public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: "Laszlo Ersek" <lersek@redhat.com>
To: "Ni, Ray" <ray.ni@intel.com>, "Zeng, Star" <star.zeng@intel.com>,
	"devel@edk2.groups.io" <devel@edk2.groups.io>
Cc: "Dong, Eric" <eric.dong@intel.com>,
	"Philippe Mathieu-Daudé" <philmd@redhat.com>,
	"Kumar, Rahul1" <rahul1.kumar@intel.com>
Subject: Re: [PATCH 2/3] UefiCpuPkg/CpuS3DataDxe: do not allocate useless register tables
Date: Wed, 13 Jan 2021 10:16:36 +0100	[thread overview]
Message-ID: <65d582e9-95cf-7ceb-e379-f1df68a57478@redhat.com> (raw)
In-Reply-To: <CO1PR11MB4930CC9FEAAA2E0BB93AA08B8CA90@CO1PR11MB4930.namprd11.prod.outlook.com>

On 01/13/21 09:12, Ni, Ray wrote:
> Star,
> Thanks for pointing that.
> RegisterCpuFeaturesLib assumes [PreSmmInit]RegisterTable array is allocated but CpuS3Data
> doesn't do that.

Sorry that I missed this; I did grep the tree for
[PreSmmInit]RegisterTable, but apparently didn't pay enough attention to
RegisterCpuFeaturesLib. OVMF does not use that library.

Ray: can you please post your fix as a standalone patch?

Or else, if it's more convenient, please just push the fix somewhere (as
a standalone patch) where I can fetch it from. It would be great if you
could write a commit message too.

Then, I will post a v2 of this series, including your fix for
RegisterCpuFeaturesLib (as a patch under your authorship).

Star and Eric can then review your patch on the list -- neither I nor
you can review your patch, as you are the author of it, and I'll be
posting it.

