/****************************************************************************** * xc_linux_restore.c * * Restore the state of a Linux session. * * Copyright (c) 2003, K A Fraser. */ #include #include #include "xg_private.h" #include "xg_save_restore.h" /* max mfn of the whole machine */ static unsigned long max_mfn; /* virtual starting address of the hypervisor */ static unsigned long hvirt_start; /* #levels of page tables used by the currrent guest */ static unsigned int pt_levels; /* total number of pages used by the current guest */ static unsigned long max_pfn; /* Live mapping of the table mapping each PFN to its current MFN. */ static unsigned long *live_p2m = NULL; /* A table mapping each PFN to its new MFN. */ static unsigned long *p2m = NULL; static ssize_t read_exact(int fd, void *buf, size_t count) { int r = 0, s; unsigned char *b = buf; while (r < count) { s = read(fd, &b[r], count - r); if ((s == -1) && (errno == EINTR)) continue; if (s <= 0) { break; } r += s; } return (r == count) ? 1 : 0; } /* ** In the state file (or during transfer), all page-table pages are ** converted into a 'canonical' form where references to actual mfns ** are replaced with references to the corresponding pfns. ** This function inverts that operation, replacing the pfn values with ** the (now known) appropriate mfn values. */ int uncanonicalize_pagetable(unsigned long type, void *page) { int i, pte_last; unsigned long pfn; uint64_t pte; pte_last = PAGE_SIZE / ((pt_levels == 2)? 4 : 8); /* Now iterate through the page table, uncanonicalizing each PTE */ for(i = 0; i < pte_last; i++) { if(pt_levels == 2) pte = ((uint32_t *)page)[i]; else pte = ((uint64_t *)page)[i]; if(pte & _PAGE_PRESENT) { pfn = (pte >> PAGE_SHIFT) & 0xffffffff; if(pfn >= max_pfn) { /* This "page table page" is probably not one; bail. */ ERR("Frame number in type %lu page table is out of range: " "i=%d pfn=0x%lx max_pfn=%lu", type >> 28, i, pfn, max_pfn); return 0; } pte &= 0xffffff0000000fffULL; pte |= (uint64_t)p2m[pfn] << PAGE_SHIFT; if(pt_levels == 2) ((uint32_t *)page)[i] = (uint32_t)pte; else ((uint64_t *)page)[i] = (uint64_t)pte; } } return 1; } int xc_linux_restore(int xc_handle, int io_fd, uint32_t dom, unsigned long nr_pfns, unsigned int store_evtchn, unsigned long *store_mfn, unsigned int console_evtchn, unsigned long *console_mfn) { DECLARE_DOM0_OP; int rc = 1, i, n; unsigned long mfn, pfn; unsigned int prev_pc, this_pc; int verify = 0; int nraces = 0; /* The new domain's shared-info frame number. */ unsigned long shared_info_frame; unsigned char shared_info_page[PAGE_SIZE]; /* saved contents from file */ shared_info_t *shared_info = (shared_info_t *)shared_info_page; /* A copy of the CPU context of the guest. */ vcpu_guest_context_t ctxt; /* A table containing the type of each PFN (/not/ MFN!). */ unsigned long *pfn_type = NULL; /* A table of MFNs to map in the current region */ unsigned long *region_mfn = NULL; /* Types of the pfns in the current region */ unsigned long region_pfn_type[MAX_BATCH_SIZE]; /* A temporary mapping, and a copy, of one frame of guest memory. */ unsigned long *page = NULL; /* A copy of the pfn-to-mfn table frame list. */ unsigned long *p2m_frame_list = NULL; /* A temporary mapping of the guest's start_info page. */ start_info_t *start_info; char *region_base; xc_mmu_t *mmu = NULL; /* used by debug verify code */ unsigned long buf[PAGE_SIZE/sizeof(unsigned long)]; struct mmuext_op pin[MAX_PIN_BATCH]; unsigned int nr_pins; max_pfn = nr_pfns; DPRINTF("xc_linux_restore start: max_pfn = %lx\n", max_pfn); if(!get_platform_info(xc_handle, dom, &max_mfn, &hvirt_start, &pt_levels)) { ERR("Unable to get platform info."); return 1; } if (mlock(&ctxt, sizeof(ctxt))) { /* needed for build dom0 op, but might as well do early */ ERR("Unable to mlock ctxt"); return 1; } /* Read the saved P2M frame list */ if(!