public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: Laszlo Ersek <lersek@redhat.com>
To: Eric Dong <eric.dong@intel.com>, edk2-devel@lists.01.org
Cc: Ruiyu Ni <ruiyu.ni@intel.com>
Subject: Re: [Patch v2 1/3] UefiCpuPkg/MpInitLib: Relocate uCode to memory to save time.
Date: Fri, 13 Jul 2018 00:13:50 +0200	[thread overview]
Message-ID: <ede5e87d-5c39-3e69-7688-98e639e15a55@redhat.com> (raw)
In-Reply-To: <debadb44-3d23-c4c8-98b8-5af3b7cdb01c@redhat.com>

On 07/12/18 23:59, Laszlo Ersek wrote:
> On 07/12/18 12:49, Eric Dong wrote:
>> Read uCode from memory has better performance than from flash.
>> But it needs extra effort to let BSP copy uCode from flash to
>> memory. Also BSP already enable cache in SEC phase, so it use
>> less time to relocate uCode from flash to memory. After
>> verification, if system has more than one processor, it will
>> reduce some time if load uCode from memory.
>>
>> This change enable this optimization.
>>
>> Cc: Laszlo Ersek <lersek@redhat.com>
>> Cc: Ruiyu Ni <ruiyu.ni@intel.com>
>> Contributed-under: TianoCore Contribution Agreement 1.1
>> Signed-off-by: Eric Dong <eric.dong@intel.com>
>> ---
>>  UefiCpuPkg/Library/MpInitLib/MpLib.c | 34 +++++++++++++++++++++++++++++++++-
>>  1 file changed, 33 insertions(+), 1 deletion(-)
>>
>> diff --git a/UefiCpuPkg/Library/MpInitLib/MpLib.c b/UefiCpuPkg/Library/MpInitLib/MpLib.c
>> index 108eea0a6f..c3cd6d7d51 100644
>> --- a/UefiCpuPkg/Library/MpInitLib/MpLib.c
>> +++ b/UefiCpuPkg/Library/MpInitLib/MpLib.c
>> @@ -1520,6 +1520,7 @@ MpInitLibInitialize (
>>    UINTN                    ApResetVectorSize;
>>    UINTN                    BackupBufferAddr;
>>    UINTN                    ApIdtBase;
>> +  VOID                     *MicrocodePatchInRam;
>>  
>>    OldCpuMpData = GetCpuMpDataFromGuidedHob ();
>>    if (OldCpuMpData == NULL) {
>> @@ -1587,8 +1588,39 @@ MpInitLibInitialize (
>>    CpuMpData->SwitchBspFlag    = FALSE;
>>    CpuMpData->CpuData          = (CPU_AP_DATA *) (CpuMpData + 1);
>>    CpuMpData->CpuInfoInHob     = (UINT64) (UINTN) (CpuMpData->CpuData + MaxLogicalProcessorNumber);
>> -  CpuMpData->MicrocodePatchAddress    = PcdGet64 (PcdCpuMicrocodePatchAddress);
>>    CpuMpData->MicrocodePatchRegionSize = PcdGet64 (PcdCpuMicrocodePatchRegionSize);
>> +  //
>> +  // If platform has more than one CPU, relocate microcode to memory to reduce
>> +  // loading microcode time.
>> +  //
>> +  MicrocodePatchInRam = NULL;
>> +  if (MaxLogicalProcessorNumber > 1) {
>> +    MicrocodePatchInRam = AllocatePages (
>> +                            EFI_SIZE_TO_PAGES (
>> +                              (UINTN)CpuMpData->MicrocodePatchRegionSize
>> +                              )
>> +                            );
>> +    ASSERT (MicrocodePatchInRam != NULL);
>> +  }
>> +  if (MicrocodePatchInRam == NULL) {
>> +    //
>> +    // there is only one processor, or no microcode patch is available, or
>> +    // memory allocation failed
>> +    //
>> +    CpuMpData->MicrocodePatchAddress = PcdGet64 (PcdCpuMicrocodePatchAddress);
>> +  } else {
>> +    //
>> +    // there are multiple processors, and a microcode patch is available, and
>> +    // memory allocation succeeded
>> +    //
>> +    CopyMem (
>> +      MicrocodePatchInRam,
>> +      (VOID *)(UINTN)PcdGet64 (PcdCpuMicrocodePatchAddress),
>> +      (UINTN)CpuMpData->MicrocodePatchRegionSize
>> +      );
>> +    CpuMpData->MicrocodePatchAddress = (UINTN)MicrocodePatchInRam;
>> +  }
>> +
>>    InitializeSpinLock(&CpuMpData->MpLock);
>>  
>>    //
>>
> 
> Reviewed-by: Laszlo Ersek <lersek@redhat.com>
> 

Sorry, I have to take that back -- please do not commit this patch.

For this version of the patch:

Nacked-by: Laszlo Ersek <lersek@redhat.com>

The ASSERT() is wrong. Again, in the code above, AllocatePages() can
return NULL not only because of memory allocation failure, but also
because the number of pages to allocate can be zero! If the platform has
no microcode patch to apply.

I knew this full well when I suggested the code, but then I forgot about
it when you mentioned the ASSERT(). I think you also knew about it, and
forgot about it too. :)

In particular, this patch would make it *impossible* to boot OVMF with
multiple processors, because OVMF *never* provides a microcode update.

So, please remove the ASSERT.

Alternatively, you could modify the ASSERT() like this:

  //
  // if we attempt actual memory allocation, we expect it to succeed
  //
  ASSERT (
    (CpuMpData->MicrocodePatchRegionSize == 0) ||
    (MicrocodePatchInRam != NULL)
    );

Thanks,
Laszlo


  reply	other threads:[~2018-07-12 22:13 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-07-12 10:49 [Patch v2 0/3] Optimize load uCode performance Eric Dong
2018-07-12 10:49 ` [Patch v2 1/3] UefiCpuPkg/MpInitLib: Relocate uCode to memory to save time Eric Dong
2018-07-12 21:59   ` Laszlo Ersek
2018-07-12 22:13     ` Laszlo Ersek [this message]
2018-07-13  0:45       ` Dong, Eric
2018-07-12 10:49 ` [Patch v2 2/3] UefiCpuPkg/MpInitLib: Use BSP uCode for APs if possible Eric Dong
2018-07-12 22:00   ` Laszlo Ersek
2018-07-12 10:49 ` [Patch v2 3/3] UefiCpuPkg/MpInitLib: Load uCode once for each core Eric Dong
2018-07-12 22:02   ` 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=ede5e87d-5c39-3e69-7688-98e639e15a55@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