public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: "Attar, AbdulLateef (Abdul Lateef)" <AbdulLateef.Attar@amd.com>
To: "Ni, Ray" <ray.ni@intel.com>,
	"devel@edk2.groups.io" <devel@edk2.groups.io>
Cc: "Grimes, Paul" <Paul.Grimes@amd.com>,
	"Kirkendall, Garrett" <Garrett.Kirkendall@amd.com>,
	"Chang, Abner" <Abner.Chang@amd.com>,
	"Dong, Eric" <eric.dong@intel.com>,
	"Kumar, Rahul R" <rahul.r.kumar@intel.com>
Subject: Re: [edk2-devel] [PATCH v6 6/6] UefiCpuPkg: Implements SmmCpuFeaturesLib for AMD Family
Date: Fri, 7 Apr 2023 06:09:34 +0000	[thread overview]
Message-ID: <PH7PR12MB64660F96474F05B86733E831E0969@PH7PR12MB6466.namprd12.prod.outlook.com> (raw)
In-Reply-To: <MN6PR11MB82447A35A29B221E2AC3A4A38C8F9@MN6PR11MB8244.namprd11.prod.outlook.com>

[AMD Official Use Only - General]

Hi Ray,
        Please see inline.
Thanks
AbduL
-----Original Message-----
From: Ni, Ray <ray.ni@intel.com>
Sent: 31 March 2023 13:29
To: devel@edk2.groups.io; Attar, AbdulLateef (Abdul Lateef) <AbdulLateef.Attar@amd.com>
Cc: Attar, AbdulLateef (Abdul Lateef) <AbdulLateef.Attar@amd.com>; Grimes, Paul <Paul.Grimes@amd.com>; Kirkendall, Garrett <Garrett.Kirkendall@amd.com>; Chang, Abner <Abner.Chang@amd.com>; Dong, Eric <eric.dong@intel.com>; Kumar, Rahul R <rahul.r.kumar@intel.com>
Subject: RE: [edk2-devel] [PATCH v6 6/6] UefiCpuPkg: Implements SmmCpuFeaturesLib for AMD Family

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.


It's a bit wired that AMD version of SmmCpuFeaturesLib calls AMD version of SmmSmramSaveStateLib.
The question naturally becomes: why not integrate the two lib implementations together to avoid creating the SmmSmramSaveStateLib library class? Because I cannot imagine AmdSmmCpuFeaturesLib calls to IntelSmmSmramSaveStateLib.
[Abdul] Initial version of this patch series has combined implementation of SmmSmramSaveStateLib and AmdSmmCpuFeaturesLib, but reviewers (Abner) suggested to separate the save state library; So that we can implement generic library for both AMD and Intel.

But I like the idea of separating the SMM save state access logic out of the SmmCpuFeaturesLib.
Do you think CpuSmm driver calling the SmmSmramSaveStateLib is better?
[Abdul] I'll make the changes and submit the patch for the same.

