public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: "Laszlo Ersek" <lersek@redhat.com>
To: devel@edk2.groups.io, kraxel@redhat.com
Cc: Tom Lendacky <thomas.lendacky@amd.com>,
	Jiewen Yao <jiewen.yao@intel.com>,
	Oliver Steffen <osteffen@redhat.com>,
	Erdem Aktas <erdemaktas@google.com>,
	Michael Roth <michael.roth@amd.com>,
	Ard Biesheuvel <ardb+tianocore@kernel.org>,
	Min Xu <min.m.xu@intel.com>
Subject: Re: [edk2-devel] [PATCH 06/10] OvmfPkg/ResetVector: add 5-level paging support
Date: Wed, 28 Feb 2024 06:33:34 +0100	[thread overview]
Message-ID: <8163c5aa-47fd-330c-ba50-c24d676a5903@redhat.com> (raw)
In-Reply-To: <20240222115435.85794-7-kraxel@redhat.com>

On 2/22/24 12:54, Gerd Hoffmann wrote:
> Add macros to check for 5-level paging and gigabyte page support.
> Enable 5-level paging for the non-confidential-computing case.
> 
> Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
> ---
>  OvmfPkg/ResetVector/ResetVector.inf       |   1 +
>  OvmfPkg/ResetVector/Ia32/PageTables64.asm | 105 ++++++++++++++++++++++
>  OvmfPkg/ResetVector/ResetVector.nasmb     |   1 +
>  3 files changed, 107 insertions(+)
> 
> diff --git a/OvmfPkg/ResetVector/ResetVector.inf b/OvmfPkg/ResetVector/ResetVector.inf
> index a4154ca90c28..65f71b05a02e 100644
> --- a/OvmfPkg/ResetVector/ResetVector.inf
> +++ b/OvmfPkg/ResetVector/ResetVector.inf
> @@ -64,3 +64,4 @@ [FixedPcd]
>    gUefiOvmfPkgTokenSpaceGuid.PcdQemuHashTableSize
>    gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSnpSecretsBase
>    gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSnpSecretsSize
> +  gEfiMdeModulePkgTokenSpaceGuid.PcdUse5LevelPageTable
> diff --git a/OvmfPkg/ResetVector/Ia32/PageTables64.asm b/OvmfPkg/ResetVector/Ia32/PageTables64.asm
> index 84a7b4efc019..825589f31193 100644
> --- a/OvmfPkg/ResetVector/Ia32/PageTables64.asm
> +++ b/OvmfPkg/ResetVector/Ia32/PageTables64.asm
> @@ -101,6 +101,97 @@ BITS    32
>      loop    .pageTableEntriesLoop4Level
>  %endmacro
>  
> +;
> +; Check whenever 5-level paging can ca used

