domain.c

来自「xen虚拟机源代码安装包」· C语言 代码 · 共 1,879 行 · 第 1/4 页

    case 'l': __arg = (unsigned long)va_arg(args, unsigned long); break;    \
    case 'h': __arg = (unsigned long)va_arg(args, void *);        break;    \
    default:  __arg = 0; BUG();                                             \
    }                                                                       \
    __arg;                                                                  \
})

DEFINE_PER_CPU(char, hc_preempted);

unsigned long hypercall_create_continuation(
    unsigned int op, const char *format, ...)
{
    struct mc_state *mcs = &this_cpu(mc_state);
    struct cpu_user_regs *regs;
    const char *p = format;
    unsigned long arg;
    unsigned int i;
    va_list args;

    va_start(args, format);

    if ( test_bit(_MCSF_in_multicall, &mcs->flags) )
    {
        __set_bit(_MCSF_call_preempted, &mcs->flags);

        for ( i = 0; *p != '\0'; i++ )
            mcs->call.args[i] = next_arg(p, args);
        if ( is_pv_32on64_domain(current->domain) )
        {
            for ( ; i < 6; i++ )
                mcs->call.args[i] = 0;
        }
    }
    else
    {
        regs       = guest_cpu_user_regs();
        regs->eax  = op;
        /*
         * For PV guest, we update EIP to re-execute 'syscall' / 'int 0x82';
         * HVM does not need this since 'vmcall' / 'vmmcall' is fault-like.
         */
        if ( !is_hvm_vcpu(current) )
            regs->eip -= 2;  /* re-execute 'syscall' / 'int 0x82' */

#ifdef __x86_64__
        if ( !is_hvm_vcpu(current) ?
             !is_pv_32on64_vcpu(current) :
             (hvm_guest_x86_mode(current) == 8) )
        {
            for ( i = 0; *p != '\0'; i++ )
            {
                arg = next_arg(p, args);
                switch ( i )
                {
                case 0: regs->rdi = arg; break;
                case 1: regs->rsi = arg; break;
                case 2: regs->rdx = arg; break;
                case 3: regs->r10 = arg; break;
                case 4: regs->r8  = arg; break;
                case 5: regs->r9  = arg; break;
                }
            }
        }
        else
#endif
        {
            if ( supervisor_mode_kernel )
                regs->eip &= ~31; /* re-execute entire hypercall entry stub */

            for ( i = 0; *p != '\0'; i++ )
            {
                arg = next_arg(p, args);
                switch ( i )
                {
                case 0: regs->ebx = arg; break;
                case 1: regs->ecx = arg; break;
                case 2: regs->edx = arg; break;
                case 3: regs->esi = arg; break;
                case 4: regs->edi = arg; break;
                case 5: regs->ebp = arg; break;
                }
            }
        }

        this_cpu(hc_preempted) = 1;
    }

    va_end(args);

    return op;
}

#ifdef CONFIG_COMPAT
int hypercall_xlat_continuation(unsigned int *id, unsigned int mask, ...)
{
    int rc = 0;
    struct mc_state *mcs = &this_cpu(mc_state);
    struct cpu_user_regs *regs;
    unsigned int i, cval = 0;
    unsigned long nval = 0;
    va_list args;

    BUG_ON(*id > 5);
    BUG_ON(mask & (1U << *id));

    va_start(args, mask);

    if ( test_bit(_MCSF_in_multicall, &mcs->flags) )
    {
        if ( !test_bit(_MCSF_call_preempted, &mcs->flags) )
            return 0;
        for ( i = 0; i < 6; ++i, mask >>= 1 )
        {
            if ( mask & 1 )
            {
                nval = va_arg(args, unsigned long);
                cval = va_arg(args, unsigned int);
                if ( cval == nval )
                    mask &= ~1U;
                else
                    BUG_ON(nval == (unsigned int)nval);
            }
            else if ( id && *id == i )
            {
                *id = mcs->call.args[i];
                id = NULL;
            }
            if ( (mask & 1) && mcs->call.args[i] == nval )
            {
                mcs->call.args[i] = cval;
                ++rc;
            }
            else
                BUG_ON(mcs->call.args[i] != (unsigned int)mcs->call.args[i]);
        }
    }
    else
    {
        regs = guest_cpu_user_regs();
        for ( i = 0; i < 6; ++i, mask >>= 1 )
        {
            unsigned long *reg;

