domain_build.c

xen虚拟机源代码安装包

    {
        l1_pgentry_t gdt_l1e;

        d->arch.is_32bit_pv = d->arch.has_32bit_shinfo = 1;
        v->vcpu_info = (void *)&d->shared_info->compat.vcpu_info[0];

        if ( nr_pages != (unsigned int)nr_pages )
            nr_pages = UINT_MAX;

        /*
         * Map compatibility Xen segments into every VCPU's GDT. See
         * arch_domain_create() for further comments.
         */
        gdt_l1e = l1e_from_page(virt_to_page(compat_gdt_table),
                                PAGE_HYPERVISOR);
        for ( i = 0; i < MAX_VIRT_CPUS; i++ )
            d->arch.mm_perdomain_pt[((i << GDT_LDT_VCPU_SHIFT) +
                                     FIRST_RESERVED_GDT_PAGE)] = gdt_l1e;
        flush_tlb_one_local(GDT_LDT_VIRT_START + FIRST_RESERVED_GDT_BYTE);
    }
#endif

    if ( parms.pae == PAEKERN_extended_cr3 )
            set_bit(VMASST_TYPE_pae_extended_cr3, &d->vm_assist);

    if ( (parms.virt_hv_start_low != UNSET_ADDR) && elf_32bit(&elf) )
    {
        unsigned long mask = (1UL << L2_PAGETABLE_SHIFT) - 1;
        value = (parms.virt_hv_start_low + mask) & ~mask;
        BUG_ON(!is_pv_32bit_domain(d));
#if defined(__i386__)
        if ( value > HYPERVISOR_VIRT_START )
            panic("Domain 0 expects too high a hypervisor start address.\n");
#else
        if ( value > __HYPERVISOR_COMPAT_VIRT_START )
            panic("Domain 0 expects too high a hypervisor start address.\n");
        HYPERVISOR_COMPAT_VIRT_START(d) =
            max_t(unsigned int, m2p_compat_vstart, value);
#endif
    }

    domain_set_alloc_bitsize(d);

    /*
     * Why do we need this? The number of page-table frames depends on the 
     * size of the bootstrap address space. But the size of the address space 
     * depends on the number of page-table frames (since each one is mapped 
     * read-only). We have a pair of simultaneous equations in two unknowns, 
     * which we solve by exhaustive search.
     */
    v_start          = parms.virt_base;
    vkern_start      = parms.virt_kstart;
    vkern_end        = parms.virt_kend;
    vinitrd_start    = round_pgup(vkern_end);
    vinitrd_end      = vinitrd_start + initrd_len;
    vphysmap_start   = round_pgup(vinitrd_end);
    vphysmap_end     = vphysmap_start + (nr_pages * (!is_pv_32on64_domain(d) ?
                                                     sizeof(unsigned long) :
                                                     sizeof(unsigned int)));
    vstartinfo_start = round_pgup(vphysmap_end);
    vstartinfo_end   = (vstartinfo_start +
                        sizeof(struct start_info) +
                        sizeof(struct dom0_vga_console_info));
    vpt_start        = round_pgup(vstartinfo_end);
    for ( nr_pt_pages = 2; ; nr_pt_pages++ )
    {
        vpt_end          = vpt_start + (nr_pt_pages * PAGE_SIZE);
        vstack_start     = vpt_end;
        vstack_end       = vstack_start + PAGE_SIZE;
        v_end            = (vstack_end + (1UL<<22)-1) & ~((1UL<<22)-1);
        if ( (v_end - vstack_end) < (512UL << 10) )
            v_end += 1UL << 22; /* Add extra 4MB to get >= 512kB padding. */
#if defined(__i386__)
        /* 5 pages: 1x 3rd + 4x 2nd level */
        if ( (((v_end - v_start + ((1UL<<L2_PAGETABLE_SHIFT)-1)) >>
               L2_PAGETABLE_SHIFT) + 5) <= nr_pt_pages )
            break;
#elif defined(__x86_64__)
#define NR(_l,_h,_s) \
    (((((_h) + ((1UL<<(_s))-1)) & ~((1UL<<(_s))-1)) - \
       ((_l) & ~((1UL<<(_s))-1))) >> (_s))
        if ( (1 + /* # L4 */
              NR(v_start, v_end, L4_PAGETABLE_SHIFT) + /* # L3 */
              (!is_pv_32on64_domain(d) ?
               NR(v_start, v_end, L3_PAGETABLE_SHIFT) : /* # L2 */
               4) + /* # compat L2 */
              NR(v_start, v_end, L2_PAGETABLE_SHIFT))  /* # L1 */
             <= nr_pt_pages )
            break;
#endif
    }

    order = get_order_from_bytes(v_end - v_start);
    if ( (1UL << order) > nr_pages )
        panic("Domain 0 allocation is too small for kernel image.\n");