(p2m_frame_list = malloc(P2M_FL_SIZE))) { ERR("Couldn't allocate p2m_frame_list array"); goto out; } if (!read_exact(io_fd, p2m_frame_list, P2M_FL_SIZE)) { ERR("read p2m_frame_list failed"); goto out; } /* We want zeroed memory so use calloc rather than malloc. */ p2m = calloc(sizeof(unsigned long), max_pfn); pfn_type = calloc(sizeof(unsigned long), max_pfn); region_mfn = calloc(sizeof(unsigned long), MAX_BATCH_SIZE); if ((p2m == NULL) || (pfn_type == NULL) || (region_mfn == NULL)) { ERR("memory alloc failed"); errno = ENOMEM; goto out; } if (mlock(region_mfn, sizeof(unsigned long) * MAX_BATCH_SIZE)) { ERR("Could not mlock region_mfn"); goto out; } /* Get the domain's shared-info frame. */ op.cmd = DOM0_GETDOMAININFO; op.u.getdomaininfo.domain = (domid_t)dom; if (xc_dom0_op(xc_handle, &op) < 0) { ERR("Could not get information on new domain"); goto out; } shared_info_frame = op.u.getdomaininfo.shared_info_frame; if(xc_domain_setmaxmem(xc_handle, dom, PFN_TO_KB(max_pfn)) != 0) { errno = ENOMEM; goto out; } if(xc_domain_memory_increase_reservation( xc_handle, dom, max_pfn, 0, 0, NULL) != 0) { ERR("Failed to increase reservation by %lx KB", PFN_TO_KB(max_pfn)); errno = ENOMEM; goto out; } DPRINTF("Increased domain reservation by %lx KB\n", PFN_TO_KB(max_pfn)); /* Build the pfn-to-mfn table. We choose MFN ordering returned by Xen. */ if (xc_get_pfn_list(xc_handle, dom, p2m, max_pfn) != max_pfn) { ERR("Did not read correct number of frame numbers for new dom"); goto out; } if(!(mmu = xc_init_mmu_updates(xc_handle, dom))) { ERR("Could not initialise for MMU updates"); goto out; } DPRINTF("Reloading memory pages: 0%%\n"); /* * Now simply read each saved frame into its new machine frame. * We uncanonicalise page tables as we go. */ prev_pc = 0; n = 0; while (1) { int j; this_pc = (n * 100) / max_pfn; if ( (this_pc - prev_pc) >= 5 ) { PPRINTF("\b\b\b\b%3d%%", this_pc); prev_pc = this_pc; } if (!read_exact(io_fd, &j, sizeof(int))) { ERR("Error when reading batch size"); goto out; } PPRINTF("batch %d\n",j); if (j == -1) { verify = 1; fprintf(stderr, "Entering page verify mode\n"); continue; } if (j == 0) break; /* our work here is done */ if (j > MAX_BATCH_SIZE) { ERR("Max batch size exceeded. Giving up."); goto out; } if (!read_exact(io_fd, region_pfn_type, j*sizeof(unsigned long))) { ERR("Error when reading region pfn types"); goto out; } for (i = 0; i < j; i++) { if ((region_pfn_type[i] & LTAB_MASK) == XTAB) region_mfn[i] = 0; /* we know map will fail, but don't care */ else region_mfn[i] = p2m[region_pfn_type[i] & ~LTAB_MASK]; } if (!