int arch_setup_meminit(struct xc_dom_image *dom)
{
int rc;
xen_pfn_t pfn, allocsz, i;
dom->shadow_enabled = 1;
dom->p2m_host = xc_dom_malloc(dom, sizeof(xen_pfn_t) * dom->total_pages);
if ( dom->p2m_host == NULL )
return -EINVAL;
/* setup initial p2m */
for ( pfn = 0; pfn < dom->total_pages; pfn++ )
dom->p2m_host[pfn] = pfn + dom->rambase_pfn;
/* allocate guest memory */
for ( i = rc = allocsz = 0;
(i < dom->total_pages) && !rc;
i += allocsz )
{
allocsz = dom->total_pages - i;
if ( allocsz > 1024*1024 )
allocsz = 1024*1024;
rc = xc_domain_populate_physmap_exact(
dom->xch, dom->guest_domid, allocsz,
0, 0, &dom->p2m_host[i]);
}
return 0;
}
char *xc_dom_strdup(struct xc_dom_image *dom, const char *str)
{
size_t len = strlen(str) + 1;
char *nstr = xc_dom_malloc(dom, len);
if ( nstr == NULL )
return NULL;
memcpy(nstr, str, len);
return nstr;
}
static int xc_dom_parse_elf_kernel(struct xc_dom_image *dom)
{
struct elf_binary *elf;
int rc;
rc = check_elf_kernel(dom, 1);
if ( rc != 0 )
return rc;
elf = xc_dom_malloc(dom, sizeof(*elf));
dom->private_loader = elf;
rc = elf_init(elf, dom->kernel_blob, dom->kernel_size);
xc_elf_set_logfile(dom->xch, elf, 1);
if ( rc != 0 )
{
xc_dom_panic(dom->xch, XC_INVALID_KERNEL, "%s: corrupted ELF image",
__FUNCTION__);
return rc;
}
/* Find the section-header strings table. */
if ( elf->sec_strtab == NULL )
{
xc_dom_panic(dom->xch, XC_INVALID_KERNEL, "%s: ELF image"
" has no shstrtab", __FUNCTION__);
return -EINVAL;
}
/* parse binary and get xen meta info */
elf_parse_binary(elf);
if ( (rc = elf_xen_parse(elf, &dom->parms)) != 0 )
return rc;
if ( elf_xen_feature_get(XENFEAT_dom0, dom->parms.f_required) )
{
xc_dom_panic(dom->xch, XC_INVALID_KERNEL, "%s: Kernel does not"
" support unprivileged (DomU) operation", __FUNCTION__);
return -EINVAL;
}
/* find kernel segment */
dom->kernel_seg.vstart = dom->parms.virt_kstart;
dom->kernel_seg.vend = dom->parms.virt_kend;
if ( dom->parms.bsd_symtab )
xc_dom_load_elf_symtab(dom, elf, 0);
dom->guest_type = xc_dom_guest_type(dom, elf);
DOMPRINTF("%s: %s: 0x%" PRIx64 " -> 0x%" PRIx64 "",
__FUNCTION__, dom->guest_type,
dom->kernel_seg.vstart, dom->kernel_seg.vend);
return 0;
}
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;
}
int arch_setup_meminit(struct xc_dom_image *dom)
{
xen_pfn_t pfn;
int rc;
/* setup initial p2m */
dom->p2m_host = xc_dom_malloc(dom, sizeof(xen_pfn_t) * dom->total_pages);
for ( pfn = 0; pfn < dom->total_pages; pfn++ )
dom->p2m_host[pfn] = pfn;
/* allocate guest memory */
rc = xc_domain_memory_populate_physmap(dom->guest_xc, dom->guest_domid,
dom->total_pages, 0, 0,
dom->p2m_host);
return rc;
}
static int xc_dom_parse_elf_kernel(struct xc_dom_image *dom)
{
struct elf_binary *elf;
int rc;
rc = check_elf_kernel(dom, 1);
if ( rc != 0 )
return rc;
elf = xc_dom_malloc(dom, sizeof(*elf));
dom->private_loader = elf;
rc = elf_init(elf, dom->kernel_blob, dom->kernel_size);
if ( xc_dom_logfile )
elf_set_logfile(elf, xc_dom_logfile, 1);
if ( rc != 0 )
{
xc_dom_panic(XC_INVALID_KERNEL, "%s: corrupted ELF image\n",
__FUNCTION__);
return rc;
}
/* Find the section-header strings table. */
if ( elf->sec_strtab == NULL )
{
xc_dom_panic(XC_INVALID_KERNEL, "%s: ELF image has no shstrtab\n",