> -----Original Message-----
> From: devel@edk2.groups.io <devel@edk2.groups.io> On Behalf Of Abdul
> Lateef Attar via groups.io
> Sent: Saturday, March 25, 2023 1:39 PM
> To: devel@edk2.groups.io
> Cc: Abdul Lateef Attar <AbdulLateef.Attar@amd.com>; Paul Grimes
> <paul.grimes@amd.com>; Garrett Kirkendall
> <garrett.kirkendall@amd.com>; Abner Chang <abner.chang@amd.com>; Dong,
> Eric <eric.dong@intel.com>; Ni, Ray <ray.ni@intel.com>; Kumar, Rahul R
> <rahul.r.kumar@intel.com>; Abdul Lateef Attar <abdattar@amd.com>
> Subject: [edk2-devel] [PATCH v6 6/6] UefiCpuPkg: Implements
> SmmCpuFeaturesLib for AMD Family
>
> From: Abdul Lateef Attar <AbdulLateef.Attar@amd.com>
>
> BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=4182
>
> Implements interfaces to read and write save state registers of AMD's
> processor family.
> Initializes processor SMMADDR and MASK depends on PcdSmrrEnable flag.
> Program or corrects the IP once control returns from SMM.
>
> Cc: Paul Grimes <paul.grimes@amd.com>
> Cc: Garrett Kirkendall <garrett.kirkendall@amd.com>
> Cc: Abner Chang <abner.chang@amd.com>
> Cc: Eric Dong <eric.dong@intel.com>
> Cc: Ray Ni <ray.ni@intel.com>
> Cc: Rahul Kumar <rahul1.kumar@intel.com>
> Signed-off-by: Abdul Lateef Attar <abdattar@amd.com>
> Reviewed-by: Abner Chang <abner.chang@amd.com>
> ---
>  .../AmdSmmCpuFeaturesLib.inf                  |   6 +
>  .../SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.c  | 106
> +++++++++++++++++-
>  2 files changed, 109 insertions(+), 3 deletions(-)
>
> diff --git
> a/UefiCpuPkg/Library/SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.inf
> b/UefiCpuPkg/Library/SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.inf
> index 4c77efc64462..9d5b8c2e972d 100644
> --- a/UefiCpuPkg/Library/SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.inf
> +++
> b/UefiCpuPkg/Library/SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.inf
> @@ -31,3 +31,9 @@ [LibraryClasses]
>    PcdLib
>    MemoryAllocationLib
>    DebugLib
> +  SmmSmramSaveStateLib
> +
> +[FeaturePcd]
> +  gUefiCpuPkgTokenSpaceGuid.PcdSmrrEnable               ## CONSUMES
> +  gUefiCpuPkgTokenSpaceGuid.PcdSmmFeatureControlEnable  ##
> CONSUMES
> +
> diff --git
> a/UefiCpuPkg/Library/SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.c
> b/UefiCpuPkg/Library/SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.c
> index c74e1a0c0c5b..af45be3e265a 100644
> --- a/UefiCpuPkg/Library/SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.c
> +++ b/UefiCpuPkg/Library/SmmCpuFeaturesLib/AmdSmmCpuFeaturesLib.c
> @@ -11,6 +11,21 @@ SPDX-License-Identifier: BSD-2-Clause-Patent
>
>  #include <Library/SmmCpuFeaturesLib.h>  #include
> <Uefi/UefiBaseType.h>
> +#include <Register/Amd/SmramSaveStateMap.h>
> +#include <Library/BaseLib.h>
> +#include <Library/DebugLib.h>
> +#include <Library/SmmSmramSaveStateLib.h>
> +
> +// EFER register LMA bit
> +#define LMA  BIT10
> +
> +// Machine Specific Registers (MSRs)
> +#define SMMADDR_ADDRESS  0xC0010112ul #define SMMMASK_ADDRESS
> +0xC0010113ul
> +#define EFER_ADDRESS     0XC0000080ul
> +
> +// The mode of the CPU at the time an SMI occurs STATIC UINT8
> +mSmmSaveStateRegisterLma;
>
>  /**
>    Read an SMM Save State register on the target processor.  If this
> function @@ -39,7 +54,7 @@ SmmCpuFeaturesReadSaveStateRegister (
>    OUT VOID                         *Buffer
>    )
>  {
> -  return EFI_SUCCESS;
> +  return SmramSaveStateReadRegister (CpuIndex, Register, Width,
> + Buffer);
>  }
>
>  /**
> @@ -67,7 +82,7 @@ SmmCpuFeaturesWriteSaveStateRegister (
>    IN CONST VOID                   *Buffer
>    )
>  {
> -  return EFI_SUCCESS;
> +  return SmramSaveStateWriteRegister (CpuIndex, Register, Width,
> + Buffer);
>  }
>
>  /**
> @@ -82,6 +97,13 @@ CpuFeaturesLibInitialization (
>    VOID
>    )
>  {
> +  UINT32  LMAValue;
> +
> +  LMAValue                 = (UINT32)AsmReadMsr64 (EFER_ADDRESS) & LMA;
> +  mSmmSaveStateRegisterLma =
> EFI_SMM_SAVE_STATE_REGISTER_LMA_32BIT;
> +  if (LMAValue) {
> +    mSmmSaveStateRegisterLma =
> EFI_SMM_SAVE_STATE_REGISTER_LMA_64BIT;
> +  }
>  }
>
>  /**
> @@ -117,6 +139,52 @@ SmmCpuFeaturesInitializeProcessor (
>    IN CPU_HOT_PLUG_DATA          *CpuHotPlugData
>    )
>  {
> +  AMD_SMRAM_SAVE_STATE_MAP  *CpuState;
> +  UINT32                    LMAValue;
> +
> +  //
> +  // Configure SMBASE.
> +  //
> +  CpuState             = (AMD_SMRAM_SAVE_STATE_MAP
> *)(UINTN)(SMM_DEFAULT_SMBASE + SMRAM_SAVE_STATE_MAP_OFFSET);
> +  CpuState->x64.SMBASE = (UINT32)CpuHotPlugData->SmBase[CpuIndex];
> +
> +  // Re-initialize the value of mSmmSaveStateRegisterLma flag which
> + might
> have been changed in PiCpuSmmDxeSmm Driver
> +  // Entry point, to make sure correct value on AMD platform is
> + assigned to
> be used by SmmCpuFeaturesLib.
> +  LMAValue                 = (UINT32)AsmReadMsr64 (EFER_ADDRESS) & LMA;
> +  mSmmSaveStateRegisterLma =
> EFI_SMM_SAVE_STATE_REGISTER_LMA_32BIT;
> +  if (LMAValue) {
> +    mSmmSaveStateRegisterLma =
> EFI_SMM_SAVE_STATE_REGISTER_LMA_64BIT;
> +  }
> +
> +  //
> +  // If SMRR is supported, then program SMRR base/mask MSRs.
> +  // The EFI_MSR_SMRR_PHYS_MASK_VALID bit is not set until the first
> normal SMI.
> +  // The code that initializes SMM environment is running in normal
> + mode  // from SMRAM region.  If SMRR is enabled here, then the SMRAM
> + region  // is protected and the normal mode code execution will fail.
> +  //
> +  if (FeaturePcdGet (PcdSmrrEnable)) {
> +    //
> +    // SMRR size cannot be less than 4-KBytes
> +    // SMRR size must be of length 2^n
> +    // SMRR base alignment cannot be less than SMRR length
> +    //
> +    if ((CpuHotPlugData->SmrrSize < SIZE_4KB) ||
> +        (CpuHotPlugData->SmrrSize != GetPowerOfTwo32 (CpuHotPlugData-
> >SmrrSize)) ||
> +        ((CpuHotPlugData->SmrrBase & ~(CpuHotPlugData->SmrrSize - 1))
> + !=
> CpuHotPlugData->SmrrBase))
> +    {
> +      //
> +      // Print message and halt if CPU is Monarch
> +      //
> +      if (IsMonarch) {
> +        DEBUG ((DEBUG_ERROR, "SMM Base/Size does not meet
> alignment/size requirement!\n"));
> +        CpuDeadLoop ();
> +      }
> +    } else {
> +      AsmWriteMsr64 (SMMADDR_ADDRESS, CpuHotPlugData->SmrrBase);
> +      AsmWriteMsr64 (SMMMASK_ADDRESS, ((~(UINT64)(CpuHotPlugData-
> >SmrrSize - 1)) | 0x6600));
> +    }
> +  }
>  }
>
>  /**
> @@ -159,7 +227,39 @@ SmmCpuFeaturesHookReturnFromSmm (
>    IN UINT64                NewInstructionPointer
>    )
>  {
> -  return 0;
> +  UINT64                    OriginalInstructionPointer;
> +  AMD_SMRAM_SAVE_STATE_MAP  *AmdCpuState;
> +
> +  AmdCpuState = (AMD_SMRAM_SAVE_STATE_MAP *)CpuState;
> +
> +  if (mSmmSaveStateRegisterLma ==
> EFI_SMM_SAVE_STATE_REGISTER_LMA_32BIT) {
> +    OriginalInstructionPointer = (UINT64)AmdCpuState->x86._EIP;
> +    AmdCpuState->x86._EIP      = (UINT32)NewInstructionPointer;
> +    //
> +    // Clear the auto HALT restart flag so the RSM instruction returns
> +    // program control to the instruction following the HLT instruction.
> +    //
> +    if ((AmdCpuState->x86.AutoHALTRestart & BIT0) != 0) {
> +      AmdCpuState->x86.AutoHALTRestart &= ~BIT0;
> +    }
> +  } else {
> +    OriginalInstructionPointer = AmdCpuState->x64._RIP;
> +    if ((AmdCpuState->x64.EFER & LMA) == 0) {
> +      AmdCpuState->x64._RIP = (UINT32)NewInstructionPointer32;
> +    } else {
> +      AmdCpuState->x64._RIP = (UINT32)NewInstructionPointer;
> +    }
> +
> +    //
> +    // Clear the auto HALT restart flag so the RSM instruction returns
> +    // program control to the instruction following the HLT instruction.
> +    //
> +    if ((AmdCpuState->x64.AutoHALTRestart & BIT0) != 0) {
> +      AmdCpuState->x64.AutoHALTRestart &= ~BIT0;
> +    }
> +  }
> +
> +  return OriginalInstructionPointer;
>  }
>
>  /**
> --
> 2.25.1
>
>
>
> 
>


