static int xc_linux_build_internal(struct xc_dom_image *dom,
xc_interface *xch, uint32_t domid,
unsigned int mem_mb,
unsigned long flags,
unsigned int store_evtchn,
unsigned long *store_mfn,
unsigned int console_evtchn,
unsigned long *console_mfn)
{
int rc;
dom->flags = flags;
dom->console_evtchn = console_evtchn;
dom->xenstore_evtchn = store_evtchn;
if ( (rc = xc_dom_boot_xen_init(dom, xch, domid)) != 0 )
goto out;
if ( (rc = xc_dom_parse_image(dom)) != 0 )
goto out;
if ( (rc = xc_dom_mem_init(dom, mem_mb)) != 0 )
goto out;
if ( (rc = xc_dom_boot_mem_init(dom)) != 0 )
goto out;
if ( (rc = xc_dom_build_image(dom)) != 0 )
goto out;
if ( (rc = xc_dom_boot_image(dom)) != 0 )
goto out;
if ( (rc = xc_dom_gnttab_init(dom)) != 0)
goto out;
*console_mfn = xc_dom_p2m_host(dom, dom->console_pfn);
*store_mfn = xc_dom_p2m_host(dom, dom->xenstore_pfn);
out:
return rc;
}
void *xc_dom_boot_domU_map(struct xc_dom_image *dom, xen_pfn_t pfn,
xen_pfn_t count)
{
int page_shift = XC_DOM_PAGE_SHIFT(dom);
privcmd_mmap_entry_t *entries;
void *ptr;
int i, rc;
int err;
entries = xc_dom_malloc(dom, count * sizeof(privcmd_mmap_entry_t));
if ( entries == NULL )
{
xc_dom_panic(XC_INTERNAL_ERROR,
"%s: failed to mmap domU pages 0x%" PRIpfn "+0x%" PRIpfn
" [malloc]\n", __FUNCTION__, pfn, count);
return NULL;
}
ptr = mmap(NULL, count << page_shift, PROT_READ | PROT_WRITE,
MAP_SHARED, dom->guest_xc, 0);
if ( ptr == MAP_FAILED )
{
err = errno;
xc_dom_panic(XC_INTERNAL_ERROR,
"%s: failed to mmap domU pages 0x%" PRIpfn "+0x%" PRIpfn
" [mmap, errno=%i (%s)]\n", __FUNCTION__, pfn, count,
err, strerror(err));
return NULL;
}
for ( i = 0; i < count; i++ )
{
entries[i].va = (uintptr_t) ptr + (i << page_shift);
entries[i].mfn = xc_dom_p2m_host(dom, pfn + i);
entries[i].npages = 1;
}
rc = xc_map_foreign_ranges(dom->guest_xc, dom->guest_domid,
entries, count);
if ( rc < 0 )
{
xc_dom_panic(XC_INTERNAL_ERROR,
"%s: failed to mmap domU pages 0x%" PRIpfn "+0x%" PRIpfn
" [xenctl, rc=%d]\n", __FUNCTION__, pfn, count, rc);
return NULL;
}
return ptr;
}
static int clear_page(struct xc_dom_image *dom, xen_pfn_t pfn)
{
xen_pfn_t dst;
int rc;
if ( pfn == 0 )
return 0;
dst = xc_dom_p2m_host(dom, pfn);
xc_dom_printf("%s: pfn 0x%" PRIpfn ", mfn 0x%" PRIpfn "\n",
__FUNCTION__, pfn, dst);
rc = xc_clear_domain_page(dom->guest_xc, dom->guest_domid, dst);
if ( rc != 0 )
xc_dom_panic(XC_INTERNAL_ERROR,
"%s: xc_clear_domain_page failed (pfn 0x%" PRIpfn
", rc=%d)\n", __FUNCTION__, pfn, rc);
return rc;
}
void *xc_dom_boot_domU_map(struct xc_dom_image *dom, xen_pfn_t pfn,
xen_pfn_t count)
{
int page_shift = XC_DOM_PAGE_SHIFT(dom);
privcmd_mmap_entry_t *entries;
void *ptr;
int i;
int err;
entries = xc_dom_malloc(dom, count * sizeof(privcmd_mmap_entry_t));
if ( entries == NULL )
{
xc_dom_panic(dom->xch, XC_INTERNAL_ERROR,
"%s: failed to mmap domU pages 0x%" PRIpfn "+0x%" PRIpfn
" [malloc]", __FUNCTION__, pfn, count);
return NULL;
}
for ( i = 0; i < count; i++ )
