/*
* Associate a virtual page frame with a given physical page frame
* and protection flags for that frame.
*/
static void set_pte_pfn_ma(unsigned long vaddr, unsigned long pfn,
pgprot_t flags)
{
pgd_t *pgd;
pud_t *pud;
pmd_t *pmd;
pte_t *pte;
pgd = swapper_pg_dir + pgd_index(vaddr);
if (pgd_none(*pgd)) {
BUG();
return;
}
pud = pud_offset(pgd, vaddr);
if (pud_none(*pud)) {
BUG();
return;
}
pmd = pmd_offset(pud, vaddr);
if (pmd_none(*pmd)) {
BUG();
return;
}
pte = pte_offset_kernel(pmd, vaddr);
/* <pfn,flags> stored as-is, to permit clearing entries */
set_pte(pte, pfn_pte_ma(pfn, flags));
/*
* It's enough to flush this one mapping.
* (PGE mappings get flushed as well)
*/
__flush_tlb_one(vaddr);
}
static int map_pte_fn(pte_t *pte, struct page *pmd_page,
unsigned long addr, void *data)
{
unsigned long **frames = (unsigned long **)data;
set_pte_at(&init_mm, addr, pte, pfn_pte_ma((*frames)[0], PAGE_KERNEL));
(*frames)++;
return 0;
}
static int increase_reservation(unsigned long nr_pages)
{
unsigned long pfn, i, flags;
struct page *page;
long rc;
struct xen_memory_reservation reservation = {
.address_bits = 0,
.extent_order = 0,
.domid = DOMID_SELF
};
if (nr_pages > ARRAY_SIZE(frame_list))
nr_pages = ARRAY_SIZE(frame_list);
balloon_lock(flags);
page = balloon_first_page();
for (i = 0; i < nr_pages; i++) {
BUG_ON(page == NULL);
frame_list[i] = page_to_pfn(page);;
page = balloon_next_page(page);
}
set_xen_guest_handle(reservation.extent_start, frame_list);
reservation.nr_extents = nr_pages;
rc = HYPERVISOR_memory_op(
XENMEM_populate_physmap, &reservation);
if (rc < nr_pages) {
int ret;
/* We hit the Xen hard limit: reprobe. */
set_xen_guest_handle(reservation.extent_start, frame_list);
reservation.nr_extents = rc;
ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
&reservation);
BUG_ON(ret != rc);
hard_limit = current_pages + rc - driver_pages;
goto out;
}
for (i = 0; i < nr_pages; i++) {
page = balloon_retrieve();
BUG_ON(page == NULL);
pfn = page_to_pfn(page);
BUG_ON(!xen_feature(XENFEAT_auto_translated_physmap) &&
phys_to_machine_mapping_valid(pfn));
/* Update P->M and M->P tables. */
set_phys_to_machine(pfn, frame_list[i]);
#ifdef CONFIG_XEN
xen_machphys_update(frame_list[i], pfn);
/* Link back into the page tables if not highmem. */
if (pfn < max_low_pfn) {
int ret;
ret = HYPERVISOR_update_va_mapping(
(unsigned long)__va(pfn << PAGE_SHIFT),
pfn_pte_ma(frame_list[i], PAGE_KERNEL),
0);
BUG_ON(ret);
}
#endif
/* Relinquish the page back to the allocator. */
ClearPageReserved(page);
set_page_count(page, 1);
__free_page(page);
}
current_pages += nr_pages;
totalram_pages = current_pages;
out:
balloon_unlock(flags);
return 0;
}
static int decrease_reservation(unsigned long nr_pages)
{
unsigned long pfn, i, flags;
struct page *page;
void *v;
int need_sleep = 0;
int ret;
struct xen_memory_reservation reservation = {
.address_bits = 0,
.extent_order = 0,
.domid = DOMID_SELF
};
if (nr_pages > ARRAY_SIZE(frame_list))
nr_pages = ARRAY_SIZE(frame_list);
for (i = 0; i < nr_pages; i++) {
if ((page = alloc_page(GFP_BALLOON)) == NULL) {
nr_pages = i;
need_sleep = 1;
break;
}
pfn = page_to_pfn(page);
frame_list[i] = pfn_to_mfn(pfn);
if (!PageHighMem(page)) {
v = phys_to_virt(pfn << PAGE_SHIFT);
scrub_pages(v, 1);
#ifdef CONFIG_XEN
ret = HYPERVISOR_update_va_mapping(
(unsigned long)v, __pte_ma(0), 0);
BUG_ON(ret);
#endif
}
#ifdef CONFIG_XEN_SCRUB_PAGES
else {
v = kmap(page);
scrub_pages(v, 1);
kunmap(page);
}
#endif
}
#ifdef CONFIG_XEN
/* Ensure that ballooned highmem pages don't have kmaps. */
kmap_flush_unused();
