On 3/1/24 08:44, Gerd Hoffmann wrote:
> When running in SEV mode keep the VC handler installed.
> Add a function to uninstall it later.
> 
> This allows using the cpuid instruction in SetCr3ForPageTables64,
> which is needed to check for la57 & 1G page support.
> 
> Signed-off-by: Gerd Hoffmann <kra...@redhat.com>
> ---
>  OvmfPkg/ResetVector/Ia32/AmdSev.asm       | 12 ++++++++++--
>  OvmfPkg/ResetVector/Ia32/PageTables64.asm |  1 +
>  OvmfPkg/ResetVector/Main.asm              |  4 ++++
>  3 files changed, 15 insertions(+), 2 deletions(-)
> 
> diff --git a/OvmfPkg/ResetVector/Ia32/AmdSev.asm 
> b/OvmfPkg/ResetVector/Ia32/AmdSev.asm
> index 23e4c5ebbe92..cbb86871636f 100644
> --- a/OvmfPkg/ResetVector/Ia32/AmdSev.asm
> +++ b/OvmfPkg/ResetVector/Ia32/AmdSev.asm
> @@ -320,9 +320,9 @@ NoSevEsVcHlt:
>  NoSevPass:
>      xor       eax, eax
>  
> -SevExit:
>      ;
> -    ; Clear exception handlers and stack
> +    ; When NOT running in SEV mode: clear exception handlers and stack here.
> +    ; Otherwise: SevClearVcHandlerAndStack must be called later.
>      ;
>      push      eax
>      mov       eax, ADDR_OF(IdtrClear)
> @@ -330,8 +330,16 @@ SevExit:
>      pop       eax
>      mov       esp, 0
>  
> +SevExit:
>      OneTimeCallRet CheckSevFeatures
>  
> +SevClearVcHandlerAndStack:
> +    ; Clear exception handlers and stack
> +    mov       eax, ADDR_OF(IdtrClear)
> +    lidt      [cs:eax]
> +    mov       esp, 0
> +    OneTimeCallRet SevClearVcHandlerAndStack
> +
>  ; Start of #VC exception handling routines
>  ;
>  
> diff --git a/OvmfPkg/ResetVector/Ia32/PageTables64.asm 
> b/OvmfPkg/ResetVector/Ia32/PageTables64.asm
> index b922c845f297..29ce155eed8d 100644
> --- a/OvmfPkg/ResetVector/Ia32/PageTables64.asm
> +++ b/OvmfPkg/ResetVector/Ia32/PageTables64.asm
> @@ -254,6 +254,7 @@ SevInit:
>      CreatePageTables4Level edx
>      ; Clear the C-bit from the GHCB page if the SEV-ES is enabled.
>      OneTimeCall   SevClearPageEncMaskForGhcbPage
> +    OneTimeCall   SevClearVcHandlerAndStack
>      jmp SetCr3
>  
>  TdxBspInit:
> diff --git a/OvmfPkg/ResetVector/Main.asm b/OvmfPkg/ResetVector/Main.asm
> index 46cfa87c4c0a..88b25db3bc9e 100644
> --- a/OvmfPkg/ResetVector/Main.asm
> +++ b/OvmfPkg/ResetVector/Main.asm
> @@ -80,7 +80,11 @@ SearchBfv:
>      ; Set the OVMF/SEV work area as appropriate.
>      ;
>      OneTimeCall CheckSevFeatures
> +    cmp         byte[WORK_AREA_GUEST_TYPE], 1
> +    jnz         NoSevIa32
> +    OneTimeCall SevClearVcHandlerAndStack
>  
> +NoSevIa32:
>      ;
>      ; Restore initial EAX value into the EAX register
>      ;

Did you miss Tom's review under v1?

https://edk2.groups.io/g/devel/message/116176

The patch is identical to its v1 counterpart, which should not be a
problem in itself (Tom mentioned a small, *optional*, simplification,
IIUC); however, I don't understand why you didn't pick up Tom's R-b.

I'm ready to merge this (adding Tom's R-b, if you, Gerd, confirm that
that's what you want).

Having deferred to Tom's judgement on this:

Acked-by: Laszlo Ersek <ler...@redhat.com>



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#116250): https://edk2.groups.io/g/devel/message/116250
Mute This Topic: https://groups.io/mt/104660115/21656
Group Owner: devel+ow...@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-


Reply via email to