__FUNCTION__);
return -EINVAL;
}
/* parse binary and get xen meta info */
elf_parse_binary(elf);
if ( (rc = elf_xen_parse(elf, &dom->parms)) != 0 )
return rc;
/* find kernel segment */
dom->kernel_seg.vstart = dom->parms.virt_kstart;
dom->kernel_seg.vend = dom->parms.virt_kend;
if ( dom->parms.bsd_symtab )
xc_dom_load_elf_symtab(dom, elf, 0);
dom->guest_type = xc_dom_guest_type(dom, elf);
xc_dom_printf("%s: %s: 0x%" PRIx64 " -> 0x%" PRIx64 "\n",
__FUNCTION__, dom->guest_type,
dom->kernel_seg.vstart, dom->kernel_seg.vend);
return 0;
}
int xc_dom_try_gunzip(struct xc_dom_image *dom, void **blob, size_t * size)
{
void *unzip;
size_t unziplen;
unziplen = xc_dom_check_gzip(*blob, *size);
if ( unziplen == 0 )
return 0;
unzip = xc_dom_malloc(dom, unziplen);
if ( unzip == NULL )
return -1;
if ( xc_dom_do_gunzip(*blob, *size, unzip, unziplen) == -1 )
return -1;
*blob = unzip;
*size = unziplen;
return 0;
}
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 *xc_dom_pfn_to_ptr(struct xc_dom_image *dom, xen_pfn_t pfn,
xen_pfn_t count)
{
struct xc_dom_phys *phys;
unsigned int page_shift = XC_DOM_PAGE_SHIFT(dom);
char *mode = "unset";
if ( pfn > dom->total_pages )
{
xc_dom_printf("%s: pfn out of range (0x%" PRIpfn " > 0x%" PRIpfn ")\n",
__FUNCTION__, pfn, dom->total_pages);
return NULL;
}
/* already allocated? */
for ( phys = dom->phys_pages; phys != NULL; phys = phys->next )
{
if ( pfn >= (phys->first + phys->count) )
continue;
if ( count )
{
/* size given: must be completely within the already allocated block */
if ( (pfn + count) <= phys->first )
continue;
if ( (pfn < phys->first) ||
((pfn + count) > (phys->first + phys->count)) )
{
xc_dom_printf("%s: request overlaps allocated block"
" (req 0x%" PRIpfn "+0x%" PRIpfn ","
" blk 0x%" PRIpfn "+0x%" PRIpfn ")\n",
__FUNCTION__, pfn, count, phys->first,
phys->count);
return NULL;
}
}
else
{
/* no size given: block must be allocated already,
just hand out a pointer to it */
if ( pfn < phys->first )
continue;
}
return phys->ptr + ((pfn - phys->first) << page_shift);
}
/* allocating is allowed with size specified only */
if ( count == 0 )
{
xc_dom_printf("%s: no block found, no size given,"
" can't malloc (pfn 0x%" PRIpfn ")\n",
__FUNCTION__, pfn);
return NULL;
}
/* not found, no overlap => allocate */
phys = xc_dom_malloc(dom, sizeof(*phys));
if ( phys == NULL )
return NULL;
memset(phys, 0, sizeof(*phys));
phys->first = pfn;
phys->count = count;
if ( dom->guest_domid )
{
mode = "domU mapping";
phys->ptr = xc_dom_boot_domU_map(dom, phys->first, phys->count);
if ( phys->ptr == NULL )
return NULL;
dom->alloc_domU_map += phys->count << page_shift;
}
else
{
int err;
mode = "anonymous memory";
phys->ptr = mmap(NULL, phys->count << page_shift,
PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON,
-1, 0);
if ( phys->ptr == MAP_FAILED )
{
err = errno;
xc_dom_panic(XC_OUT_OF_MEMORY,
"%s: oom: can't allocate 0x%" PRIpfn " pages"
" [mmap, errno=%i (%s)]\n",
__FUNCTION__, count, err, strerror(err));
return NULL;
}
dom->alloc_mem_map += phys->count << page_shift;
}
#if 1
xc_dom_printf("%s: %s: pfn 0x%" PRIpfn "+0x%" PRIpfn " at %p\n",
__FUNCTION__, mode, phys->first, phys->count, phys->ptr);
#endif
phys->next = dom->phys_pages;
dom->phys_pages = phys;
