public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: "You, Benjamin" <benjamin.you@intel.com>
To: Matt Delco <delco@google.com>
Cc: "edk2-devel@lists.01.org" <edk2-devel@lists.01.org>,
	"Ma, Maurice" <maurice.ma@intel.com>,
	"Agyeman, Prince" <prince.agyeman@intel.com>
Subject: Re: [PATCH] CorebootModulePkg/CbSupportDxe: Enable ACPI via cmd register
Date: Fri, 1 Jun 2018 02:34:01 +0000	[thread overview]
Message-ID: <E748835C6D8DB54B8E8AF33091ECC57C6221EA43@SHSMSX103.ccr.corp.intel.com> (raw)
In-Reply-To: <CAHGX9VprrQwTvLqHxSjTpm9Yxcy7RsM2+=PS0M4rB=1JMJiVww@mail.gmail.com>

Hi Matt,

Sure. I will prepare a patch for this.

Thanks,

- ben

> From: Matt Delco [mailto:delco@google.com] 
> Sent: Friday, June 1, 2018 10:13 AM
> To: You, Benjamin <benjamin.you@intel.com>
> Cc: edk2-devel@lists.01.org; Ma, Maurice <maurice.ma@intel.com>; Agyeman, Prince <prince.agyeman@intel.com>
> Subject: Re: [PATCH] CorebootModulePkg/CbSupportDxe: Enable ACPI via cmd register
> 
> Normally I would provide a patch, but I'm not sure I'd add correctly add the checks that were proposed for CbParseFadtInfo().  I'd also have to go through an company-internal review process regarding edk2's contributor license agreement (I incorrectly assumed this had already been done when I provided the initial patch).  This review might require that I place an apache2 license and company copyright on patch contributions, and I'd feel really odd about providing a patch that added a license but otherwise merely deleted the lines that modify the register.  So, for this particular issue it might be cleaner, faster, and more straightforward if I didn't provide a patch.
> 
> > On Thu, May 31, 2018 at 5:44 PM, You, Benjamin <benjamin.you@intel.com> wrote:
> > Hi, Matt,
> > 
> > Thanks. Would you like to provide an updated patch?
> > 
> > Thanks,
> > 
> > - ben
> > 
> > > From: Matt Delco [mailto:delco@google.com] 
> > > Sent: Friday, June 1, 2018 5:55 AM
> > > To: You, Benjamin <benjamin.you@intel.com>
> > > Cc: edk2-devel@lists.01.org; Ma, Maurice <maurice.ma@intel.com>; Agyeman, Prince <prince.agyeman@intel.com>
> > > Subject: Re: [PATCH] CorebootModulePkg/CbSupportDxe: Enable ACPI via cmd register
> > > 
> > > Hi,
> > > 
> > > I commented out the body of OnReadyToBoot() and the result is as expected (i.e., ACPI events are still received, presumably because the OS enabled ACPI for itself).
> > > 
> > > > On Wed, May 30, 2018 at 7:52 PM, You, Benjamin <benjamin.you@intel.com> wrote:
> > > > Hi, Matt,
> > > > 
> > > > The original SCI_EN setting code was added to cover some early buggy 
> > > > firmware implementation that fails to enable SMM, and/or fails to fill FADT 
> > > > properly, and still needs an SCI_EN to transition to ACPI mode.
> > > > 
> > > > However, as you pointed out, setting it in Payload is not the correct way
> > > >  to address the issue, which also has some side effects as in your case. 
> > > > Instead, Payload should output some reminding message to developer to fix 
> > > > the firmware that runs before Payload.
> > > > 
> > > > Thanks,
> > > > 
> > > > - ben
> > > > 
> > > > > From: Matt Delco [mailto:delco@google.com] 
> > > > > Sent: Wednesday, May 30, 2018 4:40 PM
> > > > > To: You, Benjamin <benjamin.you@intel.com>
> > > > > Cc: edk2-devel@lists.01.org; Ma, Maurice <maurice.ma@intel.com>; Agyeman, Prince <prince.agyeman@intel.com>
> > > > > Subject: Re: [PATCH] CorebootModulePkg/CbSupportDxe: Enable ACPI via cmd register
> > > > > 
> > > > > Hi, thanks for taking a look.  I agree that this seems like the OS's responsibility to enable ACPI and I was tempted to just remove the setting of SCI_EN.  However, I figured that there was likely a reason that someone bothered to add the function in the first place so I went with what seemed like the more conservative fix.  I can try testing the removal of SCI_EN though I'll likely be at a remote office for the next 2 days so it may be 3 or more days before I can update the flash on the platform to perform the experiment.