entries[i].mfn = xc_dom_p2m_host(dom, pfn + i);
ptr = xc_map_foreign_ranges(dom->xch, dom->guest_domid,
count << page_shift, PROT_READ | PROT_WRITE, 1 << page_shift,
entries, count);
if ( ptr == NULL )
{
err = errno;
xc_dom_panic(dom->xch, XC_INTERNAL_ERROR,
"%s: failed to mmap domU pages 0x%" PRIpfn "+0x%" PRIpfn
" [mmap, errno=%i (%s)]", __FUNCTION__, pfn, count,
err, strerror(err));
return NULL;
}
return ptr;
}
void kexec(void *kernel, long kernel_size, void *module, long module_size, char *cmdline, unsigned long flags)
{
struct xc_dom_image *dom;
int rc;
domid_t domid = DOMID_SELF;
xen_pfn_t pfn;
xc_interface *xc_handle;
unsigned long i;
void *seg;
xen_pfn_t boot_page_mfn = virt_to_mfn(&_boot_page);
char features[] = "";
struct mmu_update *m2p_updates;
unsigned long nr_m2p_updates;
DEBUG("booting with cmdline %s\n", cmdline);
xc_handle = xc_interface_open(0,0,0);
dom = xc_dom_allocate(xc_handle, cmdline, features);
dom->allocate = kexec_allocate;
/* We are using guest owned memory, therefore no limits. */
xc_dom_kernel_max_size(dom, 0);
xc_dom_ramdisk_max_size(dom, 0);
dom->kernel_blob = kernel;
dom->kernel_size = kernel_size;
dom->ramdisk_blob = module;
dom->ramdisk_size = module_size;
dom->flags = flags;
dom->console_evtchn = start_info.console.domU.evtchn;
dom->xenstore_evtchn = start_info.store_evtchn;
tpm_hash2pcr(dom, cmdline);
if ( (rc = xc_dom_boot_xen_init(dom, xc_handle, domid)) != 0 ) {
grub_printf("xc_dom_boot_xen_init returned %d\n", rc);
errnum = ERR_BOOT_FAILURE;
goto out;
}
if ( (rc = xc_dom_parse_image(dom)) != 0 ) {
grub_printf("xc_dom_parse_image returned %d\n", rc);
errnum = ERR_BOOT_FAILURE;
goto out;
}
#ifdef __i386__
if (strcmp(dom->guest_type, "xen-3.0-x86_32p")) {
grub_printf("can only boot x86 32 PAE kernels, not %s\n", dom->guest_type);
errnum = ERR_EXEC_FORMAT;
goto out;
}
#endif
#ifdef __x86_64__
if (strcmp(dom->guest_type, "xen-3.0-x86_64")) {
grub_printf("can only boot x86 64 kernels, not %s\n", dom->guest_type);
errnum = ERR_EXEC_FORMAT;
goto out;
}
#endif
/* equivalent of xc_dom_mem_init */
dom->arch_hooks = xc_dom_find_arch_hooks(xc_handle, dom->guest_type);
dom->total_pages = start_info.nr_pages;
/* equivalent of arch_setup_meminit */
/* setup initial p2m */
dom->p2m_host = malloc(sizeof(*dom->p2m_host) * dom->total_pages);
/* Start with our current P2M */
for (i = 0; i < dom->total_pages; i++)
dom->p2m_host[i] = pfn_to_mfn(i);
if ( (rc = xc_dom_build_image(dom)) != 0 ) {
grub_printf("xc_dom_build_image returned %d\n", rc);
errnum = ERR_BOOT_FAILURE;
goto out;
}
/* copy hypercall page */
/* TODO: domctl instead, but requires privileges */
if (dom->parms.virt_hypercall != -1) {
pfn = PHYS_PFN(dom->parms.virt_hypercall - dom->parms.virt_base);
memcpy((void *) pages[pfn], hypercall_page, PAGE_SIZE);
}
/* Equivalent of xc_dom_boot_image */
dom->shared_info_mfn = PHYS_PFN(start_info.shared_info);
if (!xc_dom_compat_check(dom)) {