(region_base = xc_map_foreign_batch( xc_handle, dom, PROT_WRITE, region_mfn, j))) { ERR("map batch failed"); goto out; } for ( i = 0; i < j; i++ ) { void *page; unsigned long pagetype; pfn = region_pfn_type[i] & ~LTAB_MASK; pagetype = region_pfn_type[i] & LTAB_MASK; if (pagetype == XTAB) /* a bogus/unmapped page: skip it */ continue; if (pfn > max_pfn) { ERR("pfn out of range"); goto out; } pfn_type[pfn] = pagetype; mfn = p2m[pfn]; /* In verify mode, we use a copy; otherwise we work in place */ page = verify ? (void *)buf : (region_base + i*PAGE_SIZE); if (!read_exact(io_fd, page, PAGE_SIZE)) { ERR("Error when reading page (type was %lx)", pagetype); goto out; } pagetype &= LTABTYPE_MASK; if(pagetype >= L1TAB && pagetype <= L4TAB) { /* ** A page table page - need to 'uncanonicalize' it, i.e. ** replace all the references to pfns with the corresponding ** mfns for the new domain. ** ** On PAE we need to ensure that PGDs are in MFNs < 4G, and ** so we may need to update the p2m after the main loop. ** Hence we defer canonicalization of L1s until then. */ if(pt_levels != 3 || pagetype != L1TAB) { if(!uncanonicalize_pagetable(pagetype, page)) { /* ** Failing to uncanonicalize a page table can be ok ** under live migration since the pages type may have ** changed by now (and we'll get an update later). */ DPRINTF("PT L%ld race on pfn=%08lx mfn=%08lx\n", pagetype >> 28, pfn, mfn); nraces++; continue; } } } else if(pagetype != NOTAB) { ERR("Bogus page type %lx page table is out of range: " "i=%d max_pfn=%lu", pagetype, i, max_pfn); goto out; } if (verify) { int res = memcmp(buf, (region_base + i*PAGE_SIZE), PAGE_SIZE); if (res) { int v; DPRINTF("************** pfn=%lx type=%lx gotcs=%08lx " "actualcs=%08lx\n", pfn, pfn_type[pfn], csum_page(region_base + i*PAGE_SIZE), csum_page(buf)); for (v = 0; v < 4; v++) { unsigned long *p = (unsigned long *) (region_base + i*PAGE_SIZE); if (buf[v] != p[v]) DPRINTF(" %d: %08lx %08lx\n", v, buf[v], p[v]); } } } if (xc_add_mmu_update(xc_handle, mmu, (((unsigned long long)mfn) << PAGE_SHIFT) | MMU_MACHPHYS_UPDATE, pfn)) { ERR("failed machpys update mfn=%lx pfn=%lx", mfn, pfn); goto out; } } /* end of 'batch' for loop */ munmap(region_base, j*PAGE_SIZE); n+= j; /* crude stats */ } DPRINTF("Received all pages (%d races)\n", nraces); if(pt_levels == 3) { /* ** XXX SMH on PAE we need to ensure PGDs are in MFNs < 4G. This ** is a little awkward and involves (a) finding all such PGDs and ** replacing them with 'lowmem' versions; (b) upating the p2m[] ** with the new info; and (c) canonicalizing all the L1s using the ** (potentially updated) p2m[]. ** ** This is relatively slow (and currently involves two passes through ** the pfn_type[] array), but at least seems to be correct. May wish ** to consider more complex approaches to optimize this later. */ int j, k; /* First pass: find all L3TABs current in > 4G mfns and get new mfns */ for (i = 0; i < max_pfn; i++) { if (((pfn_type[i] & LTABTYPE_MASK)==L3TAB) && (p2m[i]>0xfffffUL)) { unsigned long new_mfn; uint64_t l3ptes[4]; uint64_t *l3tab; l3tab = (uint64_t *) xc_map_foreign_range(xc_handle, dom, PAGE_SIZE, PROT_READ, p2m[i]); for(j = 0; j < 4; j++) l3ptes[j] = l3tab[j]; munmap(l3tab, PAGE_SIZE); if (!(new_mfn=xc_make_page_below_4G(xc_handle, dom, p2m[i]))) { ERR("Couldn't get a page below 4GB :-("); goto out; } p2m[i] = new_mfn; if (xc_add_mmu_update(xc_handle, mmu, (((unsigned long long)new_mfn) << PAGE_SHIFT) | MMU_MACHPHYS_UPDATE, i)) { ERR("Couldn't m2p on PAE root pgdir"); goto out; } l3tab = (uint64_t *) xc_map_foreign_range(xc_handle, dom, PAGE_SIZE, PROT_READ | PROT_WRITE, p2m[i]); for(j = 0; j < 4; j++) l3tab[j] = l3ptes[j]; munmap(l3tab, PAGE_SIZE); } } /* Second pass: find all L1TABs and uncanonicalize them */ j = 0; for(i = 0; i < max_pfn; i++) { if (((pfn_type[i] & LTABTYPE_MASK)==L1TAB)) { region_mfn[j] = p2m[i]; j++; } if(i == (max_pfn-1) || j == MAX_BATCH_SIZE) { if (!