return phys->ptr;
}
static int xc_dom_load_elf_symtab(struct xc_dom_image *dom,
struct elf_binary *elf, int load)
{
struct elf_binary syms;
const elf_shdr *shdr, *shdr2;
xen_vaddr_t symtab, maxaddr;
char *hdr;
size_t size;
int h, count, type, i, tables = 0;
if ( elf_swap(elf) )
{
xc_dom_printf("%s: non-native byte order, bsd symtab not supported\n",
__FUNCTION__);
return 0;
}
if ( load )
{
if ( !dom->bsd_symtab_start )
return 0;
size = dom->kernel_seg.vend - dom->bsd_symtab_start;
hdr = xc_dom_vaddr_to_ptr(dom, dom->bsd_symtab_start);
*(int *)hdr = size - sizeof(int);
}
else
{
size = sizeof(int) + elf_size(elf, elf->ehdr) +
elf_shdr_count(elf) * elf_size(elf, shdr);
hdr = xc_dom_malloc(dom, size);
if ( hdr == NULL )
return 0;
dom->bsd_symtab_start = elf_round_up(&syms, dom->kernel_seg.vend);
}
memcpy(hdr + sizeof(int),
elf->image,
elf_size(elf, elf->ehdr));
memcpy(hdr + sizeof(int) + elf_size(elf, elf->ehdr),
elf->image + elf_uval(elf, elf->ehdr, e_shoff),
elf_shdr_count(elf) * elf_size(elf, shdr));
if ( elf_64bit(elf) )
{
Elf64_Ehdr *ehdr = (Elf64_Ehdr *)(hdr + sizeof(int));
ehdr->e_phoff = 0;
ehdr->e_phentsize = 0;
ehdr->e_phnum = 0;
ehdr->e_shoff = elf_size(elf, elf->ehdr);
ehdr->e_shstrndx = SHN_UNDEF;
}
else
{
Elf32_Ehdr *ehdr = (Elf32_Ehdr *)(hdr + sizeof(int));
ehdr->e_phoff = 0;
ehdr->e_phentsize = 0;
ehdr->e_phnum = 0;
ehdr->e_shoff = elf_size(elf, elf->ehdr);
ehdr->e_shstrndx = SHN_UNDEF;
}
if ( elf_init(&syms, hdr + sizeof(int), size - sizeof(int)) )
return -1;
if ( xc_dom_logfile )
elf_set_logfile(&syms, xc_dom_logfile, 1);
symtab = dom->bsd_symtab_start + sizeof(int);
maxaddr = elf_round_up(&syms, symtab + elf_size(&syms, syms.ehdr) +
elf_shdr_count(&syms) * elf_size(&syms, shdr));
xc_dom_printf("%s/%s: bsd_symtab_start=%" PRIx64 ", kernel.end=0x%" PRIx64
" -- symtab=0x%" PRIx64 ", maxaddr=0x%" PRIx64 "\n",
__FUNCTION__, load ? "load" : "parse",
dom->bsd_symtab_start, dom->kernel_seg.vend,
symtab, maxaddr);
count = elf_shdr_count(&syms);
for ( h = 0; h < count; h++ )
{
shdr = elf_shdr_by_index(&syms, h);
type = elf_uval(&syms, shdr, sh_type);
if ( type == SHT_STRTAB )
{
/* Look for a strtab @i linked to symtab @h. */
for ( i = 0; i < count; i++ )
{
shdr2 = elf_shdr_by_index(&syms, i);
if ( (elf_uval(&syms, shdr2, sh_type) == SHT_SYMTAB) &&
(elf_uval(&syms, shdr2, sh_link) == h) )
break;
}
/* Skip symtab @h if we found no corresponding strtab @i. */
if ( i == count )
{
if ( elf_64bit(&syms) )
*(Elf64_Off*)(&shdr->e64.sh_offset) = 0;
else
*(Elf32_Off*)(&shdr->e32.sh_offset) = 0;
continue;
}
}
if ( (type == SHT_STRTAB) || (type == SHT_SYMTAB) )
{
/* Mangled to be based on ELF header location. */
if ( elf_64bit(&syms) )
*(Elf64_Off*)(&shdr->e64.sh_offset) = maxaddr - symtab;
else
*(Elf32_Off*)(&shdr->e32.sh_offset) = maxaddr - symtab;
size = elf_uval(&syms, shdr, sh_size);
maxaddr = elf_round_up(&syms, maxaddr + size);
tables++;
xc_dom_printf("%s: h=%d %s, size=0x%zx, maxaddr=0x%" PRIx64 "\n",
__FUNCTION__, h,
type == SHT_SYMTAB ? "symtab" : "strtab",
size, maxaddr);
if ( load )
{
shdr2 = elf_shdr_by_index(elf, h);
memcpy((void*)elf_section_start(&syms, shdr),
elf_section_start(elf, shdr2),
size);
}
}
/* Name is NULL. */
if ( elf_64bit(&syms) )