            switch ( i )
            {
            case 0: reg = &regs->ebx; break;
            case 1: reg = &regs->ecx; break;
            case 2: reg = &regs->edx; break;
            case 3: reg = &regs->esi; break;
            case 4: reg = &regs->edi; break;
            case 5: reg = &regs->ebp; break;
            default: BUG(); reg = NULL; break;
            }
            if ( (mask & 1) )
            {
                nval = va_arg(args, unsigned long);
                cval = va_arg(args, unsigned int);
                if ( cval == nval )
                    mask &= ~1U;
                else
                    BUG_ON(nval == (unsigned int)nval);
            }
            else if ( id && *id == i )
            {
                *id = *reg;
                id = NULL;
            }
            if ( (mask & 1) && *reg == nval )
            {
                *reg = cval;
                ++rc;
            }
            else
                BUG_ON(*reg != (unsigned int)*reg);
        }
    }

    va_end(args);

    return rc;
}
#endif

static int relinquish_memory(
    struct domain *d, struct list_head *list, unsigned long type)
{
    struct list_head *ent;
    struct page_info  *page;
    unsigned long     x, y;
    int               ret = 0;

    /* Use a recursive lock, as we may enter 'free_domheap_page'. */
    spin_lock_recursive(&d->page_alloc_lock);

    ent = list->next;
    while ( ent != list )
    {
        page = list_entry(ent, struct page_info, list);

        /* Grab a reference to the page so it won't disappear from under us. */
        if ( unlikely(!get_page(page, d)) )
        {
            /* Couldn't get a reference -- someone is freeing this page. */
            ent = ent->next;
            list_move_tail(&page->list, &d->arch.relmem_list);
            continue;
        }

        if ( test_and_clear_bit(_PGT_pinned, &page->u.inuse.type_info) )
            put_page_and_type(page);

        if ( test_and_clear_bit(_PGC_allocated, &page->count_info) )
            put_page(page);

#ifdef DOMAIN_DESTRUCT_AVOID_RECURSION
        /*
         * Forcibly drop reference counts of page tables above top most (which
         * were skipped to prevent long latencies due to deep recursion - see
         * the special treatment in free_lX_table()).
         */
        y = page->u.inuse.type_info;
        if ( (type < PGT_root_page_table) &&
             unlikely(((y + PGT_type_mask) &
                       (PGT_type_mask|PGT_validated)) == type) )
        {
            BUG_ON((y & PGT_count_mask) >=
                   (page->count_info & PGC_count_mask));
            while ( y & PGT_count_mask )
            {
                put_page_and_type(page);
                y = page->u.inuse.type_info;
            }
        }
#endif

        /*
         * Forcibly invalidate top-most, still valid page tables at this point
         * to break circular 'linear page table' references. This is okay
         * because MMU structures are not shared across domains and this domain
         * is now dead. Thus top-most valid tables are not in use so a non-zero
         * count means circular reference.
         */
        y = page->u.inuse.type_info;
        for ( ; ; )
        {
            x = y;
            if ( likely((x & (PGT_type_mask|PGT_validated)) !=
                        (type|PGT_validated)) )
                break;

            y = cmpxchg(&page->u.inuse.type_info, x, x & ~PGT_validated);
            if ( likely(y == x) )
            {
                free_page_type(page, type);
                break;
            }
        }

        /* Follow the list chain and /then/ potentially free the page. */
        ent = ent->next;
        list_move_tail(&page->list, &d->arch.relmem_list);
        put_page(page);

        if ( hypercall_preempt_check() )
        {
            ret = -EAGAIN;
            goto out;
        }
    }

    list_splice_init(&d->arch.relmem_list, list);

 out:
    spin_unlock_recursive(&d->page_alloc_lock);
    return ret;
}

static void vcpu_destroy_pagetables(struct vcpu *v)
{
    struct domain *d = v->domain;
    unsigned long pfn;

#ifdef __x86_64__
    if ( is_pv_32on64_vcpu(v) )
    {
        pfn = l4e_get_pfn(*(l4_pgentry_t *)
                          __va(pagetable_get_paddr(v->arch.guest_table)));

        if ( pfn != 0 )
        {
            if ( paging_mode_refcounts(d) )
                put_page(mfn_to_page(pfn));
            else
                put_page_and_type(mfn_to_page(pfn));
        }

        l4e_write(
            (l4_pgentry_t *)__va(pagetable_get_paddr(v->arch.guest_table)),
            l4e_empty());

        v->arch.cr3 = 0;
        return;
    }
#endif

    pfn = pagetable_get_pfn(v->arch.guest_table);
    if ( pfn != 0 )
    {
        if ( paging_mode_refcounts(d) )
            put_page(mfn_to_page(pfn));
        else
            put_page_and_type(mfn_to_page(pfn));
        v->arch.guest_table = pagetable_null();
    }

