Re: [edk2-devel] [PATCH v9 0/6] Simplify SMM Relocation Process

["Wu, Jiaxin" <jiaxin.wu@intel.com>]

I can reproduce the issue and root cause to below code. GetFirstGuidHob() can't be used after exit boot service, that's the reason why I didn't catch the issue during the S3 test in shell. I will remove those check directly, and add debug message for the mSmmRelocated value.

  ASSERT (mSmmRelocated == (BOOLEAN)(GetFirstGuidHob (&gSmmBaseHobGuid) != NULL));
  if (mSmmRelocated != (BOOLEAN)(GetFirstGuidHob (&gSmmBaseHobGuid) != NULL)) {
    DEBUG ((
      DEBUG_ERROR,
      "gSmmBaseHobGuid %a produced in normal boot but %a in S3 boot!",
      mSmmRelocated ? "is" : "is not",
      mSmmRelocated ? "is not" : "is"
      ));
    CpuDeadLoop ();
  }

Thanks,
Jiaxin 




> -----Original Message-----
> From: Wu, Jiaxin
> Sent: Friday, March 10, 2023 6:18 PM
> To: kraxel@redhat.com
> Cc: devel@edk2.groups.io; Ni, Ray <ray.ni@intel.com>; Dong, Eric
> <eric.dong@intel.com>; Zeng, Star <star.zeng@intel.com>; Laszlo Ersek
> <lersek@redhat.com>; Kumar, Rahul R <rahul.r.kumar@intel.com>
> Subject: RE: [edk2-devel] [PATCH v9 0/6] Simplify SMM Relocation Process
> 
> I will have a try with the command.
> 
> For BIOS shell trigger S3, it just needs a simple application to provide the ACPI
> FACS waking vector, then write the power management control address at
> offset 4.
> 
> Another way is that shell MM command can achieve the trigger action, for
> example like: mm -io (xxx+4) 0x3400 -w 4, but I suppose the system will crash
> since there is no waking vector.
> 
> Thanks,
> Jiaxin
> 
> > -----Original Message-----
> > From: kraxel@redhat.com <kraxel@redhat.com>
> > Sent: Friday, March 10, 2023 6:01 PM
> > To: Wu, Jiaxin <jiaxin.wu@intel.com>
> > Cc: devel@edk2.groups.io; Ni, Ray <ray.ni@intel.com>; Dong, Eric
> > <eric.dong@intel.com>; Zeng, Star <star.zeng@intel.com>; Laszlo Ersek
> > <lersek@redhat.com>; Kumar, Rahul R <rahul.r.kumar@intel.com>
> > Subject: Re: [edk2-devel] [PATCH v9 0/6] Simplify SMM Relocation Process
> >
> > On Fri, Mar 10, 2023 at 09:19:04AM +0000, Wu, Jiaxin wrote:
> > > Is this regression? I remember I tried the S3 on real silicon with the patch
> > change, don't see the issue in normal boot work flow. The only difference
> what
> > I did is to trigger S3 in the BIOS shell, but it does covered the SmmRestoreCpu
> > test.
> > >
> > > Now, I'm trying to reproduce the issue on ovmf, and build the OVMF boot
> on
> > windows qemu, but I saw below error during boot:
> > > SecCoreStartupWithStack(0xFFFCC000, 0x820000)
> > > ASSERT [SecMain]
> > c:\dev\code\edk2\MdePkg\Library\BasePcdLibNull\PcdLib.c(95):
> > ((BOOLEAN)(0==1))
> > >
> > > Build command is:
> > > build -p OvmfPkg\OvmfPkgIa32X64.dsc -a X64 -D SMM_REQUIRE -D
> > DEBUG_ON_SERIAL_PORT -D SOURCE_DEBUG_ENABLE
> > >
> > > Run qemu command is:
> > > qemu-system-x86_64 -bios
> > C:\Dev\Code\edk2\Build\Ovmf3264\DEBUG_VS2015x86\FV\OVMF.fd -hda
> > fat:rw:hda-contents -serial file:log001.log -net none
> >
> > Mine is this:
> >
> > qemu-system-x86_64 -enable-kvm -machine q35,smm=on -m 4G -net none \
> > 	-drive
> >
> if=pflash,format=raw,file=Build/Ovmf3264/DEBUG_GCC5/FV/OVMF_CODE.fd,r
> > eadonly=on \
> > 	-drive
> >
> if=pflash,format=raw,file=Build/Ovmf3264/DEBUG_GCC5/FV/OVMF_VARS.fd,s
> > napshot=on \
> > 	-chardev vc,id=fw.log -device isa-
> > debugcon,iobase=0x402,chardev=fw.log \
> > 	-cdrom /vmdisk/iso/Fedora-Workstation-Live-x86_64-37-1.7.iso \
> > 	-device virtio-tablet-pci \
> > 	-display gtk,show-tabs=on
> >
> > > Anything need to be care to build OvmfPkgIa32X64?
> >
> > It must be started with the correct flash configuration, see above.
> > The SMM-enabled builds also require the q35 machine type.
> >
> > I have both SMM_REQUIRE and SECURE_BOOT enabled, that probably
> doesn't
> > make a difference though.  I'm using the isa-debugcon instead of the
> > serial port to get the debug log, that should not matter either.
> >
> > > BTW, how can I boot linux guest in windows qemu?
> >
> > Just grab some live iso.  The fedora image used above is available from
> > getfedora.org, but any live image should work.
> >
> > How do you trigger suspend from efi shell?  There seems to be no
> > builtin command for that.
> >
> > take care,
> >   Gerd