static ssize_t
pipe_writev(struct file *filp, const struct iovec *_iov,
unsigned long nr_segs, loff_t *ppos)
{
struct inode *inode = filp->f_dentry->d_inode;
struct pipe_inode_info *info;
ssize_t ret;
int do_wakeup;
struct iovec *iov = (struct iovec *)_iov;
size_t total_len;
ssize_t chars;
total_len = iov_length(iov, nr_segs);
/* Null write succeeds. */
if (unlikely(total_len == 0))
return 0;
do_wakeup = 0;
ret = 0;
down(PIPE_SEM(*inode));
info = inode->i_pipe;
if (!PIPE_READERS(*inode)) {
send_sig(SIGPIPE, current, 0);
ret = -EPIPE;
goto out;
}
/* We try to merge small writes */
chars = total_len & (PAGE_SIZE-1); /* size of the last buffer */
if (info->nrbufs && chars != 0) {
int lastbuf = (info->curbuf + info->nrbufs - 1) & (PIPE_BUFFERS-1);
struct pipe_buffer *buf = info->bufs + lastbuf;
struct pipe_buf_operations *ops = buf->ops;
int offset = buf->offset + buf->len;
if (ops->can_merge && offset + chars <= PAGE_SIZE) {
void *addr = ops->map(filp, info, buf);
int error = pipe_iov_copy_from_user(offset + addr, iov, chars);
ops->unmap(info, buf);
ret = error;
do_wakeup = 1;
if (error)
goto out;
buf->len += chars;
total_len -= chars;
ret = chars;
if (!total_len)
goto out;
}
}
for (;;) {
int bufs;
if (!PIPE_READERS(*inode)) {
send_sig(SIGPIPE, current, 0);
if (!ret) ret = -EPIPE;
break;
}
bufs = info->nrbufs;
if (bufs < PIPE_BUFFERS) {
int newbuf = (info->curbuf + bufs) & (PIPE_BUFFERS-1);
struct pipe_buffer *buf = info->bufs + newbuf;
struct page *page = info->tmp_page;
int error;
if (!page) {
page = alloc_page(GFP_HIGHUSER);
if (unlikely(!page)) {
ret = ret ? : -ENOMEM;
break;
}
info->tmp_page = page;
}
// check_swap - check the correctness of swap & page replacement algorithm
static void
check_swap(void) {
size_t nr_free_pages_store = nr_free_pages();
size_t slab_allocated_store = slab_allocated();
size_t offset;
for (offset = 2; offset < max_swap_offset; offset ++) {
mem_map[offset] = 1;
}
struct mm_struct *mm = mm_create();
assert(mm != NULL);
extern struct mm_struct *check_mm_struct;
assert(check_mm_struct == NULL);
check_mm_struct = mm;
pgd_t *pgdir = mm->pgdir = boot_pgdir;
assert(pgdir[0] == 0);
struct vma_struct *vma = vma_create(0, PTSIZE, VM_WRITE | VM_READ);
assert(vma != NULL);
insert_vma_struct(mm, vma);
struct Page *rp0 = alloc_page(), *rp1 = alloc_page();
assert(rp0 != NULL && rp1 != NULL);
uint32_t perm = PTE_U | PTE_W;
int ret = page_insert(pgdir, rp1, 0, perm);
assert(ret == 0 && page_ref(rp1) == 1);
page_ref_inc(rp1);
ret = page_insert(pgdir, rp0, 0, perm);
assert(ret == 0 && page_ref(rp1) == 1 && page_ref(rp0) == 1);
// check try_alloc_swap_entry
swap_entry_t entry = try_alloc_swap_entry();
assert(swap_offset(entry) == 1);
mem_map[1] = 1;
assert(try_alloc_swap_entry() == 0);
// set rp1, Swap, Active, add to hash_list, active_list
swap_page_add(rp1, entry);
swap_active_list_add(rp1);
assert(PageSwap(rp1));
mem_map[1] = 0;
entry = try_alloc_swap_entry();
assert(swap_offset(entry) == 1);
assert(!PageSwap(rp1));
// check swap_remove_entry
assert(swap_hash_find(entry) == NULL);
mem_map[1] = 2;
swap_remove_entry(entry);
assert(mem_map[1] == 1);
swap_page_add(rp1, entry);
swap_inactive_list_add(rp1);
swap_remove_entry(entry);
assert(PageSwap(rp1));
assert(rp1->index == entry && mem_map[1] == 0);
// check page_launder, move page from inactive_list to active_list
assert(page_ref(rp1) == 1);
assert(nr_active_pages == 0 && nr_inactive_pages == 1);
assert(list_next(&(inactive_list.swap_list)) == &(rp1->swap_link));
page_launder();
assert(nr_active_pages == 1 && nr_inactive_pages == 0);
assert(PageSwap(rp1) && PageActive(rp1));
entry = try_alloc_swap_entry();
assert(swap_offset(entry) == 1);
assert(!PageSwap(rp1) && nr_active_pages == 0);
assert(list_empty(&(active_list.swap_list)));
// set rp1 inactive again
assert(page_ref(rp1) == 1);
swap_page_add(rp1, 0);
assert(PageSwap(rp1) && swap_offset(rp1->index) == 1);
swap_inactive_list_add(rp1);
mem_map[1] = 1;
assert(nr_inactive_pages == 1);
page_ref_dec(rp1);
size_t count = nr_free_pages();
swap_remove_entry(entry);
assert(nr_inactive_pages == 0 && nr_free_pages() == count + 1);
// check swap_out_mm
pte_t *ptep0 = get_pte(pgdir, 0, 0), *ptep1;
assert(ptep0 != NULL && pte2page(*ptep0) == rp0);
ret = swap_out_mm(mm, 0);
assert(ret == 0);
ret = swap_out_mm(mm, 10);
assert(ret == 1 && mm->swap_address == PGSIZE);
ret = swap_out_mm(mm, 10);
assert(ret == 0 && *ptep0 == entry && mem_map[1] == 1);
assert(PageDirty(rp0) && PageActive(rp0) && page_ref(rp0) == 0);
assert(nr_active_pages == 1 && list_next(&(active_list.swap_list)) == &(rp0->swap_link));
// check refill_inactive_scan()
refill_inactive_scan();
assert(!PageActive(rp0) && page_ref(rp0) == 0);
assert(nr_inactive_pages == 1 && list_next(&(inactive_list.swap_list)) == &(rp0->swap_link));
page_ref_inc(rp0);
page_launder();
assert(PageActive(rp0) && page_ref(rp0) == 1);
assert(nr_active_pages == 1 && list_next(&(active_list.swap_list)) == &(rp0->swap_link));
page_ref_dec(rp0);
refill_inactive_scan();
assert(!PageActive(rp0));
// save data in rp0
int i;
for (i = 0; i < PGSIZE; i ++) {
((char *)page2kva(rp0))[i] = (char)i;
}
page_launder();
assert(nr_inactive_pages == 0 && list_empty(&(inactive_list.swap_list)));
assert(mem_map[1] == 1);
rp1 = alloc_page();
assert(rp1 != NULL);
ret = swapfs_read(entry, rp1);
assert(ret == 0);
for (i = 0; i < PGSIZE; i ++) {
assert(((char *)page2kva(rp1))[i] == (char)i);
}
// page fault now
*(char *)0 = 0xEF;
rp0 = pte2page(*ptep0);
assert(page_ref(rp0) == 1);
assert(PageSwap(rp0) && PageActive(rp0));
entry = try_alloc_swap_entry();
assert(swap_offset(entry) == 1 && mem_map[1] == SWAP_UNUSED);
assert(!PageSwap(rp0) && nr_active_pages == 0 && nr_inactive_pages == 0);
// clear accessed flag
assert(rp0 == pte2page(*ptep0));
assert(!PageSwap(rp0));
ret = swap_out_mm(mm, 10);
assert(ret == 0);
assert(!PageSwap(rp0) && (*ptep0 & PTE_P));
// change page table
ret = swap_out_mm(mm, 10);
assert(ret == 1);
assert(*ptep0 == entry && page_ref(rp0) == 0 && mem_map[1] == 1);
count = nr_free_pages();
refill_inactive_scan();
page_launder();
assert(count + 1 == nr_free_pages());
ret = swapfs_read(entry, rp1);
assert(ret == 0 && *(char *)(page2kva(rp1)) == (char)0xEF);
free_page(rp1);
// duplictate *ptep0
ptep1 = get_pte(pgdir, PGSIZE, 0);
assert(ptep1 != NULL && *ptep1 == 0);
swap_duplicate(*ptep0);
*ptep1 = *ptep0;
// page fault again
// update for copy on write
*(char *)1 = 0x88;
*(char *)(PGSIZE) = 0x8F;
*(char *)(PGSIZE + 1) = 0xFF;
assert(pte2page(*ptep0) != pte2page(*ptep1));
assert(*(char *)0 == (char)0xEF);
assert(*(char *)1 == (char)0x88);
assert(*(char *)(PGSIZE) == (char)0x8F);