> > > > > 
> > > > > On Wed, May 30, 2018 at 1:27 AM, You, Benjamin <benjamin.you@intel.com> wrote:
> > > > > Hi Matt,
> > > > > 
> > > > > This is indeed a bug. The Payload should not set the SCI_EN bit. 
> > > > > 
> > > > > However, the Payload should not trigger an SMI either. It is OS's 
> > > > > responsibility to enable ACPI by following FADT's instruction to write 
> > > > > some value to some port.
> > > > > 
> > > > > We think a better fix might be:
> > > > > 1) Remove the SCI_EN setting code in ReadyToBoot event handler,
> > > > > 2) Add some checking in CbParseFadtInfo() in CbParseLib.c to output some 
> > > > >    warning message and ASSERT (FALSE) if ALL of the following conditions 
> > > > >    are met:
> > > > >    - HARDWARE_REDUCED_ACPI is not set
> > > > >    - SMI_CMD field is zero (indicating no need to switch mode, hence 
> > > > >      the platform being in ACPI mode)
> > > > >    - SCI_EN bit is zero (indicating not in ACPI mode)
> > > > > 
> > > > > Item 2) above is to remind the developer that in this case, there may be a 
> > > > > bug in the firmware that runs before the Payload.
> > > > > 
> > > > > Could you please try on your platform to completely remove the SCI_EN 
> > > > > setting code in ReadToBoot event handler, and see if it works properly 
> > > > > (e.g., ACPI events delivered correctly)?
> > > > > 
> > > > > Thanks,
> > > > > 
> > > > > - ben
> > > > > 
> > > > > > -----Original Message-----
> > > > > > From: Matt Delco [mailto:delco@google.com] 
> > > > > > Sent: Monday, May 28, 2018 5:57 AM
> > > > > > To: edk2-devel@lists.01.org
> > > > > > Cc: Ma, Maurice <maurice.ma@intel.com>; Agyeman, Prince <prince.agyeman@intel.com>; You, Benjamin <benjamin.you@intel.com>
> > > > > > Subject: [PATCH] CorebootModulePkg/CbSupportDxe: Enable ACPI via cmd register
> > > > > > 
> > > > > > The existing code sets the SCI_EN bit directly, which is a violation
> > > > > > of the ACPI spec ("It is the responsibility of the hardware to set or
> > > > > > reset this bit. OSPM always preserves this bit position"). The proper
> > > > > > way to cause this bit to bit set is to reference the FADT table and
> > > > > > write the value of ACPI_ENABLE to SMI_CMD.  The SMI will in turn set
> > > > > > the SCI_EN bit.
> > > > > > 
> > > > > > Prior to this change ACPI events were not being delivered because
> > > > > > ACPI was not properly enabled and the OS also did not attempt
> > > > > > to enable ACPI since it sees that SCI_EN is already set.  After this
> > > > > > change I observed that ACPI events were now being delivered properly.