grub_printf("xc_dom_compat_check failed\n");
errnum = ERR_EXEC_FORMAT;
goto out;
}
/* Move current console, xenstore and boot MFNs to the allocated place */
do_exchange(dom, dom->console_pfn, start_info.console.domU.mfn);
do_exchange(dom, dom->xenstore_pfn, start_info.store_mfn);
DEBUG("virt base at %llx\n", dom->parms.virt_base);
DEBUG("bootstack_pfn %lx\n", dom->bootstack_pfn);
_boot_target = dom->parms.virt_base + PFN_PHYS(dom->bootstack_pfn);
DEBUG("_boot_target %lx\n", _boot_target);
do_exchange(dom, PHYS_PFN(_boot_target - dom->parms.virt_base),
virt_to_mfn(&_boot_page));
/* Make sure the bootstrap page table does not RW-map any of our current
* page table frames */
kexec_allocate(dom, dom->virt_pgtab_end);
if ( (rc = xc_dom_update_guest_p2m(dom))) {
grub_printf("xc_dom_update_guest_p2m returned %d\n", rc);
errnum = ERR_BOOT_FAILURE;
goto out;
}
if ( dom->arch_hooks->setup_pgtables )
if ( (rc = dom->arch_hooks->setup_pgtables(dom))) {
grub_printf("setup_pgtables returned %d\n", rc);
errnum = ERR_BOOT_FAILURE;
goto out;
}
/* start info page */
#undef start_info
if ( dom->arch_hooks->start_info )
dom->arch_hooks->start_info(dom);
#define start_info (start_info_union.start_info)
xc_dom_log_memory_footprint(dom);
/* Unmap libxc's projection of the boot page table */
seg = xc_dom_seg_to_ptr(dom, &dom->pgtables_seg);
munmap(seg, dom->pgtables_seg.vend - dom->pgtables_seg.vstart);
/* Unmap day0 pages to avoid having a r/w mapping of the future page table */
for (pfn = 0; pfn < allocated; pfn++)
munmap((void*) pages[pfn], PAGE_SIZE);
/* Pin the boot page table base */
if ( (rc = pin_table(dom->xch,
#ifdef __i386__
MMUEXT_PIN_L3_TABLE,
#endif
#ifdef __x86_64__
MMUEXT_PIN_L4_TABLE,
#endif
xc_dom_p2m_host(dom, dom->pgtables_seg.pfn),
dom->guest_domid)) != 0 ) {
grub_printf("pin_table(%lx) returned %d\n", xc_dom_p2m_host(dom,
dom->pgtables_seg.pfn), rc);
errnum = ERR_BOOT_FAILURE;
goto out_remap;
}
/* We populate the Mini-OS page table here so that boot.S can just call
* update_va_mapping to project itself there. */
need_pgt(_boot_target);
DEBUG("day0 pages %lx\n", allocated);
DEBUG("boot target page %lx\n", _boot_target);
DEBUG("boot page %p\n", &_boot_page);
DEBUG("boot page mfn %lx\n", boot_page_mfn);
_boot_page_entry = PFN_PHYS(boot_page_mfn) | L1_PROT;
DEBUG("boot page entry %llx\n", _boot_page_entry);
_boot_oldpdmfn = virt_to_mfn(start_info.pt_base);
DEBUG("boot old pd mfn %lx\n", _boot_oldpdmfn);
DEBUG("boot pd virt %lx\n", dom->pgtables_seg.vstart);
_boot_pdmfn = dom->p2m_host[PHYS_PFN(dom->pgtables_seg.vstart - dom->parms.virt_base)];
DEBUG("boot pd mfn %lx\n", _boot_pdmfn);
_boot_stack = _boot_target + PAGE_SIZE;
DEBUG("boot stack %lx\n", _boot_stack);
_boot_start_info = dom->parms.virt_base + PFN_PHYS(dom->start_info_pfn);
DEBUG("boot start info %lx\n", _boot_start_info);
_boot_start = dom->parms.virt_entry;
DEBUG("boot start %lx\n", _boot_start);
/* Keep only useful entries */