assert(*(char *)(PGSIZE + 1) == (char)0xFF);
rp0 = pte2page(*ptep0);
rp1 = pte2page(*ptep1);
assert(!PageSwap(rp0) && PageSwap(rp1) && PageActive(rp1));
entry = try_alloc_swap_entry();
assert(!PageSwap(rp0) && !PageSwap(rp1));
assert(swap_offset(entry) == 1 && mem_map[1] == SWAP_UNUSED);
assert(list_empty(&(active_list.swap_list)));
assert(list_empty(&(inactive_list.swap_list)));
page_insert(pgdir, rp0, PGSIZE, perm | PTE_A);
// check swap_out_mm
*(char *)0 = *(char *)PGSIZE = 0xEE;
mm->swap_address = PGSIZE * 2;
ret = swap_out_mm(mm, 2);
assert(ret == 0);
assert((*ptep0 & PTE_P) && !(*ptep0 & PTE_A));
assert((*ptep1 & PTE_P) && !(*ptep1 & PTE_A));
ret = swap_out_mm(mm, 2);
assert(ret == 2);
assert(mem_map[1] == 2 && page_ref(rp0) == 0);
refill_inactive_scan();
page_launder();
assert(mem_map[1] == 2 && swap_hash_find(entry) == NULL);
// check copy entry
swap_remove_entry(entry);
*ptep1 = 0;
assert(mem_map[1] == 1);
swap_entry_t store;
ret = swap_copy_entry(entry, &store);
assert(ret == -E_NO_MEM);
mem_map[2] = SWAP_UNUSED;
ret = swap_copy_entry(entry, &store);
assert(ret == 0 && swap_offset(store) == 2 && mem_map[2] == 0);
mem_map[2] = 1;
*ptep1 = store;
assert(*(char *)PGSIZE == (char)0xEE && *(char *)(PGSIZE + 1)== (char)0x88);
*(char *)PGSIZE = 1, *(char *)(PGSIZE + 1) = 2;
assert(*(char *)0 == (char)0xEE && *(char *)1 == (char)0x88);
ret = swap_in_page(entry, &rp0);
assert(ret == 0);
ret = swap_in_page(store, &rp1);
assert(ret == 0);
assert(rp1 != rp0);
// free memory
swap_list_del(rp0), swap_list_del(rp1);
swap_page_del(rp0), swap_page_del(rp1);
assert(page_ref(rp0) == 1 && page_ref(rp1) == 1);
assert(nr_active_pages == 0 && list_empty(&(active_list.swap_list)));
assert(nr_inactive_pages == 0 && list_empty(&(inactive_list.swap_list)));
for (i = 0; i < HASH_LIST_SIZE; i ++) {
assert(list_empty(hash_list + i));
}
page_remove(pgdir, 0);
page_remove(pgdir, PGSIZE);
free_page(pa2page(PMD_ADDR(*get_pmd(pgdir, 0, 0))));
free_page(pa2page(PUD_ADDR(*get_pud(pgdir, 0, 0))));
free_page(pa2page(PGD_ADDR(*get_pgd(pgdir, 0, 0))));
pgdir[0] = 0;
mm->pgdir = NULL;
mm_destroy(mm);
check_mm_struct = NULL;
assert(nr_active_pages == 0 && nr_inactive_pages == 0);
for (offset = 0; offset < max_swap_offset; offset ++) {
mem_map[offset] = SWAP_UNUSED;
}
assert(nr_free_pages_store == nr_free_pages());
assert(slab_allocated_store == slab_allocated());
cprintf("check_swap() succeeded.\n");
}
static void xennet_alloc_rx_buffers(struct net_device *dev)
{
unsigned short id;
struct netfront_info *np = netdev_priv(dev);
struct sk_buff *skb;
struct page *page;
int i, batch_target, notify;
RING_IDX req_prod = np->rx.req_prod_pvt;
grant_ref_t ref;
unsigned long pfn;
void *vaddr;
struct xen_netif_rx_request *req;
if (unlikely(!netif_carrier_ok(dev)))
return;
/*
* Allocate skbuffs greedily, even though we batch updates to the
* receive ring. This creates a less bursty demand on the memory
* allocator, so should reduce the chance of failed allocation requests
* both for ourself and for other kernel subsystems.
*/
batch_target = np->rx_target - (req_prod - np->rx.rsp_cons);
for (i = skb_queue_len(&np->rx_batch); i < batch_target; i++) {
skb = __netdev_alloc_skb(dev, RX_COPY_THRESHOLD + NET_IP_ALIGN,
GFP_ATOMIC | __GFP_NOWARN);
if (unlikely(!skb))
goto no_skb;
/* Align ip header to a 16 bytes boundary */
skb_reserve(skb, NET_IP_ALIGN);
page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
if (!page) {
kfree_skb(skb);
no_skb:
/* Any skbuffs queued for refill? Force them out. */
if (i != 0)
goto refill;
/* Could not allocate any skbuffs. Try again later. */
mod_timer(&np->rx_refill_timer,
jiffies + (HZ/10));
break;
}
__skb_fill_page_desc(skb, 0, page, 0, 0);
skb_shinfo(skb)->nr_frags = 1;
__skb_queue_tail(&np->rx_batch, skb);
}
/* Is the batch large enough to be worthwhile? */
if (i < (np->rx_target/2)) {
if (req_prod > np->rx.sring->req_prod)
goto push;
return;
}
/* Adjust our fill target if we risked running out of buffers. */
if (((req_prod - np->rx.sring->rsp_prod) < (np->rx_target / 4)) &&
((np->rx_target *= 2) > np->rx_max_target))
np->rx_target = np->rx_max_target;
refill:
for (i = 0; ; i++) {
skb = __skb_dequeue(&np->rx_batch);
if (skb == NULL)
break;
skb->dev = dev;
id = xennet_rxidx(req_prod + i);
BUG_ON(np->rx_skbs[id]);
np->rx_skbs[id] = skb;
ref = gnttab_claim_grant_reference(&np->gref_rx_head);
BUG_ON((signed short)ref < 0);
np->grant_rx_ref[id] = ref;
pfn = page_to_pfn(skb_frag_page(&skb_shinfo(skb)->frags[0]));
vaddr = page_address(skb_frag_page(&skb_shinfo(skb)->frags[0]));
req = RING_GET_REQUEST(&np->rx, req_prod + i);
gnttab_grant_foreign_access_ref(ref,
np->xbdev->otherend_id,
pfn_to_mfn(pfn),
0);
req->id = id;
req->gref = ref;
}
wmb(); /* barrier so backend seens requests */
/* Above is a suitable barrier to ensure backend will see requests. */
np->rx.req_prod_pvt = req_prod + i;
push:
RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&np->rx, notify);
if (notify)
notify_remote_via_irq(np->netdev->irq);
}
/*
* do_no_page() tries to create a new page mapping. It aggressively
* tries to share with existing pages, but makes a separate copy if
* the "write_access" parameter is true in order to avoid the next
* page fault.
*
* As this is called only for pages that do not currently exist, we
* do not need to flush old virtual caches or the TLB.
*
* This is called with the MM semaphore held and the page table
* spinlock held. Exit with the spinlock released.
*/
static int do_no_page(struct mm_struct * mm, struct vm_area_struct * vma,
unsigned long address, int write_access, pte_t *page_table)
{
struct page * new_page;
pte_t entry;
if (!vma->vm_ops || !vma->vm_ops->nopage)
return do_anonymous_page(mm, vma, page_table, write_access, address);
spin_unlock(&mm->page_table_lock);
new_page = vma->vm_ops->nopage(vma, address & PAGE_MASK, 0);
if (new_page == NULL) /* no page was available -- SIGBUS */
return 0;
if (new_page == NOPAGE_OOM)
return -1;
/*
* Should we do an early C-O-W break?
*/
if (write_access && !(vma->vm_flags & VM_SHARED)) {
struct page * page = alloc_page(GFP_HIGHUSER);
if (!page) {
page_cache_release(new_page);
return -1;
}
copy_user_highpage(page, new_page, address);
page_cache_release(new_page);
lru_cache_add(page);
new_page = page;
}
spin_lock(&mm->page_table_lock);
/*
* This silly early PAGE_DIRTY setting removes a race
* due to the bad i386 page protection. But it's valid
* for other architectures too.
*
* Note that if write_access is true, we either now have
* an exclusive copy of the page, or this is a shared mapping,
* so we can make it writable and dirty to avoid having to
* handle that later.