*(Elf64_Half*)(&shdr->e64.sh_name) = 0;
else
*(Elf32_Word*)(&shdr->e32.sh_name) = 0;
}
if ( tables == 0 )
{
xc_dom_printf("%s: no symbol table present\n", __FUNCTION__);
dom->bsd_symtab_start = 0;
return 0;
}
if ( !load )
dom->kernel_seg.vend = maxaddr;
return 0;
}
if ( hdr_ptr == NULL )
{
DOMPRINTF("%s: xc_dom_vaddr_to_ptr(dom,dom->bsd_symtab_start"
" => NULL", __FUNCTION__);
return -1;
}
elf->caller_xdest_base = hdr_ptr;
elf->caller_xdest_size = allow_size;
hdr = ELF_REALPTR2PTRVAL(hdr_ptr);
elf_store_val(elf, unsigned, hdr, size - sizeof(unsigned));
}
else
{
char *hdr_ptr;
hdr_ptr = xc_dom_malloc(dom, size);
if ( hdr_ptr == NULL )
return 0;
elf->caller_xdest_base = hdr_ptr;
elf->caller_xdest_size = size;
hdr = ELF_REALPTR2PTRVAL(hdr_ptr);
dom->bsd_symtab_start = elf_round_up(elf, dom->kernel_seg.vend);
dom->kernel_seg.vend = elf_round_up(elf, dom->bsd_symtab_start + size);
return 0;
}
elf_memcpy_safe(elf, hdr + sizeof(unsigned),
ELF_IMAGE_BASE(elf),
elf_size(elf, elf->ehdr));
elf_memcpy_safe(elf, hdr + sizeof(unsigned) + elf_size(elf, elf->ehdr),
ELF_IMAGE_BASE(elf) + elf_uval(elf, elf->ehdr, e_shoff),
int xc_dom_boot_image(struct xc_dom_image *dom)
{
DECLARE_DOMCTL;
void *ctxt;
int rc;
xc_dom_printf("%s: called\n", __FUNCTION__);
/* misc ia64 stuff*/
if ( (rc = arch_setup_bootearly(dom)) != 0 )
return rc;
/* collect some info */
domctl.cmd = XEN_DOMCTL_getdomaininfo;
domctl.domain = dom->guest_domid;
rc = do_domctl(dom->guest_xc, &domctl);
if ( rc != 0 )
{
xc_dom_panic(XC_INTERNAL_ERROR,
"%s: getdomaininfo failed (rc=%d)\n", __FUNCTION__, rc);
return rc;
}
if ( domctl.domain != dom->guest_domid )
{
xc_dom_panic(XC_INTERNAL_ERROR,
"%s: Huh? domid mismatch (%d != %d)\n", __FUNCTION__,
domctl.domain, dom->guest_domid);
return -1;
}
dom->shared_info_mfn = domctl.u.getdomaininfo.shared_info_frame;
/* sanity checks */
if ( !xc_dom_compat_check(dom) )
return -1;
/* initial mm setup */
if ( (rc = xc_dom_update_guest_p2m(dom)) != 0 )
return rc;
if ( dom->arch_hooks->setup_pgtables )
if ( (rc = dom->arch_hooks->setup_pgtables(dom)) != 0 )
return rc;
if ( (rc = clear_page(dom, dom->console_pfn)) != 0 )
return rc;
if ( (rc = clear_page(dom, dom->xenstore_pfn)) != 0 )
return rc;
/* start info page */
if ( dom->arch_hooks->start_info )
dom->arch_hooks->start_info(dom);
/* hypercall page */
if ( (rc = setup_hypercall_page(dom)) != 0 )
return rc;
xc_dom_log_memory_footprint(dom);
/* misc x86 stuff */
if ( (rc = arch_setup_bootlate(dom)) != 0 )
return rc;
/* let the vm run */
ctxt = xc_dom_malloc(dom, PAGE_SIZE * 2 /* FIXME */ );
memset(ctxt, 0, PAGE_SIZE * 2);
if ( (rc = dom->arch_hooks->vcpu(dom, ctxt)) != 0 )
return rc;
xc_dom_unmap_all(dom);
rc = launch_vm(dom->guest_xc, dom->guest_domid, ctxt);
return rc;
}
int arch_setup_meminit(struct xc_dom_image *dom)
{
int rc;
xen_pfn_t pfn, allocsz, i;
uint64_t modbase;
/* Convenient */
const uint64_t rambase = dom->rambase_pfn << XC_PAGE_SHIFT;
const uint64_t ramsize = dom->total_pages << XC_PAGE_SHIFT;
const uint64_t ramend = rambase + ramsize;
const uint64_t kernbase = dom->kernel_seg.vstart;
const uint64_t kernend = ROUNDUP(dom->kernel_seg.vend, 21/*2MB*/);
const uint64_t kernsize = kernend - kernbase;
const uint64_t dtb_size = dom->devicetree_blob ?
