When hooking a function with libmem, the generated JMP patch does not fill the remaining overwritten bytes with NOP instructions.

Original code:

test.exe+28D5BF - 8B 04855442 A9 00-moveax,[eax*4+test.exe+694254]
test.exe+28D5C6 - C3 -ret
test.exe+28D5C7 -90-nop
test.exe+28D5C8 -53-pushebx
test.exe+28D5C9 -83 EC 1C -subesp,1C

After hooking test.exe+28D5BF, libmem correctly writes the JMP instruction, but the remaining bytes are left untouched:

test.exe+28D5BF - E9 3C2A5104 -jmp 04BA0000
test.exe+28D5C4 - A9 00C39053 -testeax,5390C300
test.exe+28D5C9 -83 EC 1C -subesp,1C

This causes the disassembly to become corrupted because the original instruction length was 7 bytes, while the JMP instruction only uses 5 bytes.

Expected behavior:

test.exe+28D5BF - E9 3C2A1C0C -jmp 0C850000
test.exe+28D5C4 -90-nop
test.exe+28D5C5 -90-nop
test.exe+28D5C6 - C3 -ret
test.exe+28D5C7 -90-nop
test.exe+28D5C8 -53-pushebx
test.exe+28D5C9 -83 EC 1C -subesp,1C

It would be helpful if libmem could automatically pad the remaining bytes with NOPs when the overwritten instruction size is larger than the JMP patch size.