*/
/* Only go through if we didn't race with anybody else... */
if (pte_none(*page_table)) {
++mm->rss;
flush_page_to_ram(new_page);
flush_icache_page(vma, new_page);
entry = mk_pte(new_page, vma->vm_page_prot);
if (write_access)
entry = pte_mkwrite(pte_mkdirty(entry));
set_pte(page_table, entry);
} else {
/* One of our sibling threads was faster, back out. */
page_cache_release(new_page);
spin_unlock(&mm->page_table_lock);
return 1;
}
/* no need to invalidate: a not-present page shouldn't be cached */
update_mmu_cache(vma, address, entry);
spin_unlock(&mm->page_table_lock);
return 2; /* Major fault */
}
};
static uint syscall_max = sizeof(syscall_vectors) / sizeof(*syscall_vectors);
void syscall_handler(regs_t* registers)
{
if(registers->eax >= syscall_max)
{
kprintf("[#%d] Unknown syscall %x\n", task_current()->pid, registers->eax);
return;
}
syscall_vectors[registers->eax](registers);
}
void sys_alloc_page(__attribute__((unused)) regs_t* registers)
{
uint page = alloc_page();
task_t* t = task_current();
uint d_ent = 0;
uint* tbl = 0;
uint dir;
for(dir = 0x10000000 / 4096 / 1024; dir < 0xffffffff / 4096 / 1024; dir++)
{
if(t->page_directory[dir] & 1)
d_ent = t->page_directory[dir] & 0xfffff000;
else
continue;
}
tbl = (uint*)d_ent;
uint tbl_i;
for(tbl_i = 0; tbl_i < 1024; tbl_i++)
{
struct netfront_dev *init_netfront(char *_nodename, void (*thenetif_rx)(unsigned char* data, int len), unsigned char rawmac[6], char **ip)
{
xenbus_transaction_t xbt;
char* err;
char* message=NULL;
struct netif_tx_sring *txs;
struct netif_rx_sring *rxs;
int retry=0;
int i;
char* msg = NULL;
char nodename[256];
char path[256];
struct netfront_dev *dev;
if (!_nodename)
snprintf(nodename, sizeof(nodename), "device/vif/%d", netfrontends);
else {
strncpy(nodename, _nodename, sizeof(nodename) - 1);
nodename[sizeof(nodename) - 1] = 0;
}
netfrontends++;
if (!thenetif_rx)
thenetif_rx = netif_rx;
#ifdef CONFIG_FRONT
printk("************************ NETFRONT for %s **********\n\n", nodename);
#endif
dev = malloc(sizeof(*dev));
memset(dev, 0, sizeof(*dev));
dev->nodename = strdup(nodename);
#ifdef HAVE_LIBC
dev->fd = -1;
#endif
#ifdef CONFIG_FRONT
printk("net TX ring size %d\n", NET_TX_RING_SIZE);
printk("net RX ring size %d\n", NET_RX_RING_SIZE);
#endif
init_SEMAPHORE(&dev->tx_sem, NET_TX_RING_SIZE);
for(i=0;i<NET_TX_RING_SIZE;i++)
{
add_id_to_freelist(i,dev->tx_freelist);
dev->tx_buffers[i].page = NULL;
}
for(i=0;i<NET_RX_RING_SIZE;i++)
{
/* TODO: that's a lot of memory */
dev->rx_buffers[i].page = (char*)alloc_page();
}
snprintf(path, sizeof(path), "%s/backend-id", nodename);
dev->dom = xenbus_read_integer(path);
#ifdef HAVE_LIBC
if (thenetif_rx == NETIF_SELECT_RX)
evtchn_alloc_unbound(dev->dom, netfront_select_handler, dev, &dev->evtchn);
else
#endif
evtchn_alloc_unbound(dev->dom, netfront_handler, dev, &dev->evtchn);
txs = (struct netif_tx_sring *) alloc_page();
rxs = (struct netif_rx_sring *) alloc_page();
memset(txs,0,PAGE_SIZE);
memset(rxs,0,PAGE_SIZE);
SHARED_RING_INIT(txs);
SHARED_RING_INIT(rxs);
FRONT_RING_INIT(&dev->tx, txs, PAGE_SIZE);
FRONT_RING_INIT(&dev->rx, rxs, PAGE_SIZE);
dev->tx_ring_ref = gnttab_grant_access(dev->dom,virt_to_mfn(txs),0);
dev->rx_ring_ref = gnttab_grant_access(dev->dom,virt_to_mfn(rxs),0);
init_rx_buffers(dev);
dev->netif_rx = thenetif_rx;
dev->events = NULL;
again:
err = xenbus_transaction_start(&xbt);
if (err) {
printk("starting transaction\n");
free(err);
}
err = xenbus_printf(xbt, nodename, "tx-ring-ref","%u",
dev->tx_ring_ref);
//printk("node = %s\n", nodename);
if (err) {
message = "writing tx ring-ref";
goto abort_transaction;
}
err = xenbus_printf(xbt, nodename, "rx-ring-ref","%u",
dev->rx_ring_ref);
if (err) {
message = "writing rx ring-ref";
goto abort_transaction;
}
err = xenbus_printf(xbt, nodename,
"event-channel", "%u", dev->evtchn);
if (err) {
message = "writing event-channel";
goto abort_transaction;
}
err = xenbus_printf(xbt, nodename, "request-rx-copy", "%u", 1);
if (err) {
message = "writing request-rx-copy";
goto abort_transaction;
}
snprintf(path, sizeof(path), "%s/state", nodename);
err = xenbus_switch_state(xbt, path, XenbusStateConnected);
if (err) {
message = "switching state";
goto abort_transaction;
}
err = xenbus_transaction_end(xbt, 0, &retry);
free(err);
if (retry) {
goto again;
printk("completing transaction\n");
}
goto done;
abort_transaction:
free(err);
err = xenbus_transaction_end(xbt, 1, &retry);
printk("Abort transaction %s\n", message);
goto error;
done:
snprintf(path, sizeof(path), "%s/backend", nodename);
msg = xenbus_read(XBT_NIL, path, &dev->backend);
snprintf(path, sizeof(path), "%s/mac", nodename);
msg = xenbus_read(XBT_NIL, path, &dev->mac);
if ((dev->backend == NULL) || (dev->mac == NULL)) {
printk("%s: backend/mac failed\n", __func__);
goto error;
}
#ifdef CONFIG_FRONT
printk("backend at %s\n",dev->backend);
printk("mac is %s\n",dev->mac);
#endif
{
XenbusState state;
char path[strlen(dev->backend) + strlen("/state") + 1];
snprintf(path, sizeof(path), "%s/state", dev->backend);
xenbus_watch_path_token(XBT_NIL, path, path, &dev->events);
err = NULL;
state = xenbus_read_integer(path);
while (err == NULL && state < XenbusStateConnected)
err = xenbus_wait_for_state_change(path, &state, &dev->events);
if (state != XenbusStateConnected) {
printk("backend not avalable, state=%d\n", state);
xenbus_unwatch_path_token(XBT_NIL, path, path);
goto error;
}
if (ip) {
snprintf(path, sizeof(path), "%s/ip", dev->backend);
xenbus_read(XBT_NIL, path, ip);
}
}
#ifdef CONFIG_FRONT
printk("**************************\n");
#endif
unmask_evtchn(dev->evtchn);
/* Special conversion specifier 'hh' needed for __ia64__. Without
this mini-os panics with 'Unaligned reference'. */
if (rawmac){
sscanf(dev->mac,"%hhx:%hhx:%hhx:%hhx:%hhx:%hhx",
&rawmac[0],&rawmac[1],&rawmac[2],
&rawmac[3],&rawmac[4],&rawmac[5]);
/*printf("MAC:%02x:%02x:%02x:%02x:%02x:%02x\n",
rawmac[0],rawmac[1],rawmac[2],
rawmac[3],rawmac[4],rawmac[5]);*/
}
return dev;
error:
free(msg);
free(err);
free_netfront(dev);
return NULL;
}
/**
* nfs_follow_referral - set up mountpoint when hitting a referral on moved error
* @dentry - parent directory
* @locations - array of NFSv4 server location information
*
*/
static struct vfsmount *nfs_follow_referral(struct dentry *dentry,
const struct nfs4_fs_locations *locations)
{
struct vfsmount *mnt = ERR_PTR(-ENOENT);
struct nfs_clone_mount mountdata = {
.sb = dentry->d_sb,
.dentry = dentry,
.authflavor = NFS_SB(dentry->d_sb)->client->cl_auth->au_flavor,
};
char *page = NULL, *page2 = NULL;
int loc, error;
if (locations == NULL || locations->nlocations <= 0)
goto out;
dprintk("%s: referral at %s/%s\n", __func__,
dentry->d_parent->d_name.name, dentry->d_name.name);
page = (char *) __get_free_page(GFP_USER);
if (!page)
goto out;
page2 = (char *) __get_free_page(GFP_USER);
if (!page2)
goto out;
/* Ensure fs path is a prefix of current dentry path */
error = nfs4_validate_fspath(dentry, locations, page, page2);
if (error < 0) {
mnt = ERR_PTR(error);
goto out;
}
for (loc = 0; loc < locations->nlocations; loc++) {
const struct nfs4_fs_location *location = &locations->locations[loc];
if (location == NULL || location->nservers <= 0 ||
location->rootpath.ncomponents == 0)
continue;
mnt = try_location(&mountdata, page, page2, location);
if (!IS_ERR(mnt))
break;
}
out:
free_page((unsigned long) page);
free_page((unsigned long) page2);
dprintk("%s: done\n", __func__);
return mnt;
}
/*
* nfs_do_refmount - handle crossing a referral on server
* @dentry - dentry of referral
*
*/
struct vfsmount *nfs_do_refmount(struct rpc_clnt *client, struct dentry *dentry)
{
struct vfsmount *mnt = ERR_PTR(-ENOMEM);
struct dentry *parent;
struct nfs4_fs_locations *fs_locations = NULL;
struct page *page;
int err;
/* BUG_ON(IS_ROOT(dentry)); */
dprintk("%s: enter\n", __func__);
page = alloc_page(GFP_KERNEL);
if (page == NULL)
goto out;
fs_locations = kmalloc(sizeof(struct nfs4_fs_locations), GFP_KERNEL);
if (fs_locations == NULL)
goto out_free;
/* Get locations */
mnt = ERR_PTR(-ENOENT);
parent = dget_parent(dentry);
dprintk("%s: getting locations for %s/%s\n",
__func__, parent->d_name.name, dentry->d_name.name);
err = nfs4_proc_fs_locations(client, parent->d_inode, &dentry->d_name, fs_locations, page);
dput(parent);
if (err != 0 ||
fs_locations->nlocations <= 0 ||
fs_locations->fs_path.ncomponents <= 0)
goto out_free;
mnt = nfs_follow_referral(dentry, fs_locations);
out_free:
__free_page(page);
kfree(fs_locations);
out:
dprintk("%s: done\n", __func__);
return mnt;
}
static int fuse_direntplus_link(struct file *file,
struct fuse_direntplus *direntplus,
u64 attr_version)
{
int err;
struct fuse_entry_out *o = &direntplus->entry_out;
struct fuse_dirent *dirent = &direntplus->dirent;
struct dentry *parent = file->f_path.dentry;
struct qstr name = { .len = dirent->namelen, .name = dirent->name };
struct dentry *dentry;
struct dentry *alias;
struct inode *dir = parent->d_inode;
struct fuse_conn *fc;
struct inode *inode;
if (!o->nodeid) {
/*
* Unlike in the case of fuse_lookup, zero nodeid does not mean
* ENOENT. Instead, it only means the userspace filesystem did
* not want to return attributes/handle for this entry.