      reply	other threads:[~2023-04-07  6:09 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-03-25  5:38 [PATCH v6 0/6] Adds AmdSmmCpuFeaturesLib for AMD Family Abdul Lateef Attar
2023-03-25  5:38 ` [PATCH v6 1/6] MdePkg: Adds AMD SMRAM save state map Abdul Lateef Attar
2023-03-25  5:38 ` [PATCH v6 2/6] UefiCpuPkg: Adds SmmSmramSaveStateLib library class Abdul Lateef Attar
2023-03-25  5:38 ` [PATCH v6 3/6] UefiCpuPkg: Implements " Abdul Lateef Attar
2023-03-25  5:38 ` [PATCH v6 4/6] UefiCpuPkg/SmmCpuFeaturesLib: Restructure arch-dependent code Abdul Lateef Attar
2023-03-25  5:38 ` [PATCH v6 5/6] UefiCpuPkg: Initial implementation of AMD's SmmCpuFeaturesLib Abdul Lateef Attar
2023-03-25  5:38 ` [PATCH v6 6/6] UefiCpuPkg: Implements SmmCpuFeaturesLib for AMD Family Abdul Lateef Attar
2023-03-31  7:59   ` [edk2-devel] " Ni, Ray
2023-04-07  6:09     ` Attar, AbdulLateef (Abdul Lateef) [this message]

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=PH7PR12MB64660F96474F05B86733E831E0969@PH7PR12MB6466.namprd12.prod.outlook.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