> > > > > > 
> > > > > > Cc: Maurice Ma <maurice.ma@intel.com>
> > > > > > Cc: Prince Agyeman <prince.agyeman@intel.com>
> > > > > > Cc: Benjamin You <benjamin.you@intel.com>
> > > > > > 
> > > > > > Contributed-under: TianoCore Contribution Agreement 1.0
> > > > > > Signed-off-by: Matt Delco <delco@google.com>
> > > > > > ---
> > > > > >  CorebootModulePkg/CbSupportDxe/CbSupportDxe.c | 79 ++++++++++++++++---
> > > > > >  .../CbSupportDxe/CbSupportDxe.inf             |  1 +
> > > > > >  2 files changed, 68 insertions(+), 12 deletions(-)
> > > > > > 
> > > > > > diff --git a/CorebootModulePkg/CbSupportDxe/CbSupportDxe.c b/CorebootModulePkg/CbSupportDxe/CbSupportDxe.c
> > > > > > index c526c9e871..4c7917ff2a 100755
> > > > > > --- a/CorebootModulePkg/CbSupportDxe/CbSupportDxe.c
> > > > > > +++ b/CorebootModulePkg/CbSupportDxe/CbSupportDxe.c
> > > > > > @@ -14,7 +14,10 @@
> > > > > >  **/
> > > > > >  #include "CbSupportDxe.h"
> > > > > >  
> > > > > > -UINTN mPmCtrlReg = 0;
> > > > > > +BOOLEAN mSmiPortFound = FALSE;
> > > > > > +UINTN mSmiCommandPort = 0;
> > > > > > +UINT8 mAcpiEnable = 0;
> > > > > > +
> > > > > >  /**
> > > > > >    Reserve MMIO/IO resource in GCD
> > > > > >  
> > > > > > @@ -107,9 +110,63 @@ OnReadyToBoot (
> > > > > >    //
> > > > > >    // Enable SCI
> > > > > >    //
> > > > > > -  IoOr16 (mPmCtrlReg, BIT0);
> > > > > > +  if (!mSmiPortFound) {
> > > > > > +    DEBUG ((DEBUG_ERROR, "SMI port not known so can not enable SCI\n"));
> > > > > > +  } else {
> > > > > > +    IoWrite8 (mSmiCommandPort, mAcpiEnable);
> > > > > > +    DEBUG ((DEBUG_ERROR, "Enable ACPI at 0x%lx with 0x%x before boot\n", (UINT64)mSmiCommandPort, (UINT32)mAcpiEnable));
> > > > > > +  }
> > > > > > +}
> > > > > > +
> > > > > > +/**
> > > > > > +  Lookup port and value for enabling ACPI
> > > > > > +
> > > > > > +  @param[in] SystemTable    A pointer to the EFI System Table.
> > > > > >  
> > > > > > -  DEBUG ((EFI_D_ERROR, "Enable SCI bit at 0x%lx before boot\n", (UINT64)mPmCtrlReg));
> > > > > > +**/
> > > > > > +VOID
> > > > > > +FindAcpiFadtTableInfo (
> > > > > > +  IN EFI_SYSTEM_TABLE   *SystemTable
> > > > > > +  )
> > > > > > +{
> > > > > > +  EFI_ACPI_2_0_ROOT_SYSTEM_DESCRIPTION_POINTER  *Rsdp;
> > > > > > +  EFI_ACPI_DESCRIPTION_HEADER                   *Rsdt;
> > > > > > +  EFI_ACPI_2_0_FIXED_ACPI_DESCRIPTION_TABLE     *Fadt;
> > > > > > +  UINTN                                         Index;
> > > > > > +  Rsdp  = NULL;
> > > > > > +  Rsdt  = NULL;
> > > > > > +  Fadt  = NULL;
> > > > > > +
> > > > > > +  for (Index = 0; Index < SystemTable->NumberOfTableEntries; Index++) {
> > > > > > +    if (CompareGuid (&(SystemTable->ConfigurationTable[Index].VendorGuid), &gEfiAcpi20TableGuid)) {
> > > > > > +      Rsdp = SystemTable->ConfigurationTable[Index].VendorTable;
> > > > > > +      break;
> > > > > > +    }
> > > > > > +  }
> > > > > > +  if (Rsdp == NULL) {
> > > > > > +    return;
> > > > > > +  }
> > > > > > +
> > > > > > +  Rsdt = (EFI_ACPI_DESCRIPTION_HEADER *)(UINTN) Rsdp->RsdtAddress;
> > > > > > +  if (Rsdt == NULL || Rsdt->Signature != EFI_ACPI_2_0_ROOT_SYSTEM_DESCRIPTION_TABLE_SIGNATURE) {
> > > > > > +    return;
> > > > > > +  }
> > > > > > +  for (Index = sizeof (EFI_ACPI_DESCRIPTION_HEADER); Index < Rsdt->Length; Index = Index + sizeof (UINT32)) {
> > > > > > +    UINT32 Data32  = *(UINT32 *) ((UINT8 *) Rsdt + Index);