*
* So do nothing.
*/
return 0;
}
if (name.name[0] == '.') {
/*
* We could potentially refresh the attributes of the directory
* and its parent?
*/
if (name.len == 1)
return 0;
if (name.name[1] == '.' && name.len == 2)
return 0;
}
fc = get_fuse_conn(dir);
name.hash = full_name_hash(name.name, name.len);
dentry = d_lookup(parent, &name);
if (dentry && dentry->d_inode) {
inode = dentry->d_inode;
if (get_node_id(inode) == o->nodeid) {
struct fuse_inode *fi;
fi = get_fuse_inode(inode);
spin_lock(&fc->lock);
fi->nlookup++;
spin_unlock(&fc->lock);
/*
* The other branch to 'found' comes via fuse_iget()
* which bumps nlookup inside
*/
goto found;
}
err = d_invalidate(dentry);
if (err)
goto out;
dput(dentry);
dentry = NULL;
}
dentry = d_alloc(parent, &name);
err = -ENOMEM;
if (!dentry)
goto out;
inode = fuse_iget(dir->i_sb, o->nodeid, o->generation,
&o->attr, entry_attr_timeout(o), attr_version);
if (!inode)
goto out;
alias = d_materialise_unique(dentry, inode);
err = PTR_ERR(alias);
if (IS_ERR(alias))
goto out;
if (alias) {
dput(dentry);
dentry = alias;
}
found:
fuse_change_attributes(inode, &o->attr, entry_attr_timeout(o),
attr_version);
fuse_change_entry_timeout(dentry, o);
err = 0;
out:
if (dentry)
dput(dentry);
return err;
}
static int parse_dirplusfile(char *buf, size_t nbytes, struct file *file,
void *dstbuf, filldir_t filldir, u64 attr_version)
{
struct fuse_direntplus *direntplus;
struct fuse_dirent *dirent;
size_t reclen;
int over = 0;
int ret;
while (nbytes >= FUSE_NAME_OFFSET_DIRENTPLUS) {
direntplus = (struct fuse_direntplus *) buf;
dirent = &direntplus->dirent;
reclen = FUSE_DIRENTPLUS_SIZE(direntplus);
if (!dirent->namelen || dirent->namelen > FUSE_NAME_MAX)
return -EIO;
if (reclen > nbytes)
break;
if (!over) {
/* We fill entries into dstbuf only as much as
it can hold. But we still continue iterating
over remaining entries to link them. If not,
we need to send a FORGET for each of those
which we did not link.
*/
over = filldir(dstbuf, dirent->name, dirent->namelen,
file->f_pos, dirent->ino,
dirent->type);
file->f_pos = dirent->off;
}
buf += reclen;
nbytes -= reclen;
ret = fuse_direntplus_link(file, direntplus, attr_version);
if (ret)
fuse_force_forget(file, direntplus->entry_out.nodeid);
}
return 0;
}
static int fuse_readdir(struct file *file, void *dstbuf, filldir_t filldir)
{
int err;
size_t nbytes;
struct page *page;
struct inode *inode = file->f_path.dentry->d_inode;
struct fuse_conn *fc = get_fuse_conn(inode);
struct fuse_req *req;
u64 attr_version = 0;
if (is_bad_inode(inode))
return -EIO;
req = fuse_get_req(fc, 1);
if (IS_ERR(req))
return PTR_ERR(req);
page = alloc_page(GFP_KERNEL);
if (!page) {
fuse_put_request(fc, req);
return -ENOMEM;
}
req->out.argpages = 1;
req->num_pages = 1;
req->pages[0] = page;
req->page_descs[0].length = PAGE_SIZE;
if (fc->do_readdirplus) {
attr_version = fuse_get_attr_version(fc);
fuse_read_fill(req, file, file->f_pos, PAGE_SIZE,
FUSE_READDIRPLUS);
} else {
fuse_read_fill(req, file, file->f_pos, PAGE_SIZE,
FUSE_READDIR);
}
fuse_request_send(fc, req);
nbytes = req->out.args[0].size;
err = req->out.h.error;
fuse_put_request(fc, req);
if (!err) {
if (fc->do_readdirplus) {
err = parse_dirplusfile(page_address(page), nbytes,
file, dstbuf, filldir,
attr_version);
} else {
err = parse_dirfile(page_address(page), nbytes, file,
dstbuf, filldir);
}
}
__free_page(page);
fuse_invalidate_attr(inode); /* atime changed */
return err;
}
struct pcifront_dev *init_pcifront(char *_nodename)
{
xenbus_transaction_t xbt;
char* err;
char* message=NULL;
int retry=0;
char* msg;
char* nodename = _nodename ? _nodename : "device/pci/0";
int dom;
struct pcifront_dev *dev;
char path[strlen(nodename) + strlen("/backend-id") + 1];
if (!_nodename && pcidev)
return pcidev;
printk("******************* PCIFRONT for %s **********\n\n\n", nodename);
snprintf(path, sizeof(path), "%s/backend-id", nodename);
dom = xenbus_read_integer(path);
if (dom == -1) {
printk("no backend\n");
return NULL;
}
dev = malloc(sizeof(*dev));
memset(dev, 0, sizeof(*dev));
dev->nodename = strdup(nodename);
dev->dom = dom;
evtchn_alloc_unbound(dev->dom, pcifront_handler, dev, &dev->evtchn);
dev->info = (struct xen_pci_sharedinfo*) alloc_page();
memset(dev->info,0,PAGE_SIZE);
dev->info_ref = gnttab_grant_access(dev->dom,virt_to_mfn(dev->info),0);
dev->events = NULL;
again:
err = xenbus_transaction_start(&xbt);
if (err) {
printk("starting transaction\n");
free(err);
}
err = xenbus_printf(xbt, nodename, "pci-op-ref","%u",
dev->info_ref);
if (err) {
message = "writing pci-op-ref";
goto abort_transaction;
}
err = xenbus_printf(xbt, nodename,
"event-channel", "%u", dev->evtchn);
if (err) {
message = "writing event-channel";
goto abort_transaction;
}
err = xenbus_printf(xbt, nodename,
"magic", XEN_PCI_MAGIC);
if (err) {
message = "writing magic";
goto abort_transaction;
}
snprintf(path, sizeof(path), "%s/state", nodename);
err = xenbus_switch_state(xbt, path, XenbusStateInitialised);
if (err) {
message = "switching state";
goto abort_transaction;
}
err = xenbus_transaction_end(xbt, 0, &retry);
if (err) free(err);
if (retry) {
goto again;
printk("completing transaction\n");
}
goto done;
abort_transaction:
free(err);
err = xenbus_transaction_end(xbt, 1, &retry);
printk("Abort transaction %s\n", message);
goto error;
done:
snprintf(path, sizeof(path), "%s/backend", nodename);
msg = xenbus_read(XBT_NIL, path, &dev->backend);
if (msg) {
printk("Error %s when reading the backend path %s\n", msg, path);
goto error;
}
printk("backend at %s\n", dev->backend);
{
char path[strlen(dev->backend) + strlen("/state") + 1];
char frontpath[strlen(nodename) + strlen("/state") + 1];
XenbusState state;
snprintf(path, sizeof(path), "%s/state", dev->backend);
xenbus_watch_path_token(XBT_NIL, path, path, &dev->events);
err = NULL;
state = xenbus_read_integer(path);
while (err == NULL && state < XenbusStateConnected)
err = xenbus_wait_for_state_change(path, &state, &dev->events);
if (state != XenbusStateConnected) {
printk("backend not avalable, state=%d\n", state);
xenbus_unwatch_path_token(XBT_NIL, path, path);
goto error;
}
snprintf(frontpath, sizeof(frontpath), "%s/state", nodename);
if ((err = xenbus_switch_state(XBT_NIL, frontpath, XenbusStateConnected))
!= NULL) {
printk("error switching state %s\n", err);
xenbus_unwatch_path_token(XBT_NIL, path, path);
goto error;
}
}
unmask_evtchn(dev->evtchn);
printk("**************************\n");
if (!_nodename)
pcidev = dev;
return dev;
error:
free(err);
free_pcifront(dev);
return NULL;
}
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);
spin_lock_irqsave(&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) {
if (rc > 0) {
int ret;
/* We hit the Xen hard limit: reprobe. */
reservation.nr_extents = rc;
ret = HYPERVISOR_memory_op(XENMEM_decrease_reservation,
&reservation);
BUG_ON(ret != rc);
}
if (rc >= 0)
balloon_stats.hard_limit = (balloon_stats.current_pages + rc -
balloon_stats.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));
set_phys_to_machine(pfn, frame_list[i]);
/* 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),
mfn_pte(frame_list[i], PAGE_KERNEL),
0);
BUG_ON(ret);
}
/* Relinquish the page back to the allocator. */
ClearPageReserved(page);
init_page_count(page);
__free_page(page);
}
balloon_stats.current_pages += nr_pages;
totalram_pages = balloon_stats.current_pages;
out:
spin_unlock_irqrestore(&balloon_lock, flags);
return 0;
}
static int decrease_reservation(unsigned long nr_pages)
{
unsigned long pfn, i, flags;
struct page *page;
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);
scrub_page(page);
if (!PageHighMem(page)) {
ret = HYPERVISOR_update_va_mapping(
(unsigned long)__va(pfn << PAGE_SHIFT),
__pte_ma(0), 0);
BUG_ON(ret);
}
}
/* Ensure that ballooned highmem pages don't have kmaps. */
kmap_flush_unused();
flush_tlb_all();
spin_lock_irqsave(&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);
balloon_stats.current_pages -= nr_pages;
totalram_pages = balloon_stats.current_pages;
spin_unlock_irqrestore(&balloon_lock, 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(struct work_struct *work)
{
int need_sleep = 0;
long credit;
mutex_lock(&balloon_mutex);
do {
credit = current_target() - balloon_stats.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() != balloon_stats.current_pages)