(1) typo in comment

> +;
> +; Argument: jump label for 4-level paging
> +;
> +%macro Check5LevelPaging 1
> +    ; check for cpuid leaf 0x07
> +    mov     eax, 0x00
> +    cpuid
> +    cmp     eax, 0x07
> +    jb      %1
> +
> +    ; check for la57 (aka 5-level paging)
> +    mov     eax, 0x07
> +    mov     ecx, 0x00
> +    cpuid
> +    bt      ecx, 16
> +    jnc     %1
> +
> +    ; check for cpuid leaf 0x80000001
> +    mov     eax, 0x80000000
> +    cpuid
> +    cmp     eax, 0x80000001
> +    jb      %1
> +
> +    ; check for 1g pages
> +    mov     eax, 0x80000001
> +    cpuid
> +    bt      edx, 26
> +    jnc     %1
> +%endmacro
> +
> +;
> +; Create page tables for 5-level paging with gigabyte pages
> +;
> +; Argument: upper 32 bits of the page table entries
> +;
> +; We have 6 pages available for the early page tables,
> +; we use four of them:
> +;    PT_ADDR(0)      - level 5 directory
> +;    PT_ADDR(0x1000) - level 4 directory
> +;    PT_ADDR(0x2000) - level 2 directory (0 -> 1GB)
> +;    PT_ADDR(0x3000) - level 3 directory
> +;
> +; The level 2 directory for the first gigabyte has the same
> +; physical address in both 4-level and 5-level paging mode,
> +; SevClearPageEncMaskForGhcbPage depends on this.
> +;
> +; The 1 GB -> 4 GB range is mapped using 1G pages in the
> +; level 3 directory.
> +;
> +%macro CreatePageTables5Level 1
> +    ; level 5
> +    mov     dword[PT_ADDR (0)], PT_ADDR (0x1000) + PAGE_PDE_DIRECTORY_ATTR
> +    mov     dword[PT_ADDR (4)], %1
> +
> +    ; level 4
> +    mov     dword[PT_ADDR (0x1000)], PT_ADDR (0x3000) + PAGE_PDE_DIRECTORY_ATTR
> +    mov     dword[PT_ADDR (0x1004)], %1
> +
> +    ; level 3 (1x -> level 2, 3x 1GB)
> +    mov     dword[PT_ADDR (0x3000)], PT_ADDR (0x2000) + PAGE_PDE_DIRECTORY_ATTR
> +    mov     dword[PT_ADDR (0x3004)], %1
> +    mov     dword[PT_ADDR (0x3008)], (1 << 30) + PAGE_PDE_LARGEPAGE_ATTR
> +    mov     dword[PT_ADDR (0x300c)], %1
> +    mov     dword[PT_ADDR (0x3010)], (2 << 30) + PAGE_PDE_LARGEPAGE_ATTR
> +    mov     dword[PT_ADDR (0x3014)], %1
> +    mov     dword[PT_ADDR (0x3018)], (3 << 30) + PAGE_PDE_LARGEPAGE_ATTR
> +    mov     dword[PT_ADDR (0x301c)], %1
> +
> +    ;
> +    ; level 2 (512 * 2MB entries => 1GB)
> +    ;
> +    mov     ecx, 0x200
> +.pageTableEntriesLoop5Level:

(2) suggest "..@pageTableEntriesLoop5Level" as label name

> +    mov     eax, ecx
> +    dec     eax
> +    shl     eax, 21
> +    add     eax, PAGE_PDE_LARGEPAGE_ATTR
> +    mov     dword[ecx * 8 + PT_ADDR (0x2000 - 8)], eax
> +    mov     dword[(ecx * 8 + PT_ADDR (0x2000 - 8)) + 4], %1
> +    loop    .pageTableEntriesLoop5Level
> +%endmacro
> +
> +%macro Enable5LevelPaging 0
> +    ; set la57 bit in cr4
> +    mov     eax, cr4
> +    bts     eax, 12
> +    mov     cr4, eax
> +%endmacro
> +
>  ;
>  ; Modified:  EAX, EBX, ECX, EDX
>  ;
> @@ -125,6 +216,12 @@ SetCr3ForPageTables64:
>      ; normal (non-CoCo) workflow
>      ;
>      ClearOvmfPageTables
> +%if PG_5_LEVEL
> +    Check5LevelPaging Paging4Level
> +    CreatePageTables5Level 0
> +    jmp SetCr3La57
> +Paging4Level:
> +%endif
>      CreatePageTables4Level 0
>      jmp SetCr3
>  
> @@ -151,6 +248,14 @@ TdxBspInit:
>      OneTimeCall TdxPostBuildPageTables
>      jmp SetCr3
>  
> +    ;
> +    ; common workflow
> +    ;
> +%if PG_5_LEVEL
> +SetCr3La57:
> +    Enable5LevelPaging
> +%endif
> +
>  SetCr3:
>      ;
>      ; Set CR3 now that the paging structures are available

(3) I don't like SetCr3La57.

It saves minimal code duplication, but makes the control flow much
harder to follow.

(3.1) From this patch, we have one (unconditional) jump to SetCr3La57. I
suggest simply inlining that, like this:

%if PG_5_LEVEL
    Check5LevelPaging Paging4Level
    CreatePageTables5Level 0
    Enable5LevelPaging                 <-------- inline
    jmp SetCr3                         <-------- here
Paging4Level:
%endif

This adds one extra line before the unconditional jump, but removes 8
lines at the destination, *plus* it removes a quirky jump / reused code
path.

(3.2) From patch #8 ("OvmfPkg/ResetVector: wire up 5-level paging for
TDX"), we have two jumps to SetCr3La57; one conditional and one
unconditional.

- The unconditional jump can be replaced with inlining (invoke the
Enable5LevelPaging macro, then jump to SetCr3).