Thanks!
Laszlo

> 
> Can following change fix the problem?
> 
> --- a/UefiCpuPkg/Library/RegisterCpuFeaturesLib/RegisterCpuFeaturesLib.c
> +++ b/UefiCpuPkg/Library/RegisterCpuFeaturesLib/RegisterCpuFeaturesLib.c
> @@ -937,21 +937,19 @@ GetAcpiCpuData (
>    EFI_PROCESSOR_INFORMATION            ProcessorInfoBuffer;
>  
>    AcpiCpuData = (ACPI_CPU_DATA *) (UINTN) PcdGet64 (PcdCpuS3DataAddress);
> -  if (AcpiCpuData != NULL) {
> -    return AcpiCpuData;
> -  }
> -
> -  AcpiCpuData  = AllocatePages (EFI_SIZE_TO_PAGES (sizeof (ACPI_CPU_DATA)));
> -  ASSERT (AcpiCpuData != NULL);
> +  if (AcpiCpuData == NULL) {
> +    AcpiCpuData  = AllocatePages (EFI_SIZE_TO_PAGES (sizeof (ACPI_CPU_DATA)));
> +    ASSERT (AcpiCpuData != NULL);
>  
> -  //
> -  // Set PcdCpuS3DataAddress to the base address of the ACPI_CPU_DATA structure
> -  //
> -  Status = PcdSet64S (PcdCpuS3DataAddress, (UINT64)(UINTN)AcpiCpuData);
> -  ASSERT_EFI_ERROR (Status);
> +    //
> +    // Set PcdCpuS3DataAddress to the base address of the ACPI_CPU_DATA structure
> +    //
> +    Status = PcdSet64S (PcdCpuS3DataAddress, (UINT64)(UINTN)AcpiCpuData);
> +    ASSERT_EFI_ERROR (Status);
>  
> -  GetNumberOfProcessor (&NumberOfCpus, &NumberOfEnabledProcessors);
> -  AcpiCpuData->NumberOfCpus = (UINT32)NumberOfCpus;
> +    GetNumberOfProcessor (&NumberOfCpus, &NumberOfEnabledProcessors);
> +    AcpiCpuData->NumberOfCpus = (UINT32)NumberOfCpus;
> +  }
>  
>    //
>    // Allocate buffer for empty RegisterTable and PreSmmInitRegisterTable for all CPUs
> 
> Thanks,
> Ray
> 
>> -----Original Message-----
>> From: Zeng, Star <star.zeng@intel.com>
>> Sent: Wednesday, January 13, 2021 3:17 PM
>> To: Ni, Ray <ray.ni@intel.com>; Laszlo Ersek <lersek@redhat.com>;
>> devel@edk2.groups.io
>> Cc: Dong, Eric <eric.dong@intel.com>; Philippe Mathieu-Daudé
>> <philmd@redhat.com>; Kumar, Rahul1 <rahul1.kumar@intel.com>; Zeng, Star
>> <star.zeng@intel.com>
>> Subject: RE: [PATCH 2/3] UefiCpuPkg/CpuS3DataDxe: do not allocate useless
>> register tables
>>
>> DxeRegisterCpuFeaturesLib still has some execution sequence dependency to
>> UefiCpuPkg CpuS3DataDxe.
>> There is ASSERT to happen at
>> https://github.com/tianocore/edk2/blob/master/UefiCpuPkg/Library/RegisterC
>> puFeaturesLib/RegisterCpuFeaturesLib.c#L1059 when CpuS3DataDxe with this
>> patch runs before DxeRegisterCpuFeaturesLib.
>>
>> Thanks,
>> Star
>> -----Original Message-----
>> From: Ni, Ray <ray.ni@intel.com>
>> Sent: Wednesday, January 13, 2021 2:12 PM
>> To: Laszlo Ersek <lersek@redhat.com>; devel@edk2.groups.io
>> Cc: Dong, Eric <eric.dong@intel.com>; Philippe Mathieu-Daudé
>> <philmd@redhat.com>; Kumar, Rahul1 <rahul1.kumar@intel.com>; Zeng, Star
>> <star.zeng@intel.com>
>> Subject: RE: [PATCH 2/3] UefiCpuPkg/CpuS3DataDxe: do not allocate useless
>> register tables
>>
>> Reviewed-by: Ray Ni <ray.ni@intel.com>
>>
>> (I guess you don't want to put Redhat copyright in the patch 1&2 so the
>> copyright year is not updated.
>> Since it's a simple change that make the logic more neat, I am ok with that.)
>>
>> Will you create a pull request to merge all 3 patches together?
>>
>>> -----Original Message-----
>>> From: Laszlo Ersek <lersek@redhat.com>
>>> Sent: Wednesday, January 13, 2021 3:20 AM
>>> To: devel@edk2.groups.io
>>> Cc: Dong, Eric <eric.dong@intel.com>; Philippe Mathieu-Daudé
>>> <philmd@redhat.com>; Kumar, Rahul1 <rahul1.kumar@intel.com>; Ni, Ray
>>> <ray.ni@intel.com>; Zeng, Star <star.zeng@intel.com>
>>> Subject: [PATCH 2/3] UefiCpuPkg/CpuS3DataDxe: do not allocate useless
>>> register tables
>>>
>>> CpuS3DataDxe allocates the "RegisterTable" and "PreSmmInitRegisterTable"
>>> arrays in ACPI_CPU_DATA just so every processor in the system can have
>>> its own empty register table, matched by APIC ID. This has never been
>>> useful in practice.
>>>
>>> Given commit e992cc3f4859 ("UefiCpuPkg PiSmmCpuDxeSmm: Reduce
>> SMRAM
>>> consumption in CpuS3.c", 2021-01-11), simply leave both
>>> "AcpiCpuData->RegisterTable" and "AcpiCpuData->PreSmmInitRegisterTable"
>>> initialized to the zero address. This simplifies the driver, and saves
>>> both normal RAM (boot services data type memory) and -- in
>>> PiSmmCpuDxeSmm
>>> -- SMRAM.
>>>
>>> Cc: Eric Dong <eric.dong@intel.com>
>>> Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
>>> Cc: Rahul Kumar <rahul1.kumar@intel.com>
>>> Cc: Ray Ni <ray.ni@intel.com>
>>> Cc: Star Zeng <star.zeng@intel.com>
>>> Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=3159
>>> Signed-off-by: Laszlo Ersek <lersek@redhat.com>
>>> ---
>>>
>>> Notes:
>>>     Tested by temporarily replacing OvmfPkgPkg/CpuS3DataDxe in the
>>> OVMF
>>> IA32
>>>     and IA32X64 platforms with this driver -- this driver works OK in OVMF
>>>     as long as no CPUs are hot-plugged.
>>>
>>>  UefiCpuPkg/CpuS3DataDxe/CpuS3Data.c | 32 --------------------
>>>  1 file changed, 32 deletions(-)
>>>
>>> diff --git a/UefiCpuPkg/CpuS3DataDxe/CpuS3Data.c
>>> b/UefiCpuPkg/CpuS3DataDxe/CpuS3Data.c
>>> index 2be335d91903..078af36cfb41 100644
>>> --- a/UefiCpuPkg/CpuS3DataDxe/CpuS3Data.c
>>> +++ b/UefiCpuPkg/CpuS3DataDxe/CpuS3Data.c
>>> @@ -165,10 +165,6 @@ CpuS3DataInitialize (
>>>    UINTN                      NumberOfCpus;
>>>    UINTN                      NumberOfEnabledProcessors;
>>>    VOID                       *Stack;
>>> -  UINTN                      TableSize;
>>> -  CPU_REGISTER_TABLE         *RegisterTable;
>>> -  UINTN                      Index;
>>> -  EFI_PROCESSOR_INFORMATION  ProcessorInfoBuffer;
>>>    UINTN                      GdtSize;
>>>    UINTN                      IdtSize;
>>>    VOID                       *Gdt;
>>> @@ -255,34 +251,6 @@ CpuS3DataInitialize (
>>>      AcpiCpuData->PreSmmInitRegisterTable = OldAcpiCpuData-
>>>> PreSmmInitRegisterTable;
>>>      AcpiCpuData->ApLocation              = OldAcpiCpuData->ApLocation;
>>>      CopyMem (&AcpiCpuData->CpuStatus, &OldAcpiCpuData->CpuStatus,
>>> sizeof (CPU_STATUS_INFORMATION));
>>> -  } else {
>>> -    //
>>> -    // Allocate buffer for empty RegisterTable and PreSmmInitRegisterTable
>> for
>>> all CPUs
>>> -    //
>>> -    TableSize = 2 * NumberOfCpus * sizeof (CPU_REGISTER_TABLE);
>>> -    RegisterTable = (CPU_REGISTER_TABLE *)AllocateZeroPages (TableSize);
>>> -    ASSERT (RegisterTable != NULL);
>>> -
>>> -    for (Index = 0; Index < NumberOfCpus; Index++) {
>>> -      Status = MpServices->GetProcessorInfo (
>>> -                           MpServices,
>>> -                           Index,
>>> -                           &ProcessorInfoBuffer
>>> -                           );
>>> -      ASSERT_EFI_ERROR (Status);
>>> -
>>> -      RegisterTable[Index].InitialApicId      =
>>> (UINT32)ProcessorInfoBuffer.ProcessorId;
>>> -      RegisterTable[Index].TableLength        = 0;
>>> -      RegisterTable[Index].AllocatedSize      = 0;
>>> -      RegisterTable[Index].RegisterTableEntry = 0;
>>> -
>>> -      RegisterTable[NumberOfCpus + Index].InitialApicId      =
>>> (UINT32)ProcessorInfoBuffer.ProcessorId;
>>> -      RegisterTable[NumberOfCpus + Index].TableLength        = 0;
>>> -      RegisterTable[NumberOfCpus + Index].AllocatedSize      = 0;
>>> -      RegisterTable[NumberOfCpus + Index].RegisterTableEntry = 0;
>>> -    }
>>> -    AcpiCpuData->RegisterTable           =
>>> (EFI_PHYSICAL_ADDRESS)(UINTN)RegisterTable;
>>> -    AcpiCpuData->PreSmmInitRegisterTable =
>>> (EFI_PHYSICAL_ADDRESS)(UINTN)(RegisterTable + NumberOfCpus);
>>>    }
>>>
>>>    //
>>> --
>>> 2.19.1.3.g30247aa5d201
>>>
> 