mod_timer(&balloon_timer, jiffies + HZ);
mutex_unlock(&balloon_mutex);
}
/* filemap_write_and_wait(inode->i_mapping); */
if ( inode->i_mapping->nrpages
&& filemap_fdatawrite(inode->i_mapping) != -EIO)
filemap_fdatawait(inode->i_mapping);
#endif
rc = VbglR0SfClose(&client_handle, &sf_g->map, sf_r->handle);
if (RT_FAILURE(rc))
LogFunc(("VbglR0SfClose failed rc=%Rrc\n", rc));
kfree(sf_r);
sf_i->file = NULL;
sf_i->handle = SHFL_HANDLE_NIL;
file->private_data = NULL;
return 0;
}
#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 25)
static int sf_reg_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 0)
static struct page *sf_reg_nopage(struct vm_area_struct *vma, unsigned long vaddr, int *type)
# define SET_TYPE(t) *type = (t)
#else /* LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 0) */
static struct page *sf_reg_nopage(struct vm_area_struct *vma, unsigned long vaddr, int unused)
# define SET_TYPE(t)
#endif
{
struct page *page;
char *buf;
loff_t off;
uint32_t nread = PAGE_SIZE;
int err;
struct file *file = vma->vm_file;
struct inode *inode = GET_F_DENTRY(file)->d_inode;
struct sf_glob_info *sf_g = GET_GLOB_INFO(inode->i_sb);
struct sf_reg_info *sf_r = file->private_data;
TRACE();
#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 25)
if (vmf->pgoff > vma->vm_end)
return VM_FAULT_SIGBUS;
#else
if (vaddr > vma->vm_end)
{
SET_TYPE(VM_FAULT_SIGBUS);
return NOPAGE_SIGBUS;
}
#endif
/* Don't use GFP_HIGHUSER as long as sf_reg_read_aux() calls VbglR0SfRead()
* which works on virtual addresses. On Linux cannot reliably determine the
* physical address for high memory, see rtR0MemObjNativeLockKernel(). */
page = alloc_page(GFP_USER);
if (!page) {
LogRelFunc(("failed to allocate page\n"));
#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 25)
return VM_FAULT_OOM;
#else
SET_TYPE(VM_FAULT_OOM);
return NOPAGE_OOM;
#endif
}
buf = kmap(page);
#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 25)
off = (vmf->pgoff << PAGE_SHIFT);
#else
off = (vaddr - vma->vm_start) + (vma->vm_pgoff << PAGE_SHIFT);
#endif
err = sf_reg_read_aux(__func__, sf_g, sf_r, buf, &nread, off);
if (err)
{
kunmap(page);
put_page(page);
#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 25)
return VM_FAULT_SIGBUS;
#else
SET_TYPE(VM_FAULT_SIGBUS);
return NOPAGE_SIGBUS;
#endif
}
BUG_ON (nread > PAGE_SIZE);
if (!nread)
{
#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 25)
clear_user_page(page_address(page), vmf->pgoff, page);
#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 0)
clear_user_page(page_address(page), vaddr, page);
#else
clear_user_page(page_address(page), vaddr);
#endif
}
else
memset(buf + nread, 0, PAGE_SIZE - nread);
flush_dcache_page(page);
kunmap(page);
#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 25)
vmf->page = page;
return 0;
#else
SET_TYPE(VM_FAULT_MAJOR);
return page;
#endif
}
int do_pgfault(struct mm_struct *mm, machine_word_t error_code, uintptr_t addr)
{
if (mm == NULL) {
assert(current != NULL);
/* Chen Yuheng
* give handler a chance to deal with it
*/
kprintf
("page fault in kernel thread: pid = %d, name = %s, %d %08x.\n",
current->pid, current->name, error_code, addr);
return -E_KILLED;
}
bool need_unlock = 1;
if (!try_lock_mm(mm)) {
if (current != NULL && mm->locked_by == current->pid) {
need_unlock = 0;
} else {
lock_mm(mm);
}
}
int ret = -E_INVAL;
struct vma_struct *vma = find_vma(mm, addr);
if (vma == NULL || vma->vm_start > addr) {
goto failed;
}
if (vma->vm_flags & VM_STACK) {
if (addr < vma->vm_start + PGSIZE) {
goto failed;
}
}
//kprintf("@ %x %08x\n", vma->vm_flags, vma->vm_start);
//assert((vma->vm_flags & VM_IO)==0);
if (vma->vm_flags & VM_IO) {
ret = -E_INVAL;
goto failed;
}
switch (error_code & 3) {
default:
/* default is 3: write, present */
case 2: /* write, not present */
if (!(vma->vm_flags & VM_WRITE)) {
goto failed;
}
break;
case 1: /* read, present */
goto failed;
case 0: /* read, not present */
if (!(vma->vm_flags & (VM_READ | VM_EXEC))) {
goto failed;
}
}
pte_perm_t perm, nperm;
#ifdef ARCH_ARM
/* ARM9 software emulated PTE_xxx */
perm = PTE_P | PTE_U;
if (vma->vm_flags & VM_WRITE) {
perm |= PTE_W;
}
#else
ptep_unmap(&perm);
ptep_set_u_read(&perm);
if (vma->vm_flags & VM_WRITE) {
ptep_set_u_write(&perm);
}
#endif
addr = ROUNDDOWN(addr, PGSIZE);
ret = -E_NO_MEM;
pte_t *ptep;
if ((ptep = get_pte(mm->pgdir, addr, 1)) == NULL) {
goto failed;
}
if (ptep_invalid(ptep)) {
#ifdef UCONFIG_BIONIC_LIBC
if (vma->mfile.file != NULL) {
struct file *file = vma->mfile.file;
off_t old_pos = file->pos, new_pos =
vma->mfile.offset + addr - vma->vm_start;
#ifdef SHARE_MAPPED_FILE
struct mapped_addr *maddr =
find_maddr(file, new_pos, NULL);
if (maddr == NULL) {
#endif // SHARE_MAPPED_FILE
struct Page *page;
if ((page = alloc_page()) == NULL) {
assert(false);
goto failed;
}
nperm = perm;
#ifdef ARCH_ARM
/* ARM9 software emulated PTE_xxx */
nperm &= ~PTE_W;
#else
ptep_unset_s_write(&nperm);
#endif
page_insert_pte(mm->pgdir, page, ptep, addr,
nperm);
if ((ret =
filestruct_setpos(file, new_pos)) != 0) {
assert(false);
goto failed;
}
filestruct_read(file, page2kva(page), PGSIZE);
if ((ret =
filestruct_setpos(file, old_pos)) != 0) {
assert(false);
goto failed;
}
#ifdef SHARE_MAPPED_FILE
if ((maddr = (struct mapped_addr *)
kmalloc(sizeof(struct mapped_addr))) !=
NULL) {
maddr->page = page;
maddr->offset = new_pos;
page->maddr = maddr;
list_add(&
(file->node->mapped_addr_list),
&(maddr->list));
} else {
assert(false);
}
} else {
nperm = perm;
#ifdef ARCH_ARM
/* ARM9 software emulated PTE_xxx */
nperm &= ~PTE_W;
#else
ptep_unset_s_write(&nperm);
#endif
page_insert_pte(mm->pgdir, maddr->page, ptep,
addr, nperm);
}
#endif //SHARE_MAPPED_FILE
} else
#endif //UCONFIG_BIONIC_LIBC
if (!(vma->vm_flags & VM_SHARE)) {
if (pgdir_alloc_page(mm->pgdir, addr, perm) == NULL) {
goto failed;
}
#ifdef UCONFIG_BIONIC_LIBC
if (vma->vm_flags & VM_ANONYMOUS) {
memset((void *)addr, 0, PGSIZE);
}
#endif //UCONFIG_BIONIC_LIBC
} else { //shared mem
lock_shmem(vma->shmem);
uintptr_t shmem_addr =
addr - vma->vm_start + vma->shmem_off;
pte_t *sh_ptep =
shmem_get_entry(vma->shmem, shmem_addr, 1);
if (sh_ptep == NULL || ptep_invalid(sh_ptep)) {
unlock_shmem(vma->shmem);
goto failed;
}
unlock_shmem(vma->shmem);
if (ptep_present(sh_ptep)) {
page_insert(mm->pgdir, pa2page(*sh_ptep), addr,
perm);
} else {
#ifdef UCONFIG_SWAP
swap_duplicate(*ptep);
ptep_copy(ptep, sh_ptep);
#else
panic("NO SWAP\n");
#endif
}
}
} else { //a present page, handle copy-on-write (cow)
struct Page *page, *newpage = NULL;
bool cow =
((vma->vm_flags & (VM_SHARE | VM_WRITE)) == VM_WRITE),
may_copy = 1;
#if 1
if (!(!ptep_present(ptep)
|| ((error_code & 2) && !ptep_u_write(ptep) && cow))) {
//assert(PADDR(mm->pgdir) == rcr3());
kprintf("%p %p %d %d %x\n", *ptep, addr, error_code,
cow, vma->vm_flags);
assert(0);
}
#endif
if (cow) {
newpage = alloc_page();
}
if (ptep_present(ptep)) {
page = pte2page(*ptep);
} else {
#ifdef UCONFIG_SWAP
if ((ret = swap_in_page(*ptep, &page)) != 0) {
if (newpage != NULL) {
free_page(newpage);
}
goto failed;
}
#else
assert(0);
#endif
if (!(error_code & 2) && cow) {
#ifdef ARCH_ARM
//#warning ARM9 software emulated PTE_xxx
perm &= ~PTE_W;
#else
ptep_unset_s_write(&perm);
#endif
may_copy = 0;
}
}
if (cow && may_copy) {
#ifdef UCONFIG_SWAP
if (page_ref(page) + swap_page_count(page) > 1) {
#else
if (page_ref(page) > 1) {
#endif
if (newpage == NULL) {
goto failed;
}
memcpy(page2kva(newpage), page2kva(page),
PGSIZE);
//kprintf("COW!\n");
page = newpage, newpage = NULL;
}
}
#ifdef UCONFIG_BIONIC_LIBC
else if (vma->mfile.file != NULL) {
#ifdef UCONFIG_SWAP
assert(page_reg(page) + swap_page_count(page) == 1);
#else
assert(page_ref(page) == 1);
#endif
#ifdef SHARE_MAPPED_FILE
off_t offset = vma->mfile.offset + addr - vma->vm_start;
struct mapped_addr *maddr =
find_maddr(vma->mfile.file, offset, page);
if (maddr != NULL) {
list_del(&(maddr->list));
kfree(maddr);
page->maddr = NULL;
assert(find_maddr(vma->mfile.file, offset, page)
== NULL);
} else {
}
#endif //SHARE_MAPPED_FILE
}
#endif //UCONFIG_BIONIC_LIBC
else {
}
page_insert(mm->pgdir, page, addr, perm);
if (newpage != NULL) {
free_page(newpage);
}
}
ret = 0;
failed:
if (need_unlock) {
unlock_mm(mm);
}
return ret;
}
/*
* Must not be called with IRQs off. This should only be used on the
* slow path.