- The conditional one can be reworked like this:

%if PG_5_LEVEL
    cmp       eax, TDX_AP_5_LEVEL
    jne       CheckForSev
    Enable5LevelPaging
    jmp       SetCr3
CheckForSev:
%endif

This way, the conditional jump is *local*; way easier to follow. The
"CheckForSev" label much resembles the "Paging4Level" label above.

The patch looks fine otherwise.

Thanks!
Laszlo


> diff --git a/OvmfPkg/ResetVector/ResetVector.nasmb b/OvmfPkg/ResetVector/ResetVector.nasmb
> index 366a70fb9992..2bd80149e58b 100644
> --- a/OvmfPkg/ResetVector/ResetVector.nasmb
> +++ b/OvmfPkg/ResetVector/ResetVector.nasmb
> @@ -53,6 +53,7 @@
>  
>  %define WORK_AREA_GUEST_TYPE          (FixedPcdGet32 (PcdOvmfWorkAreaBase))
>  %define PT_ADDR(Offset)               (FixedPcdGet32 (PcdOvmfSecPageTablesBase) + (Offset))
> +%define PG_5_LEVEL                    (FixedPcdGetBool (PcdUse5LevelPageTable))
>  
>  %define GHCB_PT_ADDR                  (FixedPcdGet32 (PcdOvmfSecGhcbPageTableBase))
>  %define GHCB_BASE                     (FixedPcdGet32 (PcdOvmfSecGhcbBase))



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#116085): https://edk2.groups.io/g/devel/message/116085
Mute This Topic: https://groups.io/mt/104506793/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-



  reply	other threads:[~2024-02-28  5:33 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-22 11:54 [edk2-devel] [PATCH 00/10] OvmfPkg/ResetVector: cleanup and add 5-level paging support Gerd Hoffmann
2024-02-22 11:54 ` [edk2-devel] [PATCH 01/10] OvmfPkg/ResetVector: improve page table flag names Gerd Hoffmann
2024-02-22 11:54 ` [edk2-devel] [PATCH 02/10] OvmfPkg/ResetVector: add ClearOvmfPageTables macro Gerd Hoffmann
2024-02-28  4:09   ` Laszlo Ersek
2024-02-28  8:22     ` Gerd Hoffmann
2024-02-29  7:42       ` Laszlo Ersek
2024-02-22 11:54 ` [edk2-devel] [PATCH 03/10] OvmfPkg/ResetVector: add CreatePageTables4Level macro Gerd Hoffmann
2024-02-28  4:14   ` Laszlo Ersek
2024-02-22 11:54 ` [edk2-devel] [PATCH 04/10] OvmfPkg/ResetVector: split TDX BSP workflow Gerd Hoffmann
2024-02-28  4:34   ` Laszlo Ersek
2024-02-22 11:54 ` [edk2-devel] [PATCH 05/10] OvmfPkg/ResetVector: split SEV and non-CoCo workflows Gerd Hoffmann
2024-02-28  4:51   ` Laszlo Ersek
2024-02-22 11:54 ` [edk2-devel] [PATCH 06/10] OvmfPkg/ResetVector: add 5-level paging support Gerd Hoffmann
2024-02-28  5:33   ` Laszlo Ersek [this message]
2024-02-22 11:54 ` [edk2-devel] [PATCH 07/10] OvmfPkg/ResetVector: print post codes for 4/5 level paging Gerd Hoffmann
2024-02-28  5:35   ` Laszlo Ersek
2024-02-22 11:54 ` [edk2-devel] [PATCH 08/10] OvmfPkg/ResetVector: wire up 5-level paging for TDX Gerd Hoffmann
2024-02-28  5:44   ` Laszlo Ersek
2024-02-22 11:54 ` [edk2-devel] [PATCH 09/10] OvmfPkg/ResetVector: leave SEV VC handler installed longer Gerd Hoffmann
2024-02-28  5:52   ` Laszlo Ersek
2024-02-29 15:47   ` Lendacky, Thomas via groups.io
2024-02-22 11:54 ` [edk2-devel] [PATCH 10/10] OvmfPkg/ResetVector: wire up 5-level paging for SEV Gerd Hoffmann
2024-02-28  5:51   ` Laszlo Ersek

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-list from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=8163c5aa-47fd-330c-ba50-c24d676a5903@redhat.com \
    --to=devel@edk2.groups.io \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox