Linux and QEMU branches updated

Hi everyone,

A few weeks ago we discussed the plan to update the Linux and QEMU development branches for the SVSM to the versions Roy provided.

Today I made this change and updated the svsm branches in both repositories. I also updated the SVSM tree to work with the new KVM base, which includes updated documentation, launch script, and disabling one of the unit-tests[1].

So hereby I ask everyone to update their development/test environments to use the latest Linux, QEMU and SVSM branches from the project.

Why is this important?

There is a PR pending for the SVSM which introduces the use of the TPR for managing IRQ priorities. This is part of the ongoing work to enable IRQs in the SVSM and add the ability to send IPIs, which is required to enable other platforms besides AMD SEV-SNP.

With the old 6.8-based kernel the TPR was shared across VMPLs, which made the SVSM observe values set by the Linux kernel. This breaks the SVSM as soon as Linux initialized the local APIC.

Going forward, when the SVSM TPR PR gets merged, the SVSM will no longer work on the 6.8 kernel, so please everyone make sure to update your development/test environments and let me know in case you run into any problems.

Regards,

Joerg

[1] That test fails now because it relies on an intercept bit no longer set.

Hi Jörg,

I tested all components, I can boot the guest OS and use the vTPM, but PCRs were strange:

# tpm2_pcrread
  sha1:
  sha256:
    0 : 0x0000000000000000000000000000000000000000000000000000000000000000
    1 : 0x0000000000000000000000000000000000000000000000000000000000000000
    2 : 0x0000000000000000000000000000000000000000000000000000000000000000
    3 : 0x0000000000000000000000000000000000000000000000000000000000000000
    4 : 0x0000000000000000000000000000000000000000000000000000000000000000
    5 : 0x0000000000000000000000000000000000000000000000000000000000000000
    6 : 0x0000000000000000000000000000000000000000000000000000000000000000
    7 : 0x0000000000000000000000000000000000000000000000000000000000000000
    8 : 0x0000000000000000000000000000000000000000000000000000000000000000
    9 : 0x0000000000000000000000000000000000000000000000000000000000000000
    10: 0xBF0D858E3904704B36740BC2DDCF4820B93A9323C1098338B7C38E338735257B
    11: 0x0000000000000000000000000000000000000000000000000000000000000000
    12: 0x0000000000000000000000000000000000000000000000000000000000000000
    13: 0x0000000000000000000000000000000000000000000000000000000000000000
    14: 0x0000000000000000000000000000000000000000000000000000000000000000
    15: 0x0000000000000000000000000000000000000000000000000000000000000000
    16: 0x0000000000000000000000000000000000000000000000000000000000000000
    17: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    18: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    19: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    20: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    21: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    22: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    23: 0x0000000000000000000000000000000000000000000000000000000000000000

I noticed that in the svsm branch of edk2 the TPM driver is no longer there, so I tried the PR [1] that Oliver opened upstream, and it seems to be working:

# tpm2_pcrread
  sha1:
  sha256:
    0 : 0xEE32461432E54C50E78118642B66B770AEDE2DDBE86256AD39BF6C1E3A0C2270
    1 : 0x3E56433CCE1C5ADE10DFBEB087645E4E4CE1440D9623F9106841B660812B9E95
    2 : 0x3D458CFE55CC03EA1F443F1562BEEC8DF51C75E14A9FCF9A7234A13F198E7969
    3 : 0x3D458CFE55CC03EA1F443F1562BEEC8DF51C75E14A9FCF9A7234A13F198E7969
    4 : 0xE202304E038D4985F817949AC2BEC2A6E969B81BA893AFC2E92C5158292C587E
    5 : 0x5CD9240D9D327C6907507075091BAF761CB81D95571C8A49F72B317B2A1A3F31
    6 : 0x3D458CFE55CC03EA1F443F1562BEEC8DF51C75E14A9FCF9A7234A13F198E7969
    7 : 0x127C18EBA2300E30767FAFE71F4E5975776F665D22C7CA9017C7C24846B96FA1
    8 : 0x5AD47B20CE56F0ED9985BA9E338D833C2E7C29981B158C1BCD15B616579D5555
    9 : 0x9269DA5E4B8F2CA44BE308734FC233C9603F090EADFB4C2E5CBF30331AFB0BC3
    10: 0x32A87359AFD16F31B3C13BE34054D16C11562A4B2FEC0837B8E24CD2374E3F44
    11: 0x0000000000000000000000000000000000000000000000000000000000000000
    12: 0x0000000000000000000000000000000000000000000000000000000000000000
    13: 0x0000000000000000000000000000000000000000000000000000000000000000
    14: 0x17CDEFD9548F4383B67A37A901673BF3C8DED6F619D36C8007562DE1D93C81CC
    15: 0x0000000000000000000000000000000000000000000000000000000000000000
    16: 0x0000000000000000000000000000000000000000000000000000000000000000
    17: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    18: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    19: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    20: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    21: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    22: 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
    23: 0x0000000000000000000000000000000000000000000000000000000000000000

Should we include that PR in our fork?

Thanks,
Stefano

[1] https://github.com/tianocore/edk2/pull/6527

Hi Stefano,

Yes, please also open a PR against the svsm branch of our EDK2 fork. I wll merge it then so that we get back the vTPM driver.

Regards,

Joerg

-> https://github.com/coconut-svsm/edk2/pull/5

Thanks, merged now.