry/exit").
>
> Signed-off-by: Michael Ellerman <[hidden email]>

Applied, thanks.

jon