(region_base = xc_map_foreign_batch( xc_handle, dom, PROT_READ | PROT_WRITE, region_mfn, j))) { ERR("map batch failed"); goto out; } for(k = 0; k < j; k++) { if(!uncanonicalize_pagetable(L1TAB, region_base + k*PAGE_SIZE)) { ERR("failed uncanonicalize pt!"); goto out; } } munmap(region_base, j*PAGE_SIZE); j = 0; } } } if (xc_finish_mmu_updates(xc_handle, mmu)) { ERR("Error doing finish_mmu_updates()"); goto out; } /* * Pin page tables. Do this after writing to them as otherwise Xen * will barf when doing the type-checking. */ nr_pins = 0; for (i = 0; i < max_pfn; i++) { if (i == (max_pfn-1) || nr_pins == MAX_PIN_BATCH) { if (xc_mmuext_op(xc_handle, pin, nr_pins, dom) < 0) { ERR("Failed to pin batch of %d page tables", nr_pins); goto out; } nr_pins = 0; } if ( (pfn_type[i] & LPINTAB) == 0 ) continue; switch(pfn_type[i]) { case (L1TAB|LPINTAB): pin[nr_pins].cmd = MMUEXT_PIN_L1_TABLE; break; case (L2TAB|LPINTAB): pin[nr_pins].cmd = MMUEXT_PIN_L2_TABLE; break; case (L3TAB|LPINTAB): pin[nr_pins].cmd = MMUEXT_PIN_L3_TABLE; break; case (L4TAB|LPINTAB): pin[nr_pins].cmd = MMUEXT_PIN_L4_TABLE; break; default: continue; } pin[nr_pins].arg1.mfn = p2m[i]; nr_pins++; } DPRINTF("\b\b\b\b100%%\n"); DPRINTF("Memory reloaded.\n"); /* Get the list of PFNs that are not in the psuedo-phys map */ { unsigned int count; unsigned long *pfntab; int rc; if (!read_exact(io_fd, &count, sizeof(count))) { ERR("Error when reading pfn count"); goto out; } if(!(pfntab = malloc(sizeof(unsigned long) * count))) { ERR("Out of memory"); goto out; } if (!read_exact(io_fd, pfntab, sizeof(unsigned long)*count)) { ERR("Error when reading pfntab"); goto out; } for (i = 0; i < count; i++) { unsigned long pfn = pfntab[i]; if(pfn > max_pfn) /* shouldn't happen - continue optimistically */ continue; pfntab[i] = p2m[pfn]; p2m[pfn] = INVALID_P2M_ENTRY; // not in pseudo-physical map } if (count > 0) { struct xen_memory_reservation reservation = { .extent_start = pfntab, .nr_extents = count, .extent_order = 0, .domid = dom }; if ((rc = xc_memory_op(xc_handle, XENMEM_decrease_reservation, &reservation)) != count) { ERR("Could not decrease reservation : %d", rc); goto out; } else DPRINTF("Decreased reservation by %d pages\n", count); } } if (!read_exact(io_fd, &ctxt, sizeof(ctxt)) || !read_exact(io_fd, shared_info_page, PAGE_SIZE)) { ERR("Error when reading ctxt or shared info page"); goto out; } /* Uncanonicalise the suspend-record frame number and poke resume rec. */ pfn = ctxt.user_regs.edx; if ((pfn >= max_pfn) || (pfn_type[pfn] != NOTAB)) { ERR("Suspend record frame number is bad"); goto out; } ctxt.user_regs.edx = mfn = p2m[pfn]; start_info = xc_map_foreign_range( xc_handle, dom, PAGE_SIZE, PROT_READ | PROT_WRITE, mfn); start_info->nr_pages = max_pfn; start_info->shared_info = shared_info_frame << PAGE_SHIFT; start_info->flags = 0; *store_mfn = start_info->store_mfn = p2m[start_info->store_mfn]; start_info->store_evtchn = store_evtchn; *console_mfn = start_info->console_mfn = p2m[start_info->console_mfn]; start_info->console_evtchn = console_evtchn; munmap(start_info, PAGE_SIZE); /* Uncanonicalise each GDT frame