Work-out what are we spending most of the time in OVMF

[csegarragonz]

According to the plot below, the biggest difference between SEV and non-SEV OVMF boots is in the MdeModulePkg/Core/Dxe/DxeMain/DxeMain.c method.

In particular, in these lines of OVMF code..

At first sight, nothing in those lines seems particularly suspicious. Annoyingly, we can not profile any further as the GetPerformanceCounter() method returns zero until we call the AcpiTimerLib constructor as part of ProcessLibraryConstructor at the very end of the linked code block. Thus, we still have no way to further discern where the overheads are coming from.

Note

The ProcessLibraryConstructor method is auto-generated during compilation, and can be found in the ./Build directory.
In my case, the file is located here: Build/AmdSev/RELEASE_GCC5/X64/MdeModulePkg/Core/Dxe/DxeMain/DEBUG/AutoGen.c.