for (nr_m2p_updates = pfn = 0; pfn < start_info.nr_pages; pfn++)
if (dom->p2m_host[pfn] != pfn_to_mfn(pfn))
nr_m2p_updates++;
m2p_updates = malloc(sizeof(*m2p_updates) * nr_m2p_updates);
for (i = pfn = 0; pfn < start_info.nr_pages; pfn++)
if (dom->p2m_host[pfn] != pfn_to_mfn(pfn)) {
m2p_updates[i].ptr = PFN_PHYS(dom->p2m_host[pfn]) | MMU_MACHPHYS_UPDATE;
m2p_updates[i].val = pfn;
i++;
}
for (i = 0; i < blk_nb; i++)
shutdown_blkfront(blk_dev[i]);
if (net_dev)
shutdown_netfront(net_dev);
if (kbd_dev)
shutdown_kbdfront(kbd_dev);
stop_kernel();
/* Update M2P */
if ((rc = HYPERVISOR_mmu_update(m2p_updates, nr_m2p_updates, NULL, DOMID_SELF)) < 0) {
xprintk("Could not update M2P\n");
ASSERT(0);
}
xprintk("go!\n");
/* Jump to trampoline boot page */
_boot();
ASSERT(0);
out_remap:
for (pfn = 0; pfn < allocated; pfn++)
do_map_frames(pages[pfn], &pages_mfns[pfn], 1, 0, 0, DOMID_SELF, 0, L1_PROT);
out:
xc_dom_release(dom);
for (pfn = 0; pfn < allocated; pfn++)
free_page((void*)pages[pfn]);
free(pages);
free(pages_mfns);
pages = NULL;
pages_mfns = NULL;
allocated = 0;
xc_interface_close(xc_handle );
}
int xc_dom_gnttab_hvm_seed(xc_interface *xch, domid_t domid,
xen_pfn_t console_gpfn,
xen_pfn_t xenstore_gpfn,
domid_t console_domid,
domid_t xenstore_domid)
{
int rc;
xen_pfn_t scratch_gpfn;
struct xen_add_to_physmap xatp = {
.domid = domid,
.space = XENMAPSPACE_grant_table,
.idx = 0,
};
struct xen_remove_from_physmap xrfp = {
.domid = domid,
};
rc = xc_core_arch_get_scratch_gpfn(xch, domid, &scratch_gpfn);
if ( rc < 0 )
{
xc_dom_panic(xch, XC_INTERNAL_ERROR,
"%s: failed to get a scratch gfn "
"[errno=%d]\n",
__FUNCTION__, errno);
return -1;
}
xatp.gpfn = scratch_gpfn;
xrfp.gpfn = scratch_gpfn;
xc_dom_printf(xch, "%s: called, pfn=0x%"PRI_xen_pfn, __FUNCTION__,
scratch_gpfn);
rc = do_memory_op(xch, XENMEM_add_to_physmap, &xatp, sizeof(xatp));
if ( rc != 0 )
{
xc_dom_panic(xch, XC_INTERNAL_ERROR,
"%s: failed to add gnttab to physmap "
"[errno=%d]\n",
__FUNCTION__, errno);
return -1;
}
rc = xc_dom_gnttab_seed(xch, domid,
console_gpfn, xenstore_gpfn,
console_domid, xenstore_domid);
if (rc != 0)
{
xc_dom_panic(xch, XC_INTERNAL_ERROR,
"%s: failed to seed gnttab entries\n",
__FUNCTION__);
(void) do_memory_op(xch, XENMEM_remove_from_physmap, &xrfp, sizeof(xrfp));
return -1;
}
rc = do_memory_op(xch, XENMEM_remove_from_physmap, &xrfp, sizeof(xrfp));
if (rc != 0)
{
xc_dom_panic(xch, XC_INTERNAL_ERROR,
"%s: failed to remove gnttab from physmap "
"[errno=%d]\n",
__FUNCTION__, errno);
return -1;
}
return 0;
}
int xc_dom_gnttab_init(struct xc_dom_image *dom)
{
if ( xc_dom_feature_translated(dom) ) {
return xc_dom_gnttab_hvm_seed(dom->xch, dom->guest_domid,
dom->console_pfn, dom->xenstore_pfn,
dom->console_domid, dom->xenstore_domid);
} else {
return xc_dom_gnttab_seed(dom->xch, dom->guest_domid,
xc_dom_p2m_host(dom, dom->console_pfn),
xc_dom_p2m_host(dom, dom->xenstore_pfn),
dom->console_domid, dom->xenstore_domid);
}
}
