Measured boot reports volume full error when chaining EFI binary from the Sovereign Boot Provisioning Wizard

[miczyg1]

For some reason when chaining a trusted bootloader from the Sovereign Boot Provisioning Wizard, the TCG2 measurement protocol responsible for measured boot report VOLUME_FULL error when logging a PCR extend event to the event log:

ReadPcr - 05
Supported PCRs - Count = 00000004
GetSupportedAndActivePcrs - Count = 00000004
ReadPcr - HashAlg = 0x0004, Pcr[05], digest = C8 30 29 00 14 D1 85 61 B1 72 2B 34 5E 3A 75 15 7A 29 50 CB 
ReadPcr - HashAlg = 0x000B, Pcr[05], digest = 47 32 D9 B2 9E 22 91 2A 70 2A 3B D0 87 3C 53 EF 4B D1 CE D0 49 89 A8 44 72 66 EE 4F 97 DC 6C 90 
ReadPcr - HashAlg = 0x000C, Pcr[05], digest = F5 76 F1 A6 94 64 2D 28 1D 58 49 1C FD F7 B9 6E 99 D0 09 3B FD 89 91 B2 3E A9 5B 4F E4 BC 2D 0C 41 7F 6E 73 AC 39 12 AD 4B 21 D9 1B 41 AC 64 CB 
ReadPcr - HashAlg = 0x000D, Pcr[05], digest = CB 9B D8 6C 2B CC 00 18 30 0A AA 89 1B D7 42 ED A4 55 9C FF 6A 51 6F A5 B2 19 A0 F5 A4 EA A3 C6 C3 D3 C0 81 7B 5E CD 97 43 48 5F 88 39 29 58 EC 86 9A A2 96 B7 F5 63 B9 0B 02 5D 78 11 10 61 A7 
SupportedEventLogs - 0x00000002
  LogFormat - 0x00000002
DxeTpm2MeasureBootHandler - Tcg2 MeasureGptTable - Volume Full
The measured image path is PciRoot(0x0)/Pci(0x1F,0x2)/Sata(0x1,0xFFFF,0x0)/HD(1,GPT,F106CED9-E27E-964C-9964-46BAAC7A60D4,0x800,0x7F000)/\EFI\ubuntu\shimx64.efi.
ReadPcr - 04
Supported PCRs - Count = 00000004
GetSupportedAndActivePcrs - Count = 00000004
ReadPcr - HashAlg = 0x0004, Pcr[04], digest = 0A 94 68 02 EF 40 A5 36 FF 4F 1C AC 29 B5 F8 12 8D 76 AA FA 
ReadPcr - HashAlg = 0x000B, Pcr[04], digest = F3 69 0B 55 2D EE B3 81 E0 C1 8F 7D 0D AB DF 06 08 ED CA 4C 63 13 24 B6 99 BA C2 B7 80 02 A4 62 
ReadPcr - HashAlg = 0x000C, Pcr[04], digest = 44 2E 43 83 1E 92 78 20 4E F1 4F 41 28 AD 4D A4 78 CA 86 B9 FA CB 5A FE 6B 3D DA 26 3A 60 3E 74 98 90 33 1F 1E B6 08 84 5B 89 0E BB B2 0A 3E 7D 
ReadPcr - HashAlg = 0x000D, Pcr[04], digest = 1E D3 78 63 3A 54 10 4D C6 F9 8E 7A 05 DD 76 B8 4D 46 0C 6B DB 89 E7 C0 C2 A8 92 95 03 87 ED 9E F7 CC EE D0 01 11 78 61 46 62 CB 8F 66 1D A5 E3 83 0B 2B 6F 1D 77 F9 5B 64 4A E8 9C 0B 91 97 47 
SupportedEventLogs - 0x00000002
  LogFormat - 0x00000002
DxeTpm2MeasureBootHandler - Tcg2 MeasurePeImage - Volume Full
DxeTpm2MeasureBootHandler - Success

It can be seen on the screen when booting a SHIM:

As a workaround, the TPM drivers have been disabled in the QEMU builds enabling Sovereign Boot Provisioning Wizard. The issue will be investigated in next project phases.

[miczyg1]

It is just as the error says, the TPM event log becomes full very quickly with Sovereign Boot. It is because the DB/DBX is appended with full certificates and its size grows quickly too. Each time dbx is written, it is being measured and logged to the TPM event log. As we can see in the below log:

VariableDxe: MeasureVariable (Pcr - 7, EventType - 80000001, VariableName - dbx, VendorGuid - D719B2CB-3D3A-4596-A3BC-DAD00E67656F)
ReadPcr - 07
Supported PCRs - Count = 00000004
GetSupportedAndActivePcrs - Count = 00000004
ReadPcr - HashAlg = 0x0004, Pcr[07], digest = AA 51 9A C6 32 E9 AA 12 40 31 6F FF FB 53 0A 3B AE 0E 43 EF 
ReadPcr - HashAlg = 0x000B, Pcr[07], digest = 05 9D F5 EB C6 C8 F1 7E 8E 43 83 F9 B5 91 94 1D EA C0 1D 32 78 43 0B 67 5A 68 B8 67 53 72 10 1F 
ReadPcr - HashAlg = 0x000C, Pcr[07], digest = EC C0 7D C5 85 16 CE 2A FD 8F 9D A2 8B 06 70 0D 90 C6 11 05 98 D5 F5 E1 9E 90 34 04 5D 6E F1 AC 53 F4 ED 03 F3 25 35 4E 1D 82 D5 F7 2F 39 1D 1D 
ReadPcr - HashAlg = 0x000D, Pcr[07], digest = F7 7C 94 68 22 14 89 61 DB C7 DF 32 3E E5 43 40 31 EF 8F 2A 83 89 C8 FA AC DC DE 33 24 8F 75 50 40 D0 26 E1 FD FB 9A 16 9B D8 39 2F F2 50 67 75 99 2D 33 5E ED B2 2A A8 C7 C4 43 CE 3E 49 3F 57 
SupportedEventLogs - 0x00000002
  LogFormat - 0x00000002
  Laml       - 0x10000
  NewLogSize - 0x519E
  LogSize    - 0xCD98
TcgCommLogEvent - Out of Resources
TcgDxeLogEvent:1012 TcgCommLogEvent Out of Resources

The lgo size is 64KB Laml - 0x10000. Currently occupied log space is LogSize - 0xCD98. The incoming new event size is NewLogSize - 0x519E. It shows hwo quickly the log can be overflown.

I guess we can increase its size to 16MB or so for Sovereign Boot to avoid this... But it should be triggered mainly during the boot with provisioning being conducted, as the SB variables are updated frequently. On normal boots when SV is provisioned, the log should not overflow that much.

[miczyg1]

Fixed by commit Dasharo/edk2@64c721e
In Dasharo/edk2#270