  parent reply	other threads:[~2021-01-13  9:16 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-01-12 19:19 [PATCH 0/3] UefiCpuPkg, OvmfPkg: do not allocate useless register tables for S3 resume Laszlo Ersek
2021-01-12 19:19 ` [PATCH 1/3] UefiCpuPkg/AcpiCpuData: update comments on register table fields Laszlo Ersek
2021-01-13  5:00   ` Ni, Ray
2021-01-15  9:28   ` Philippe Mathieu-Daudé
2021-01-15  9:34     ` Laszlo Ersek
2021-01-12 19:19 ` [PATCH 2/3] UefiCpuPkg/CpuS3DataDxe: do not allocate useless register tables Laszlo Ersek
2021-01-13  6:12   ` Ni, Ray
2021-01-13  7:16     ` Zeng, Star
2021-01-13  8:12       ` Ni, Ray
2021-01-13  8:40         ` Zeng, Star
2021-01-13  9:16         ` Laszlo Ersek [this message]
2021-01-14  1:55         ` Zeng, Star
2021-01-14 17:35           ` Laszlo Ersek
2021-01-15  8:33             ` [edk2-devel] " Ni, Ray
2021-01-15  8:37               ` Laszlo Ersek
2021-01-19 12:52                 ` Ni, Ray
2021-01-19 13:48                   ` Laszlo Ersek
2021-01-13  9:10     ` Laszlo Ersek
2021-01-12 19:19 ` [PATCH 3/3] OvmfPkg/CpuS3DataDxe: " Laszlo Ersek
2021-01-14 12:53   ` Ard Biesheuvel

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=65d582e9-95cf-7ceb-e379-f1df68a57478@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