*
* Copy a foreign granted page to local memory.
*/
int gnttab_copy_grant_page(grant_ref_t ref, struct page **pagep)
{
struct gnttab_unmap_and_replace unmap;
mmu_update_t mmu;
struct page *page;
struct page *new_page;
void *new_addr;
void *addr;
paddr_t pfn;
maddr_t mfn;
maddr_t new_mfn;
int err;
page = *pagep;
if (!get_page_unless_zero(page))
return -ENOENT;
err = -ENOMEM;
new_page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
if (!new_page)
goto out;
new_addr = page_address(new_page);
addr = page_address(page);
memcpy(new_addr, addr, PAGE_SIZE);
pfn = page_to_pfn(page);
mfn = pfn_to_mfn(pfn);
new_mfn = virt_to_mfn(new_addr);
write_seqlock(&gnttab_dma_lock);
/* Make seq visible before checking page_mapped. */
smp_mb();
/* Has the page been DMA-mapped? */
if (unlikely(page_mapped(page))) {
write_sequnlock(&gnttab_dma_lock);
put_page(new_page);
err = -EBUSY;
goto out;
}
if (!xen_feature(XENFEAT_auto_translated_physmap))
set_phys_to_machine(pfn, new_mfn);
gnttab_set_replace_op(&unmap, (unsigned long)addr,
(unsigned long)new_addr, ref);
err = HYPERVISOR_grant_table_op(GNTTABOP_unmap_and_replace,
&unmap, 1);
BUG_ON(err);
BUG_ON(unmap.status != GNTST_okay);
write_sequnlock(&gnttab_dma_lock);
if (!xen_feature(XENFEAT_auto_translated_physmap)) {
set_phys_to_machine(page_to_pfn(new_page), INVALID_P2M_ENTRY);
mmu.ptr = (new_mfn << PAGE_SHIFT) | MMU_MACHPHYS_UPDATE;
mmu.val = pfn;
err = HYPERVISOR_mmu_update(&mmu, 1, NULL, DOMID_SELF);
BUG_ON(err);
}
new_page->mapping = page->mapping;
new_page->index = page->index;
set_bit(PG_foreign, &new_page->flags);
*pagep = new_page;
SetPageForeign(page, gnttab_page_free);
page->mapping = NULL;
out:
put_page(page);
return err;
}
static int lgx_alloc_buffer(struct inno_lgx *lgx)
{
//int i,j;
//int page_num;
//struct page *page = NULL;
int i;
struct inno_buffer *inno_buf = &lgx->inno_buffer;
//int ret = 0;
memset(inno_buf, 0, sizeof(struct inno_buffer));
sema_init(&inno_buf->sem,1);
down(&inno_buf->sem);
// alloc buffer
#if 0 //xingyu buffer issue
page_num = PAGE_ALIGN(INNO_BUFFER_SIZE) / PAGE_SIZE;
inno_buf->pages = (struct page **)kzalloc(page_num * sizeof(struct page *), GFP_KERNEL);
if (!inno_buf->pages) {
inno_err("lgx_alloc_buffer:No enough memory");
ret = -ENOMEM;
goto alloc_node_error;
}
for (i = 0; i < page_num; i++) {
page = alloc_page(GFP_KERNEL);
if (!page) {
inno_err("lgx_alloc_buffer:No enough page");
ret = -ENOMEM;
goto alloc_pages_error;
}
//SetPageReserved(page);
inno_buf->pages[i] = page;
}
inno_buf->page_num = page_num;
inno_buf->bufsize = page_num * PAGE_SIZE;
inno_buf->vaddr = vmap(inno_buf->pages, page_num, VM_MAP, PAGE_KERNEL);
/* check if the memory map is OK. */
if (!inno_buf->vaddr) {
inno_err("lgx_alloc_buffer:vmap() failure");
ret = -EFAULT;
goto vmap_error;
}
memset(inno_buf->vaddr, 0, inno_buf->bufsize);
inno_buf->start = inno_buf->vaddr;
#else
#ifndef _buffer_global // buffer alloc modify xingyu 0714
inno_buf->vaddr = kmalloc(INNO_BUFFER_SIZE, GFP_KERNEL| GFP_DMA);
if(inno_buf->vaddr == NULL || (int)inno_buf->vaddr %4 !=0){
inno_err("inno_buf->vaddr kmalloc fail");
up(&inno_buf->sem);
return -1;
}
#else
for(i=0; lgx_ids_table[i].name!=NULL; i++){
if(lgx->ids == &lgx_ids_table[i]){
inno_buf->vaddr =i*INNO_BUFFER_SIZE+g_inno_buffer;
inno_msg("use global mem");
break;
}
}
#endif
inno_buf->start = inno_buf->vaddr;
inno_buf->bufsize = INNO_BUFFER_SIZE;
#endif
up(&inno_buf->sem);
return 0;
#if 0 //xingyu buffer issue
//vmap_error:
//alloc_pages_error:
for (j = 0; j < i; j++) {
page = inno_buf->pages[j];
//ClearPageReserved(page);
__free_page(page);
}
kfree(inno_buf->pages);
//alloc_node_error:
return ret;
#endif
}
static int __dune_vm_page_walk(ptent_t *dir, void *start_va, void *end_va,
page_walk_cb cb, const void *arg, int level,
int create)
{
// XXX: Using PA == VA
int i, ret;
//level==3, start_idx is pml4. level==2, start_idx is pdpt. level==1, start_idx is pd. level==0, start_idx is pt. commented by wenjia zhao.
int start_idx = PDX(level, start_va);
int end_idx = PDX(level, end_va);
//discard the 0-47 of va. commented by wenjia zhao
void *base_va = (void *) ((unsigned long)
start_va & ~(PDADDR(level + 1, 1) - 1));
assert(level >= 0 && level <= NPTLVLS);
assert(end_idx < NPTENTRIES);
for (i = start_idx; i <= end_idx; i++) {
void *n_start_va, *n_end_va;
void *cur_va = base_va + PDADDR(level, i);
ptent_t *pte = &dir[i];
if (level == 0) {
if (create == CREATE_NORMAL || *pte) {
ret = cb(arg, pte, cur_va);
if (ret)
return ret;
}
continue;
}
if (level == 1) {
if (create == CREATE_BIG || pte_big(*pte)) {
ret = cb(arg, pte, cur_va);
if (ret)
return ret;
continue;
}
}
if (level == 2) {
if (create == CREATE_BIG_1GB || pte_big(*pte)) {
ret = cb(arg, pte, cur_va);
if (ret)
return ret;
continue;
}
}
if (!pte_present(*pte)) {
ptent_t *new_pte;
if (!create)
continue;
new_pte = alloc_page();
if (!new_pte)
return -ENOMEM;
memset(new_pte, 0, PGSIZE);
//construct the page that pml4 pointed. commented by wenjia zhao.