flush_tlb_all();
#endif
balloon_lock(flags);
/* No more mappings: invalidate P2M and add to balloon. */
for (i = 0; i < nr_pages; i++) {
pfn = mfn_to_pfn(frame_list[i]);
set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
balloon_append(pfn_to_page(pfn));
}
set_xen_guest_handle(reservation.extent_start, frame_list);
reservation.nr_extents = nr_pages;
ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
BUG_ON(ret != nr_pages);
current_pages -= nr_pages;
totalram_pages = current_pages;
balloon_unlock(flags);
return need_sleep;
}
/*
* We avoid multiple worker processes conflicting via the balloon mutex.
* We may of course race updates of the target counts (which are protected
* by the balloon lock), or with changes to the Xen hard limit, but we will
* recover from these in time.
*/
static void balloon_process(void *unused)
{
int need_sleep = 0;
long credit;
down(&balloon_mutex);
do {
credit = current_target() - current_pages;
if (credit > 0)
need_sleep = (increase_reservation(credit) != 0);
if (credit < 0)
need_sleep = (decrease_reservation(-credit) != 0);
#ifndef CONFIG_PREEMPT
if (need_resched())
schedule();
#endif
} while ((credit != 0) && !need_sleep);
/* Schedule more work if there is some still to be done. */
if (current_target() != current_pages)
mod_timer(&balloon_timer, jiffies + HZ);
up(&balloon_mutex);
}
/* Set up the grant operations for this fragment. If it's a flipping
interface, we also set up the unmap request from here. */
static u16 netbk_gop_frag(netif_t *netif, struct netbk_rx_meta *meta,
int i, struct netrx_pending_operations *npo,
struct page *page, unsigned long size,
unsigned long offset)
{
mmu_update_t *mmu;
gnttab_transfer_t *gop;
gnttab_copy_t *copy_gop;
multicall_entry_t *mcl;
netif_rx_request_t *req;
unsigned long old_mfn, new_mfn;
old_mfn = virt_to_mfn(page_address(page));
req = RING_GET_REQUEST(&netif->rx, netif->rx.req_cons + i);
if (netif->copying_receiver) {
/* The fragment needs to be copied rather than
flipped. */
meta->copy = 1;
copy_gop = npo->copy + npo->copy_prod++;
copy_gop->flags = GNTCOPY_dest_gref;
if (PageForeign(page)) {
struct pending_tx_info *src_pend =
&pending_tx_info[page->index];
copy_gop->source.domid = src_pend->netif->domid;
copy_gop->source.u.ref = src_pend->req.gref;
copy_gop->flags |= GNTCOPY_source_gref;
} else {
copy_gop->source.domid = DOMID_SELF;
copy_gop->source.u.gmfn = old_mfn;
}
copy_gop->source.offset = offset;
copy_gop->dest.domid = netif->domid;
copy_gop->dest.offset = 0;
copy_gop->dest.u.ref = req->gref;
copy_gop->len = size;
} else {
meta->copy = 0;
if (!xen_feature(XENFEAT_auto_translated_physmap)) {
new_mfn = alloc_mfn();
/*
* Set the new P2M table entry before
* reassigning the old data page. Heed the
* comment in pgtable-2level.h:pte_page(). :-)
*/
set_phys_to_machine(page_to_pfn(page), new_mfn);
mcl = npo->mcl + npo->mcl_prod++;
MULTI_update_va_mapping(mcl,
(unsigned long)page_address(page),
pfn_pte_ma(new_mfn, PAGE_KERNEL),
0);
mmu = npo->mmu + npo->mmu_prod++;
mmu->ptr = ((maddr_t)new_mfn << PAGE_SHIFT) |
MMU_MACHPHYS_UPDATE;
mmu->val = page_to_pfn(page);
}
gop = npo->trans + npo->trans_prod++;
gop->mfn = old_mfn;
gop->domid = netif->domid;
gop->ref = req->gref;
}
return req->id;
}
void __init adjust_boot_vcpu_info(void)
{
unsigned long lpfn, rpfn, lmfn, rmfn;
pte_t *lpte, *rpte;
unsigned int level;
mmu_update_t mmu[2];
/*
* setup_vcpu_info() cannot be used more than once for a given (v)CPU,
* hence we must swap the underlying MFNs of the two pages holding old
* and new vcpu_info of the boot CPU.