#ifdef __x86_64__
    /* Drop ref to guest_table_user (from MMUEXT_NEW_USER_BASEPTR) */
    pfn = pagetable_get_pfn(v->arch.guest_table_user);
    if ( pfn != 0 )
    {
        if ( !is_pv_32bit_vcpu(v) )
        {
            if ( paging_mode_refcounts(d) )
                put_page(mfn_to_page(pfn));
            else
                put_page_and_type(mfn_to_page(pfn));
        }
        v->arch.guest_table_user = pagetable_null();
    }
#endif

    v->arch.cr3 = 0;
}

int domain_relinquish_resources(struct domain *d)
{
    int ret;
    struct vcpu *v;

    BUG_ON(!cpus_empty(d->domain_dirty_cpumask));

    switch ( d->arch.relmem )
    {
    case RELMEM_not_started:
        /* Tear down paging-assistance stuff. */
        paging_teardown(d);

        for_each_vcpu ( d, v )
        {
            /* Drop the in-use references to page-table bases. */
            vcpu_destroy_pagetables(v);

            /*
             * Relinquish GDT mappings. No need for explicit unmapping of the
             * LDT as it automatically gets squashed with the guest mappings.
             */
            destroy_gdt(v);

            unmap_vcpu_info(v);
        }

        d->arch.relmem = RELMEM_xen;
        /* fallthrough */

        /* Relinquish every page of memory. */
    case RELMEM_xen:
        ret = relinquish_memory(d, &d->xenpage_list, ~0UL);
        if ( ret )
            return ret;
#if CONFIG_PAGING_LEVELS >= 4
        d->arch.relmem = RELMEM_l4;
        /* fallthrough */

    case RELMEM_l4:
        ret = relinquish_memory(d, &d->page_list, PGT_l4_page_table);
        if ( ret )
            return ret;
#endif
#if CONFIG_PAGING_LEVELS >= 3
        d->arch.relmem = RELMEM_l3;
        /* fallthrough */

    case RELMEM_l3:
        ret = relinquish_memory(d, &d->page_list, PGT_l3_page_table);
        if ( ret )
            return ret;
#endif
        d->arch.relmem = RELMEM_l2;
        /* fallthrough */

    case RELMEM_l2:
        ret = relinquish_memory(d, &d->page_list, PGT_l2_page_table);
        if ( ret )
            return ret;
        d->arch.relmem = RELMEM_done;
        /* fallthrough */

    case RELMEM_done:
#ifdef DOMAIN_DESTRUCT_AVOID_RECURSION
        ret = relinquish_memory(d, &d->page_list, PGT_l1_page_table);
        if ( ret )
            return ret;
#endif
        break;

    default:
        BUG();
    }

    /* Free page used by xen oprofile buffer. */
    free_xenoprof_pages(d);

    if ( is_hvm_domain(d) )
        hvm_domain_relinquish_resources(d);

    return 0;
}

void arch_dump_domain_info(struct domain *d)
{
    paging_dump_domain_info(d);
}

void arch_dump_vcpu_info(struct vcpu *v)
{
    paging_dump_vcpu_info(v);
}

void domain_cpuid(
    struct domain *d,
    unsigned int  input,
    unsigned int  sub_input,
    unsigned int  *eax,
    unsigned int  *ebx,
    unsigned int  *ecx,
    unsigned int  *edx)
{
    cpuid_input_t *cpuid;
    int i;

    for ( i = 0; i < MAX_CPUID_INPUT; i++ )
    {
        cpuid = &d->arch.cpuids[i];

        if ( (cpuid->input[0] == input) &&
             ((cpuid->input[1] == XEN_CPUID_INPUT_UNUSED) ||
              (cpuid->input[1] == sub_input)) )
        {
            *eax = cpuid->eax;
            *ebx = cpuid->ebx;
            *ecx = cpuid->ecx;
            *edx = cpuid->edx;
            return;
        }
    }

    *eax = *ebx = *ecx = *edx = 0;
}

/*
 * Local variables:
 * mode: C
 * c-set-style: "BSD"
 * c-basic-offset: 4
 * tab-width: 4
 * indent-tabs-mode: nil
 * End:
 */