*pte = PTE_ADDR(new_pte) | PTE_DEF_FLAGS;
}
n_start_va = (i == start_idx) ? start_va : cur_va;
n_end_va = (i == end_idx) ? end_va : cur_va + PDADDR(level, 1) - 1;
ret = __dune_vm_page_walk((ptent_t *) PTE_ADDR(dir[i]),
n_start_va, n_end_va, cb, arg,
level - 1, create);
if (ret)
return ret;
}
return 0;
}
static ssize_t
pipe_write(struct kiocb *iocb, const struct iovec *_iov,
unsigned long nr_segs, loff_t ppos)
{
struct file *filp = iocb->ki_filp;
struct inode *inode = filp->f_path.dentry->d_inode;
struct pipe_inode_info *pipe;
ssize_t ret;
int do_wakeup;
struct iovec *iov = (struct iovec *)_iov;
size_t total_len;
ssize_t chars;
total_len = iov_length(iov, nr_segs);
/* Null write succeeds. */
if (unlikely(total_len == 0))
return 0;
do_wakeup = 0;
ret = 0;
mutex_lock(&inode->i_mutex);
pipe = inode->i_pipe;
if (!pipe->readers) {
send_sig(SIGPIPE, current, 0);
ret = -EPIPE;
goto out;
}
/* We try to merge small writes */
chars = total_len & (PAGE_SIZE-1); /* size of the last buffer */
if (pipe->nrbufs && chars != 0) {
int lastbuf = (pipe->curbuf + pipe->nrbufs - 1) &
(PIPE_BUFFERS-1);
struct pipe_buffer *buf = pipe->bufs + lastbuf;
const struct pipe_buf_operations *ops = buf->ops;
int offset = buf->offset + buf->len;
if (ops->can_merge && offset + chars <= PAGE_SIZE) {
int error, atomic = 1;
void *addr;
error = ops->confirm(pipe, buf);
if (error)
goto out;
iov_fault_in_pages_read(iov, chars);
redo1:
addr = ops->map(pipe, buf, atomic);
error = pipe_iov_copy_from_user(offset + addr, iov,
chars, atomic);
ops->unmap(pipe, buf, addr);
ret = error;
do_wakeup = 1;
if (error) {
if (atomic) {
atomic = 0;
goto redo1;
}
goto out;
}
buf->len += chars;
total_len -= chars;
ret = chars;
if (!total_len)
goto out;
}
}
for (;;) {
int bufs;
if (!pipe->readers) {
send_sig(SIGPIPE, current, 0);
if (!ret)
ret = -EPIPE;
break;
}
bufs = pipe->nrbufs;
if (bufs < PIPE_BUFFERS) {
int newbuf = (pipe->curbuf + bufs) & (PIPE_BUFFERS-1);
struct pipe_buffer *buf = pipe->bufs + newbuf;
struct page *page = pipe->tmp_page;
char *src;
int error, atomic = 1;
if (!page) {
page = alloc_page(GFP_HIGHUSER | __GFP_UBC);
if (unlikely(!page)) {
ret = ret ? : -ENOMEM;
break;
}
pipe->tmp_page = page;
}
static void run_tool(uint8_t devices) {
uint8_t highlighted;
for (uint8_t i = 0; i < 4; i++) {
if (ide_device_is_present(i) && ide_device_get_type(i) == 0) {
highlighted = i;
break;
}
}
bool selected[4] = {0,0,0,0};
bool selecting = true;
uint8_t selectedNum = 0;
while (selecting) {
console_clear();
print_header();
console_puts("Please select the drives which you would like to erase.\n");
for (uint8_t i = 0; i < 4; i++) {
if (ide_device_is_present(i) && ide_device_get_type(i) == 0) {
uint8_t foreground = 0x7, background = 0x0;
if (selected[i]) foreground = 0xC;
if (highlighted == i) {
background = 0xF;
if (foreground == 0x7) foreground = 0x0;
}
console_color((background << 4) + foreground);
console_putsf(" - ATA Device %u (%s) %7uMB - %s\n", i, ide_device_is_dma_capable(i) ? "DMA" : "PIO", ide_device_get_size(i) / 1024 / 2, ide_device_get_string(i, IDE_STRING_MODEL));
console_color(0x07);
}
}
for (int i = 0; i < 10 - devices; i++) console_puts("\n");
console_puts("Use the \"ARROW KEYS\" to highlight a drive.\n");
console_puts("Press \"SPACE\" to select/deselect the currently highlighted drive.\n\n");
console_puts("NOTE: You must select at least one drive to continue.\n\n");
console_puts("Press \"ENTER\" to continue.\n");
console_puts("Press \"ESCAPE\" to go back.");
switch (get_key()) {
case '\n':
if (selectedNum > 0) {
selecting = false;
}
break;
case '\1':
return;
case '\3':
do {
highlighted--;
if (highlighted > 3) highlighted = 3;
} while (!(ide_device_is_present(highlighted) && ide_device_get_type(highlighted) == 0));
break;
case '\5':
do {
highlighted++;
if (highlighted == 4) highlighted = 0;
} while (!(ide_device_is_present(highlighted) && ide_device_get_type(highlighted) == 0));
break;
case ' ':
selected[highlighted] = !selected[highlighted];
selectedNum += selected[highlighted] ? 1 : -1;
break;
}
}
console_clear();
print_header();
console_puts("Please select the drives which you would like to erase.\n");
for (uint8_t i = 0; i < 4; i++) {
if (ide_device_is_present(i) && ide_device_get_type(i) == 0) {
uint8_t foreground = 0x7, background = 0x0;
if (selected[i]) foreground = 0xC;
console_color((background << 4) + foreground);
console_putsf(" - ATA Device %u (%s) %7uMB - %s\n", i, ide_device_is_dma_capable(i) ? "DMA" : "PIO", ide_device_get_size(i) / 1024 / 2, ide_device_get_string(i, IDE_STRING_MODEL));
console_color(0x07);
}
}
console_puts("\n\nPlease enter the number of passes to be performed (max 99, 0 to cancel): ");
uint8_t idx = 0;
char raw_passes[3];
char c;
do {
c = get_key();
if (c == '\x7f' && idx > 0) {
idx--;
console_cursor(console_row(), console_col() - 1);
console_puts(" ");
console_cursor(console_row(), console_col() - 1);
} else if (c >= '0' && c <= '9' && idx < 2) {
console_putsf("%c", c);
raw_passes[idx] = c;
idx++;
}
} while (!(c == '\n' && idx != 0));
raw_passes[idx] = '\0';
uint32_t passes = atoi(raw_passes);
if (passes == 0) return;
console_puts("\n\nPress \"ENTER\" to ");
console_color(0x0C);
console_puts("WIPE THE ATA DISKS SELECTED ABOVE");
console_color(0x07);
console_putsf(" in %u pass", passes);
if (passes != 1) console_puts("es");
console_puts(".\nPress \"ESCAPE\" to cancel and go back.\n\n");
bool wait = true;
while (wait) {
switch (get_key()) {
case '\1':
return;
break;
case '\n':
wait = false;
}
}
console_clear();
console_puts("Starting Data Erasure Tool...\n\n");
uint8_t *space = (uint8_t *) page_to_virt(alloc_page(0));
for (uint32_t i = 1; i < 64; i++) {
alloc_page(0);
console_putsf("Initializing... %3u%%\r", ((i + 1) * 100) / 64);
}
console_puts("Initializing... 100%%\n");
uint8_t bit = 0x00;
for (uint32_t pass = 0; pass < passes; pass++) {
console_clear();
console_puts("Starting Data Erasure Tool...\n\n");
console_putsf("Initializing... 100%%\n", space);
console_putsf("\nWriting Pass %02u of %02u...\n", pass + 1, passes);
for (uint32_t i = 0; i < (64 * 4 * 1024); i++) {
space[i] = bit;
}
int device = 0;
for (uint32_t i = 0; i < 4; i++) {
if (selected[i]) {
device++;
uint32_t count = 0;
uint32_t written = 0;
while(written < ide_device_get_size(i)) {
count++;
console_putsf(" - Writing to device %u... %3u%% %c\r", device, (written * 100) / ide_device_get_size(i), twirls[(count / 10) % 4]);
written += ide_write_sectors_same(i, ide_device_get_size(i) - written, written, space) / 512;
}
console_putsf(" - Writing to device %u... 100%% \n", device);
}
}
bit = ~bit;
}
console_color(0x0A);
console_puts("\nOperation completed successfully! ");
console_color(0x07);
console_puts("You may now turn the computer off.");
beep();
die();
}
// LAB2: below code is used to check the first fit allocation algorithm (your EXERCISE 1)
// NOTICE: You SHOULD NOT CHANGE basic_check, default_check functions!
static void
default_check(void) {
int count = 0, total = 0;
list_entry_t *le = &free_list;
while ((le = list_next(le)) != &free_list) {
struct Page *p = le2page(le, page_link);
assert(PageProperty(p));
count ++, total += p->property;
}
assert(total == nr_free_pages());
basic_check();
struct Page *p0 = alloc_pages(5), *p1, *p2;
assert(p0 != NULL);
assert(!PageProperty(p0));
list_entry_t free_list_store = free_list;
list_init(&free_list);
assert(list_empty(&free_list));
assert(alloc_page() == NULL);
unsigned int nr_free_store = nr_free;
nr_free = 0;
free_pages(p0 + 2, 3);
assert(alloc_pages(4) == NULL);
assert(PageProperty(p0 + 2) && p0[2].property == 3);
assert((p1 = alloc_pages(3)) != NULL);
assert(alloc_page() == NULL);
assert(p0 + 2 == p1);
p2 = p0 + 1;
free_page(p0);
free_pages(p1, 3);
assert(PageProperty(p0) && p0->property == 1);
assert(PageProperty(p1) && p1->property == 3);
assert((p0 = alloc_page()) == p2 - 1);
free_page(p0);
assert((p0 = alloc_pages(2)) == p2 + 1);
free_pages(p0, 2);
free_page(p2);
assert((p0 = alloc_pages(5)) != NULL);
assert(alloc_page() == NULL);
assert(nr_free == 0);
nr_free = nr_free_store;
free_list = free_list_store;
free_pages(p0, 5);
le = &free_list;
while ((le = list_next(le)) != &free_list) {
struct Page *p = le2page(le, page_link);
count --, total -= p->property;
}
assert(count == 0);
assert(total == 0);
}
/**
* Compute the speed of specified hash function
*
* Run a speed test on the given hash algorithm on buffer of the given size.
* The speed is stored internally in the cfs_crypto_hash_speeds[] array, and
* is available through the cfs_crypto_hash_speed() function.