*
* Do *not* use __get_cpu_var() or percpu_{write,...}() here, as the per-
* CPU segment didn't get reloaded yet. Using percpu_read(), as in
* arch_use_lazy_mmu_mode(), though undesirable, is safe except for the
* accesses to variables that were updated in setup_percpu_areas().
*/
lpte = lookup_address((unsigned long)&per_cpu_var(vcpu_info)
+ (__per_cpu_load - __per_cpu_start),
&level);
rpte = lookup_address((unsigned long)&per_cpu(vcpu_info, 0), &level);
BUG_ON(!lpte || !(pte_flags(*lpte) & _PAGE_PRESENT));
BUG_ON(!rpte || !(pte_flags(*rpte) & _PAGE_PRESENT));
lmfn = __pte_mfn(*lpte);
rmfn = __pte_mfn(*rpte);
if (lmfn == rmfn)
return;
lpfn = mfn_to_local_pfn(lmfn);
rpfn = mfn_to_local_pfn(rmfn);
printk(KERN_INFO
"Swapping MFNs for PFN %lx and %lx (MFN %lx and %lx)\n",
lpfn, rpfn, lmfn, rmfn);
xen_l1_entry_update(lpte, pfn_pte_ma(rmfn, pte_pgprot(*lpte)));
xen_l1_entry_update(rpte, pfn_pte_ma(lmfn, pte_pgprot(*rpte)));
#ifdef CONFIG_X86_64
if (HYPERVISOR_update_va_mapping((unsigned long)__va(lpfn<<PAGE_SHIFT),
pfn_pte_ma(rmfn, PAGE_KERNEL_RO), 0))
BUG();
#endif
if (HYPERVISOR_update_va_mapping((unsigned long)__va(rpfn<<PAGE_SHIFT),
pfn_pte_ma(lmfn, PAGE_KERNEL),
UVMF_TLB_FLUSH))
BUG();
set_phys_to_machine(lpfn, rmfn);
set_phys_to_machine(rpfn, lmfn);
mmu[0].ptr = ((uint64_t)lmfn << PAGE_SHIFT) | MMU_MACHPHYS_UPDATE;
mmu[0].val = rpfn;
mmu[1].ptr = ((uint64_t)rmfn << PAGE_SHIFT) | MMU_MACHPHYS_UPDATE;
mmu[1].val = lpfn;
if (HYPERVISOR_mmu_update(mmu, 2, NULL, DOMID_SELF))
BUG();
/*
* Copy over all contents of the page just replaced, except for the
* vcpu_info itself, as it may have got updated after having been
* copied from __per_cpu_load[].
*/
memcpy(__va(rpfn << PAGE_SHIFT),
__va(lpfn << PAGE_SHIFT),
(unsigned long)&per_cpu_var(vcpu_info) & (PAGE_SIZE - 1));
level = (unsigned long)(&per_cpu_var(vcpu_info) + 1) & (PAGE_SIZE - 1);
if (level)
memcpy(__va(rpfn << PAGE_SHIFT) + level,
__va(lpfn << PAGE_SHIFT) + level,
PAGE_SIZE - level);
}