> > > > > > +    Fadt    = (EFI_ACPI_2_0_FIXED_ACPI_DESCRIPTION_TABLE *) (UINT32 *) (UINTN) Data32;
> > > > > > +    if (Fadt->Header.Signature == EFI_ACPI_2_0_FIXED_ACPI_DESCRIPTION_TABLE_SIGNATURE) {
> > > > > > +      break;
> > > > > > +    }
> > > > > > +  }
> > > > > > +
> > > > > > +  if (Fadt == NULL || Fadt->Header.Signature != EFI_ACPI_2_0_FIXED_ACPI_DESCRIPTION_TABLE_SIGNATURE) {
> > > > > > +    return;
> > > > > > +  }
> > > > > > +
> > > > > > +  mSmiPortFound = TRUE;
> > > > > > +  mSmiCommandPort = Fadt->SmiCmd;
> > > > > > +  mAcpiEnable = Fadt->AcpiEnable;
> > > > > > +  return;
> > > > > >  }
> > > > > >  
> > > > > >  /**
> > > > > > @@ -133,7 +190,6 @@ CbDxeEntryPoint (
> > > > > >    EFI_EVENT  ReadyToBootEvent;
> > > > > >    EFI_HOB_GUID_TYPE  *GuidHob;
> > > > > >    SYSTEM_TABLE_INFO  *pSystemTableInfo;
> > > > > > -  ACPI_BOARD_INFO    *pAcpiBoardInfo;
> > > > > >    FRAME_BUFFER_INFO  *FbInfo;
> > > > > >  
> > > > > >    Status = EFI_SUCCESS;
> > > > > > @@ -172,14 +228,14 @@ CbDxeEntryPoint (
> > > > > >    }
> > > > > >  
> > > > > >    //
> > > > > > -  // Find the acpi board information guid hob
> > > > > > +  // Find the SMI port in the FADT table in acpi
> > > > > >    //
> > > > > > -  GuidHob = GetFirstGuidHob (&gUefiAcpiBoardInfoGuid);
> > > > > > -  ASSERT (GuidHob != NULL);
> > > > > > -  pAcpiBoardInfo = (ACPI_BOARD_INFO *)GET_GUID_HOB_DATA (GuidHob);
> > > > > > -
> > > > > > -  mPmCtrlReg = (UINTN)pAcpiBoardInfo->PmCtrlRegBase;
> > > > > > -  DEBUG ((EFI_D_ERROR, "PmCtrlReg at 0x%lx\n", (UINT64)mPmCtrlReg));
> > > > > > +  FindAcpiFadtTableInfo(SystemTable);
> > > > > > +  if (mSmiPortFound) {
> > > > > > +    DEBUG ((DEBUG_ERROR, "Found ACPI enable info in FADT: 0x%lx 0x%x\n", (UINT64)mSmiCommandPort, (UINT32)mAcpiEnable));
> > > > > > +  } else {
> > > > > > +    DEBUG ((DEBUG_ERROR, "Failed to find FADT info for enabling ACPI\n"));
> > > > > > +  }
> > > > > >  
> > > > > >    //
> > > > > >    // Find the frame buffer information and update PCDs
> > > > > > @@ -212,4 +268,3 @@ CbDxeEntryPoint (
> > > > > >  
> > > > > >    return EFI_SUCCESS;
> > > > > >  }
> > > > > > -
> > > > > > diff --git a/CorebootModulePkg/CbSupportDxe/CbSupportDxe.inf b/CorebootModulePkg/CbSupportDxe/CbSupportDxe.inf
> > > > > > index 99245183ea..e334dafdb7 100644
> > > > > > --- a/CorebootModulePkg/CbSupportDxe/CbSupportDxe.inf
> > > > > > +++ b/CorebootModulePkg/CbSupportDxe/CbSupportDxe.inf
> > > > > > @@ -51,6 +51,7 @@
> > > > > >  
> > > > > >  [Guids]
> > > > > >    gEfiAcpiTableGuid
> > > > > > +  gEfiAcpi20TableGuid
> > > > > >    gEfiSmbiosTableGuid
> > > > > >    gUefiSystemTableInfoGuid
> > > > > >    gUefiAcpiBoardInfoGuid
> > > > > > -- 
> > > > > > 2.17.0.921.gf22659ad46-goog



      reply	other threads:[~2018-06-01  2:34 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-05-27 21:57 [PATCH] CorebootModulePkg/CbSupportDxe: Enable ACPI via cmd register Matt Delco
2018-05-30  8:27 ` You, Benjamin
2018-05-30  8:40   ` Matt Delco
2018-05-31  2:52     ` You, Benjamin
2018-05-31 21:55       ` Matt Delco
2018-06-01  0:44         ` You, Benjamin
2018-06-01  2:13           ` Matt Delco
2018-06-01  2:34             ` You, Benjamin [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=E748835C6D8DB54B8E8AF33091ECC57C6221EA43@SHSMSX103.ccr.corp.intel.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