#ifdef __i386__
    /* Ensure that our low-memory 1:1 mapping covers the allocation. */
    page = alloc_domheap_pages(d, order, MEMF_bits(30));
#else
    page = alloc_domheap_pages(d, order, 0);
#endif
    if ( page == NULL )
        panic("Not enough RAM for domain 0 allocation.\n");
    alloc_spfn = page_to_mfn(page);
    alloc_epfn = alloc_spfn + d->tot_pages;

    printk("PHYSICAL MEMORY ARRANGEMENT:\n"
           " Dom0 alloc.:   %"PRIpaddr"->%"PRIpaddr,
           pfn_to_paddr(alloc_spfn), pfn_to_paddr(alloc_epfn));
    if ( d->tot_pages < nr_pages )
        printk(" (%lu pages to be allocated)",
               nr_pages - d->tot_pages);
    printk("\nVIRTUAL MEMORY ARRANGEMENT:\n"
           " Loaded kernel: %p->%p\n"
           " Init. ramdisk: %p->%p\n"
           " Phys-Mach map: %p->%p\n"
           " Start info:    %p->%p\n"
           " Page tables:   %p->%p\n"
           " Boot stack:    %p->%p\n"
           " TOTAL:         %p->%p\n",
           _p(vkern_start), _p(vkern_end),
           _p(vinitrd_start), _p(vinitrd_end),
           _p(vphysmap_start), _p(vphysmap_end),
           _p(vstartinfo_start), _p(vstartinfo_end),
           _p(vpt_start), _p(vpt_end),
           _p(vstack_start), _p(vstack_end),
           _p(v_start), _p(v_end));
    printk(" ENTRY ADDRESS: %p\n", _p(parms.virt_entry));

    if ( ((v_end - v_start)>>PAGE_SHIFT) > nr_pages )
    {
        printk("Initial guest OS requires too much space\n"
               "(%luMB is greater than %luMB limit)\n",
               (v_end-v_start)>>20, nr_pages>>(20-PAGE_SHIFT));
        return -ENOMEM;
    }

    mpt_alloc = (vpt_start - v_start) +
        (unsigned long)pfn_to_paddr(alloc_spfn);

#if defined(__i386__)
    /*
     * Protect the lowest 1GB of memory. We use a temporary mapping there
     * from which we copy the kernel and ramdisk images.
     */
    if ( v_start < (1UL<<30) )
    {
        printk("Initial loading isn't allowed to lowest 1GB of memory.\n");
        return -EINVAL;
    }

    /* WARNING: The new domain must have its 'processor' field filled in! */
    l3start = l3tab = (l3_pgentry_t *)mpt_alloc; mpt_alloc += PAGE_SIZE;
    l2start = l2tab = (l2_pgentry_t *)mpt_alloc; mpt_alloc += 4*PAGE_SIZE;
    memcpy(l2tab, idle_pg_table_l2, 4*PAGE_SIZE);
    for (i = 0; i < 4; i++) {
        l3tab[i] = l3e_from_paddr((u32)l2tab + i*PAGE_SIZE, L3_PROT);
        l2tab[(LINEAR_PT_VIRT_START >> L2_PAGETABLE_SHIFT)+i] =
            l2e_from_paddr((u32)l2tab + i*PAGE_SIZE, __PAGE_HYPERVISOR);
    }
    v->arch.guest_table = pagetable_from_paddr((unsigned long)l3start);

    for ( i = 0; i < PDPT_L2_ENTRIES; i++ )
        l2tab[l2_linear_offset(PERDOMAIN_VIRT_START) + i] =
            l2e_from_page(virt_to_page(d->arch.mm_perdomain_pt) + i,
                          __PAGE_HYPERVISOR);

    l2tab += l2_linear_offset(v_start);
    mfn = alloc_spfn;
    for ( count = 0; count < ((v_end-v_start)>>PAGE_SHIFT); count++ )
    {
        if ( !((unsigned long)l1tab & (PAGE_SIZE-1)) )
        {
            l1start = l1tab = (l1_pgentry_t *)mpt_alloc;
            mpt_alloc += PAGE_SIZE;
            *l2tab = l2e_from_paddr((unsigned long)l1start, L2_PROT);
            l2tab++;
            clear_page(l1tab);
            if ( count == 0 )
                l1tab += l1_table_offset(v_start);
        }
        *l1tab = l1e_from_pfn(mfn, L1_PROT);
        l1tab++;
        
        page = mfn_to_page(mfn);
        if ( !get_page_and_type(page, d, PGT_writable_page) )
            BUG();

        mfn++;
    }

    /* Pages that are part of page tables must be read only. */
    l2tab = l2start + l2_linear_offset(vpt_start);
    l1start = l1tab = (l1_pgentry_t *)(u32)l2e_get_paddr(*l2tab);
    l1tab += l1_table_offset(vpt_start);
    for ( count = 0; count < nr_pt_pages; count++ ) 
    {
        page = mfn_to_page(l1e_get_pfn(*l1tab));
        if ( !opt_dom0_shadow )
            l1e_remove_flags(*l1tab, _PAGE_RW);
        else
            if ( !get_page_type(page, PGT_writable_page) )
                BUG();

        switch ( count )
        {
        case 0:
            page->u.inuse.type_info &= ~PGT_type_mask;
            page->u.inuse.type_info |= PGT_l3_page_table;
            get_page(page, d); /* an extra ref because of readable mapping */