ROUNDUP(dom->devicetree_size, XC_PAGE_SHIFT) : 0;
const uint64_t ramdisk_size = dom->ramdisk_blob ?
ROUNDUP(dom->ramdisk_size, XC_PAGE_SHIFT) : 0;
const uint64_t modsize = dtb_size + ramdisk_size;
const uint64_t ram128mb = rambase + (128<<20);
if ( modsize + kernsize > ramsize )
{
DOMPRINTF("%s: Not enough memory for the kernel+dtb+initrd",
__FUNCTION__);
return -1;
}
if ( ramsize > GUEST_RAM_SIZE - NR_MAGIC_PAGES*XC_PAGE_SIZE )
{
DOMPRINTF("%s: ram size is too large for guest address space: "
"%"PRIx64" > %llx",
__FUNCTION__, ramsize,
GUEST_RAM_SIZE - NR_MAGIC_PAGES*XC_PAGE_SIZE);
return -1;
}
rc = set_mode(dom->xch, dom->guest_domid, dom->guest_type);
if ( rc )
return rc;
dom->shadow_enabled = 1;
dom->p2m_host = xc_dom_malloc(dom, sizeof(xen_pfn_t) * dom->total_pages);
if ( dom->p2m_host == NULL )
return -EINVAL;
/* setup initial p2m */
for ( pfn = 0; pfn < dom->total_pages; pfn++ )
dom->p2m_host[pfn] = pfn + dom->rambase_pfn;
/* allocate guest memory */
for ( i = rc = allocsz = 0;
(i < dom->total_pages) && !rc;
i += allocsz )
{
allocsz = dom->total_pages - i;
if ( allocsz > 1024*1024 )
allocsz = 1024*1024;
rc = xc_domain_populate_physmap_exact(
dom->xch, dom->guest_domid, allocsz,
0, 0, &dom->p2m_host[i]);
}
/*
* We try to place dtb+initrd at 128MB or if we have less RAM
* as high as possible. If there is no space then fallback to
* just before the kernel.
*
* If changing this then consider
* xen/arch/arm/kernel.c:place_modules as well.
*/
if ( ramend >= ram128mb + modsize && kernend < ram128mb )
modbase = ram128mb;
else if ( ramend - modsize > kernend )
modbase = ramend - modsize;
else if (kernbase - rambase > modsize )
modbase = kernbase - modsize;
else
return -1;
DOMPRINTF("%s: placing boot modules at 0x%" PRIx64, __FUNCTION__, modbase);
/*
* Must map DTB *after* initrd, to satisfy order of calls to
* xc_dom_alloc_segment in xc_dom_build_image, which must map
* things at monotonolically increasing addresses.
*/
if ( ramdisk_size )
{
dom->ramdisk_seg.vstart = modbase;
dom->ramdisk_seg.vend = modbase + ramdisk_size;
DOMPRINTF("%s: ramdisk: 0x%" PRIx64 " -> 0x%" PRIx64 "",
__FUNCTION__,
dom->ramdisk_seg.vstart, dom->ramdisk_seg.vend);
modbase += ramdisk_size;
}
if ( dtb_size )
{
dom->devicetree_seg.vstart = modbase;
dom->devicetree_seg.vend = modbase + dtb_size;
DOMPRINTF("%s: devicetree: 0x%" PRIx64 " -> 0x%" PRIx64 "",
__FUNCTION__,
dom->devicetree_seg.vstart, dom->devicetree_seg.vend);
modbase += dtb_size;
}
return 0;
}