ELF Coredump recoding between ISAs

[compor]

In order to facilitate executable migration across ISAs, we will use GDB to generate core dump files.

Currently, this can be broken down to:

• [x] #131 (addressed by #132)
  • [x] Use GDB to generate coredump files at specific migration points
  • [x] Use GDB to generate execution timings
• [x] Coredump file transformation AArch64 to X86_64
  • [x] Transform ELF header between archs
  • [x] Transform ELF segments between archs
  • [x] Register and state remapping between archs
  • [x] prstatus note (process satate/registers)
  • [x] fpregset note (FP registers)
  • [x] siginfo note (signal information)
  • [x] auxv note (auxiliary vector)
  • [x] ntfile note (mapped files)
  • [x] nt_x86_xsave note (X86 XSAVE extended state) optional/not required
• [ ] Coredump file transformation X86_64 to AArch64
  • [ ] prstatus note (process satate/registers)
  • [ ] fpregset note (FP registers)
  • [ ] siginfo note (signal information)
  • [ ] auxv note (auxiliary vector)
  • [ ] ntfile note (mapped files)
  • [x] nt_x86_xsave note (X86 XSAVE extended state) optional/not required
  • [x] prpsinfo note (process info) optional?
  • [ ] Address space size transformation (this might not need to be actionable)

[compor]

It seems that pyelftools do not support editing/saving changes in ELF format and are mainly used for reading/examining ELF files.

Thus, using a modified version of the criu coredump generator seems the most viable approach ATM.

[compor]

Not sure what to do with this ATM, it requires a bit of cleanup and then it can be partially merged. However, not sure what might be the point of this.

Let's decide later.

[compor]

Candidate for dropping.