            /* Get another ref to L3 page so that it can be pinned. */
            page->u.inuse.type_info++;
            page->count_info++;
            set_bit(_PGT_pinned, &page->u.inuse.type_info);
            break;
        case 1 ... 4:
            page->u.inuse.type_info &= ~PGT_type_mask;
            page->u.inuse.type_info |= PGT_l2_page_table;
            if ( count == 4 )
                page->u.inuse.type_info |= PGT_pae_xen_l2;
            get_page(page, d); /* an extra ref because of readable mapping */
            break;
        default:
            page->u.inuse.type_info &= ~PGT_type_mask;
            page->u.inuse.type_info |= PGT_l1_page_table;
            get_page(page, d); /* an extra ref because of readable mapping */
            break;
        }
        if ( !((unsigned long)++l1tab & (PAGE_SIZE - 1)) )
            l1start = l1tab = (l1_pgentry_t *)(u32)l2e_get_paddr(*++l2tab);
    }

#elif defined(__x86_64__)

    /* Overlap with Xen protected area? */
    if ( !is_pv_32on64_domain(d) ?
         ((v_start < HYPERVISOR_VIRT_END) &&
          (v_end > HYPERVISOR_VIRT_START)) :
         (v_end > HYPERVISOR_COMPAT_VIRT_START(d)) )
    {
        printk("DOM0 image overlaps with Xen private area.\n");
        return -EINVAL;
    }

    if ( is_pv_32on64_domain(d) )
    {
        v->arch.guest_context.failsafe_callback_cs = FLAT_COMPAT_KERNEL_CS;
        v->arch.guest_context.event_callback_cs    = FLAT_COMPAT_KERNEL_CS;
    }

    /* WARNING: The new domain must have its 'processor' field filled in! */
    if ( !is_pv_32on64_domain(d) )
    {
        maddr_to_page(mpt_alloc)->u.inuse.type_info = PGT_l4_page_table;
        l4start = l4tab = __va(mpt_alloc); mpt_alloc += PAGE_SIZE;
    }
    else
    {
        page = alloc_domheap_page(NULL, 0);
        if ( !page )
            panic("Not enough RAM for domain 0 PML4.\n");
        page->u.inuse.type_info = PGT_l4_page_table|PGT_validated|1;
        l4start = l4tab = page_to_virt(page);
    }
    copy_page(l4tab, idle_pg_table);
    l4tab[0] = l4e_empty(); /* zap trampoline mapping */
    l4tab[l4_table_offset(LINEAR_PT_VIRT_START)] =
        l4e_from_paddr(__pa(l4start), __PAGE_HYPERVISOR);
    l4tab[l4_table_offset(PERDOMAIN_VIRT_START)] =
        l4e_from_paddr(__pa(d->arch.mm_perdomain_l3), __PAGE_HYPERVISOR);
    v->arch.guest_table = pagetable_from_paddr(__pa(l4start));
    if ( is_pv_32on64_domain(d) )
        v->arch.guest_table_user = v->arch.guest_table;

    l4tab += l4_table_offset(v_start);
    mfn = alloc_spfn;
    for ( count = 0; count < ((v_end-v_start)>>PAGE_SHIFT); count++ )
    {
        if ( !((unsigned long)l1tab & (PAGE_SIZE-1)) )
        {
            maddr_to_page(mpt_alloc)->u.inuse.type_info = PGT_l1_page_table;
            l1start = l1tab = __va(mpt_alloc); mpt_alloc += PAGE_SIZE;
            clear_page(l1tab);
            if ( count == 0 )
                l1tab += l1_table_offset(v_start);
            if ( !((unsigned long)l2tab & (PAGE_SIZE-1)) )
            {
                maddr_to_page(mpt_alloc)->u.inuse.type_info = PGT_l2_page_table;
                l2start = l2tab = __va(mpt_alloc); mpt_alloc += PAGE_SIZE;
                clear_page(l2tab);
                if ( count == 0 )
                    l2tab += l2_table_offset(v_start);
                if ( !((unsigned long)l3tab & (PAGE_SIZE-1)) )
                {
                    maddr_to_page(mpt_alloc)->u.inuse.type_info =
                        PGT_l3_page_table;
                    l3start = l3tab = __va(mpt_alloc); mpt_alloc += PAGE_SIZE;
                    clear_page(l3tab);
                    if ( count == 0 )
                        l3tab += l3_table_offset(v_start);
                    *l4tab = l4e_from_paddr(__pa(l3start), L4_PROT);
                    l4tab++;
                }
                *l3tab = l3e_from_paddr(__pa(l2start), L3_PROT);
                l3tab++;
            }
            *l2tab = l2e_from_paddr(__pa(l1start), L2_PROT);
            l2tab++;
        }
        *l1tab = l1e_from_pfn(mfn, (!is_pv_32on64_domain(d) ?
                                    L1_PROT : COMPAT_L1_PROT));
        l1tab++;