number. */ if (ctxt.gdt_ents > 8192) { ERR("GDT entry count out of range"); goto out; } for (i = 0; i < ctxt.gdt_ents; i += 512) { pfn = ctxt.gdt_frames[i]; if ((pfn >= max_pfn) || (pfn_type[pfn] != NOTAB)) { ERR("GDT frame number is bad"); goto out; } ctxt.gdt_frames[i] = p2m[pfn]; } /* Uncanonicalise the page table base pointer. */ pfn = ctxt.ctrlreg[3] >> PAGE_SHIFT; if (pfn >= max_pfn) { ERR("PT base is bad: pfn=%lu max_pfn=%lu type=%08lx", pfn, max_pfn, pfn_type[pfn]); goto out; } if ((pt_levels == 2) && ((pfn_type[pfn]<ABTYPE_MASK) != L2TAB)) { ERR("PT base is bad. pfn=%lu nr=%lu type=%08lx %08lx", pfn, max_pfn, pfn_type[pfn], (unsigned long)L2TAB); goto out; } if ((pt_levels == 3) && ((pfn_type[pfn]<ABTYPE_MASK) != L3TAB)) { ERR("PT base is bad. pfn=%lu nr=%lu type=%08lx %08lx", pfn, max_pfn, pfn_type[pfn], (unsigned long)L3TAB); goto out; } ctxt.ctrlreg[3] = p2m[pfn] << PAGE_SHIFT; /* clear any pending events and the selector */ memset(&(shared_info->evtchn_pending[0]), 0, sizeof (shared_info->evtchn_pending)); for ( i = 0; i < MAX_VIRT_CPUS; i++ ) shared_info->vcpu_info[i].evtchn_pending_sel = 0; /* Copy saved contents of shared-info page. No checking needed. */ page = xc_map_foreign_range( xc_handle, dom, PAGE_SIZE, PROT_WRITE, shared_info_frame); memcpy(page, shared_info, sizeof(shared_info_t)); munmap(page, PAGE_SIZE); /* Uncanonicalise the pfn-to-mfn table frame-number list. */ for (i = 0; i < P2M_FL_ENTRIES; i++) { pfn = p2m_frame_list[i]; if ((pfn >= max_pfn) || (pfn_type[pfn] != NOTAB)) { ERR("PFN-to-MFN frame number is bad"); goto out; } p2m_frame_list[i] = p2m[pfn]; } /* Copy the P2M we've constructed to the 'live' P2M */ if (!(live_p2m = xc_map_foreign_batch(xc_handle, dom, PROT_WRITE, p2m_frame_list, P2M_FL_ENTRIES))) { ERR("Couldn't map p2m table"); goto out; } memcpy(live_p2m, p2m, P2M_SIZE); munmap(live_p2m, P2M_SIZE); /* * Safety checking of saved context: * 1. user_regs is fine, as Xen checks that on context switch. * 2. fpu_ctxt is fine, as it can't hurt Xen. * 3. trap_ctxt needs the code selectors checked. * 4. ldt base must be page-aligned, no more than 8192 ents, ... * 5. gdt already done, and further checking is done by Xen. * 6. check that kernel_ss is safe. * 7. pt_base is already done. * 8. debugregs are checked by Xen. * 9. callback code selectors need checking. */ for ( i = 0; i < 256; i++ ) { ctxt.trap_ctxt[i].vector = i; if ((ctxt.trap_ctxt[i].cs & 3) == 0) ctxt.trap_ctxt[i].cs = FLAT_KERNEL_CS; } if ((ctxt.kernel_ss & 3) == 0) ctxt.kernel_ss = FLAT_KERNEL_DS; #if defined(__i386__) if ((ctxt.event_callback_cs & 3) == 0) ctxt.event_callback_cs = FLAT_KERNEL_CS; if ((ctxt.failsafe_callback_cs & 3) == 0) ctxt.failsafe_callback_cs = FLAT_KERNEL_CS; #endif if (((ctxt.ldt_base & (PAGE_SIZE - 1)) != 0) || (ctxt.ldt_ents > 8192) || (ctxt.ldt_base > hvirt_start) || ((ctxt.ldt_base + ctxt.ldt_ents*8) > hvirt_start)) { ERR("Bad LDT base or size"); goto out; } DPRINTF("Domain ready to be built.\n"); op.cmd = DOM0_SETVCPUCONTEXT; op.u.setvcpucontext.domain = (domid_t)dom; op.u.setvcpucontext.vcpu = 0; op.u.setvcpucontext.ctxt = &ctxt; rc = xc_dom0_op(xc_handle, &op); if (rc != 0) { ERR("Couldn't build the domain"); goto out; } out: if ( (rc != 0) && (dom != 0) ) xc_domain_destroy(xc_handle, dom); free(mmu); free(p2m); free(pfn_type); DPRINTF("Restore exit with rc=%d\n", rc); return rc; }