编译动态链接内核

Question

I am attempting to implement KASLR in the xv6 kernel, so i need to recompile/link the kernel in a dynamic way(relocation table). Since the kernel is currently hardcoded, even if I change the kernel code and data mappings, the instructions will still reference hardcoded virtual addresses. I need a way for the instructions to reference other parts of the elf without hardcoded addresses. An example of this is below: This is a sample of how it looks after current compilation:

8010018e:   83 ec 0c                sub    $0xc,%esp
80100191:   68 a7 79 10 80          push   $0x801079a7
80100196:   e8 b5 01 00 00          call   80100350 <panic>
8010019b:   90                      nop
8010019c:   8d 74 26 00             lea    0x0(%esi,%eiz,1),%esi.

As you can see, the instructions such as the push instruction reference hardcoded v addresses($0x801079a7). This is because this elf is compiled in a static way with no relocations:(readelf printout)

 Program Headers:
    Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align
    LOAD           0x001000 0x80100000 0x00100000 0x0b516 0x15668 RWE 0x1000
    GNU_STACK      0x000000 0x00000000 0x00000000 0x00000 0x00000 RWE 0x10
    Section to Segment mapping:
    Segment Sections...
    00     .text .rodata .stab .stabstr .data .bss
    01
    There is no dynamic section in this file.
    There are no relocations in this file.

I need advice on how to recompile a statically linked elf binary to one with relocation and then be able to map the kernel anywhere in virtual memory. Additionally, what would I have to modify in the bootloader for parsing the elf to load it correctly with relocation? Will I have to custom parse a GOT included in the kernel elf?

The following is a linker script to compile the kernel and essentially set its base hardcoded virtual address: 0x80100000

OUTPUT_FORMAT("elf32-i386", "elf32-i386", "elf32-i386")
OUTPUT_ARCH(i386)
ENTRY(_start)

SECTIONS
{
    /* Link the kernel at this address: "." means the current address */
        /* Must be equal to KERNLINK */
    . = 0x80100000;

    .text : AT(0x100000) {
        *(.text .stub .text.* .gnu.linkonce.t.*)
    }

    PROVIDE(etext = .); /* Define the 'etext' symbol to this value */

    .rodata : {
        *(.rodata .rodata.* .gnu.linkonce.r.*)
    }

    /* Include debugging information in kernel memory */
    .stab : {
        PROVIDE(__STAB_BEGIN__ = .);
        *(.stab);
        PROVIDE(__STAB_END__ = .);
        BYTE(0)     /* Force the linker to allocate space
                   for this section */
    }

    .stabstr : {
        PROVIDE(__STABSTR_BEGIN__ = .);
        *(.stabstr);
        PROVIDE(__STABSTR_END__ = .);
        BYTE(0)     /* Force the linker to allocate space
                   for this section */
    }

    /* Adjust the address for the data segment to the next page */
    . = ALIGN(0x1000);

    /* Conventionally, Unix linkers provide pseudo-symbols
     * etext, edata, and end, at the end of the text, data, and bss.
     * For the kernel mapping, we need the address at the beginning
     * of the data section, but that's not one of the conventional
     * symbols, because the convention started before there was a
     * read-only rodata section between text and data. */
    PROVIDE(data = .);

    /* The data segment */
    .data : {
        *(.data)
    }

    PROVIDE(edata = .);

    .bss : {
        *(.bss)
    }

    PROVIDE(end = .);

    /DISCARD/ : {
        *(.eh_frame .note.GNU-stack)
    }
}

Answer 1

看来我必须修改xv6引导程序并重写许多操作系统才能完成动态内核。