eld currently behaves differently in case of the Common/standalone/Patching/PatchSections/PatchSections.test in windows and linux for RISC-V (both 64 and 32 bit).
The output seems to have an arbitrary order especially in case of:
82: Note: Internal Section : .rela.pgot Alignment : 0x4 Size : 0x0 Flags : NONE
83: Note: Internal Section : .pgot Alignment : 0x4 Size : 0x0 Flags : SHF_ALLOC|SHF_WRITE
The above output is for the RISC-V backend on windows. On Linux the order seems to be flipped; and thus, eld exhibits non-deterministic behavior in the case of this test.