*
* \param[in] hash_alg hash algorithm id (CFS_HASH_ALG_*)
* \param[in] buf data buffer on which to compute the hash
* \param[in] buf_len length of \buf on which to compute hash
*/
static void cfs_crypto_performance_test(enum cfs_crypto_hash_alg hash_alg)
{
int buf_len = max(PAGE_SIZE, 1048576UL);
void *buf;
unsigned long start, end;
int bcount, err = 0;
struct page *page;
unsigned char hash[CFS_CRYPTO_HASH_DIGESTSIZE_MAX];
unsigned int hash_len = sizeof(hash);
page = alloc_page(GFP_KERNEL);
if (page == NULL) {
err = -ENOMEM;
goto out_err;
}
buf = kmap(page);
memset(buf, 0xAD, PAGE_SIZE);
kunmap(page);
for (start = jiffies, end = start + msecs_to_jiffies(MSEC_PER_SEC),
bcount = 0;
time_before(jiffies, end) && err == 0; bcount++) {
struct cfs_crypto_hash_desc *hdesc;
int i;
hdesc = cfs_crypto_hash_init(hash_alg, NULL, 0);
if (IS_ERR(hdesc)) {
err = PTR_ERR(hdesc);
break;
}
for (i = 0; i < buf_len / PAGE_SIZE; i++) {
err = cfs_crypto_hash_update_page(hdesc, page, 0,
PAGE_SIZE);
if (err != 0)
break;
}
err = cfs_crypto_hash_final(hdesc, hash, &hash_len);
if (err != 0)
break;
}
end = jiffies;
__free_page(page);
out_err:
if (err != 0) {
cfs_crypto_hash_speeds[hash_alg] = err;
CDEBUG(D_INFO, "Crypto hash algorithm %s test error: rc = %d\n",
cfs_crypto_hash_name(hash_alg), err);
} else {
unsigned long tmp;
tmp = ((bcount * buf_len / jiffies_to_msecs(end - start)) *
1000) / (1024 * 1024);
cfs_crypto_hash_speeds[hash_alg] = (int)tmp;
CDEBUG(D_CONFIG, "Crypto hash algorithm %s speed = %d MB/s\n",
cfs_crypto_hash_name(hash_alg),
cfs_crypto_hash_speeds[hash_alg]);
}
}
/*
* Message allocation uses this to build up regions of a message.
*
* @bytes - the number of bytes needed.
* @gfp - the waiting behaviour of the allocation
*
* @gfp is always ored with __GFP_HIGHMEM. Callers must be prepared to
* kmap the pages, etc.
*
* If @bytes is at least a full page then this just returns a page from
* alloc_page().
*
* If @bytes is a partial page then this stores the unused region of the
* page in a per-cpu structure. Future partial-page allocations may be
* satisfied from that cached region. This lets us waste less memory on
* small allocations with minimal complexity. It works because the transmit
* path passes read-only page regions down to devices. They hold a page
* reference until they are done with the region.
*/
int rds_page_remainder_alloc(struct scatterlist *scat, unsigned long bytes,
gfp_t gfp)
{
struct rds_page_remainder *rem;
unsigned long flags;
struct page *page;
int ret;
gfp |= __GFP_HIGHMEM;
/* jump straight to allocation if we're trying for a huge page */
if (bytes >= PAGE_SIZE) {
page = alloc_page(gfp);
if (page == NULL) {
ret = -ENOMEM;
} else {
sg_set_page(scat, page, PAGE_SIZE, 0);
ret = 0;
}
goto out;
}
rem = &per_cpu(rds_page_remainders, get_cpu());
local_irq_save(flags);
while (1) {
/* avoid a tiny region getting stuck by tossing it */
if (rem->r_page && bytes > (PAGE_SIZE - rem->r_offset)) {
rds_stats_inc(s_page_remainder_miss);
__free_page(rem->r_page);
rem->r_page = NULL;
}
/* hand out a fragment from the cached page */
if (rem->r_page && bytes <= (PAGE_SIZE - rem->r_offset)) {
sg_set_page(scat, rem->r_page, bytes, rem->r_offset);
get_page(sg_page(scat));
if (rem->r_offset != 0)
rds_stats_inc(s_page_remainder_hit);
rem->r_offset += bytes;
if (rem->r_offset == PAGE_SIZE) {
__free_page(rem->r_page);
rem->r_page = NULL;
}
ret = 0;
break;
}
/* alloc if there is nothing for us to use */
local_irq_restore(flags);
put_cpu();
page = alloc_page(gfp);
rem = &per_cpu(rds_page_remainders, get_cpu());
local_irq_save(flags);
if (page == NULL) {
ret = -ENOMEM;
break;
}
/* did someone race to fill the remainder before us? */
if (rem->r_page) {
__free_page(page);
continue;
}
/* otherwise install our page and loop around to alloc */
rem->r_page = page;
rem->r_offset = 0;
}
local_irq_restore(flags);
put_cpu();
out:
rdsdebug("bytes %lu ret %d %p %u %u\n", bytes, ret,
ret ? NULL : sg_page(scat), ret ? 0 : scat->offset,
ret ? 0 : scat->length);
return ret;
}
/*
* zswap_get_swap_cache_page
*
* This is an adaption of read_swap_cache_async()
*
* This function tries to find a page with the given swap entry
* in the swapper_space address space (the swap cache). If the page
* is found, it is returned in retpage. Otherwise, a page is allocated,
* added to the swap cache, and returned in retpage.
*
* If success, the swap cache page is returned in retpage
* Returns ZSWAP_SWAPCACHE_EXIST if page was already in the swap cache
* Returns ZSWAP_SWAPCACHE_NEW if the new page needs to be populated,
* the new page is added to swapcache and locked
* Returns ZSWAP_SWAPCACHE_FAIL on error
*/
static int zswap_get_swap_cache_page(swp_entry_t entry,
struct page **retpage)
{
struct page *found_page, *new_page = NULL;
struct address_space *swapper_space = swap_address_space(entry);
int err;
*retpage = NULL;
do {
/*
* First check the swap cache. Since this is normally
* called after lookup_swap_cache() failed, re-calling
* that would confuse statistics.
*/
found_page = find_get_page(swapper_space, entry.val);
if (found_page)
break;
/*
* Get a new page to read into from swap.
*/
if (!new_page) {
new_page = alloc_page(GFP_KERNEL);
if (!new_page)
break; /* Out of memory */
}
/*
* call radix_tree_preload() while we can wait.
*/
err = radix_tree_preload(GFP_KERNEL);
if (err)
break;
/*
* Swap entry may have been freed since our caller observed it.
*/
err = swapcache_prepare(entry);
if (err == -EEXIST) { /* seems racy */
radix_tree_preload_end();
continue;
}
if (err) { /* swp entry is obsolete ? */
radix_tree_preload_end();
break;
}
/* May fail (-ENOMEM) if radix-tree node allocation failed. */
__set_page_locked(new_page);
SetPageSwapBacked(new_page);
err = __add_to_swap_cache(new_page, entry);
if (likely(!err)) {
radix_tree_preload_end();
lru_cache_add_anon(new_page);
*retpage = new_page;
return ZSWAP_SWAPCACHE_NEW;
}
radix_tree_preload_end();
ClearPageSwapBacked(new_page);
__clear_page_locked(new_page);
/*
* add_to_swap_cache() doesn't return -EEXIST, so we can safely
* clear SWAP_HAS_CACHE flag.
*/
swapcache_free(entry, NULL);
} while (err != -ENOMEM);
if (new_page)
page_cache_release(new_page);
if (!found_page)
return ZSWAP_SWAPCACHE_FAIL;
*retpage = found_page;
return ZSWAP_SWAPCACHE_EXIST;
}
/* read a page from a file.
* We both read the page, and attach buffers to the page to record the
* address of each block (using bmap). These addresses will be used
* to write the block later, completely bypassing the filesystem.
* This usage is similar to how swap files are handled, and allows us
* to write to a file with no concerns of memory allocation failing.
*/
static struct page *read_page(struct file *file, unsigned long index,
struct bitmap *bitmap,
unsigned long count)
{
struct page *page = NULL;
struct inode *inode = file->f_dentry->d_inode;
struct buffer_head *bh;
sector_t block;
PRINTK("read bitmap file (%dB @ %Lu)\n", (int)PAGE_SIZE,
(unsigned long long)index << PAGE_SHIFT);
page = alloc_page(GFP_KERNEL);
if (!page)
page = ERR_PTR(-ENOMEM);
if (IS_ERR(page))
goto out;
bh = alloc_page_buffers(page, 1<<inode->i_blkbits, 0);
if (!bh) {
put_page(page);
page = ERR_PTR(-ENOMEM);
goto out;
}
attach_page_buffers(page, bh);
block = index << (PAGE_SHIFT - inode->i_blkbits);
while (bh) {
if (count == 0)
bh->b_blocknr = 0;
else {
bh->b_blocknr = bmap(inode, block);
if (bh->b_blocknr == 0) {
/* Cannot use this file! */
free_buffers(page);
page = ERR_PTR(-EINVAL);
goto out;
}
bh->b_bdev = inode->i_sb->s_bdev;
if (count < (1<<inode->i_blkbits))
count = 0;
else
count -= (1<<inode->i_blkbits);
bh->b_end_io = end_bitmap_write;
bh->b_private = bitmap;
atomic_inc(&bitmap->pending_writes);
set_buffer_locked(bh);
set_buffer_mapped(bh);
submit_bh(READ, bh);
}
block++;
bh = bh->b_this_page;
}
page->index = index;
wait_event(bitmap->write_wait,
atomic_read(&bitmap->pending_writes)==0);
if (bitmap->flags & BITMAP_WRITE_ERROR) {
free_buffers(page);
page = ERR_PTR(-EIO);
}
out:
if (IS_ERR(page))
printk(KERN_ALERT "md: bitmap read error: (%dB @ %Lu): %ld\n",
(int)PAGE_SIZE,
(unsigned long long)index << PAGE_SHIFT,
PTR_ERR(page));
return page;
}
static int vmw_gmr_build_descriptors(struct list_head *desc_pages,
struct page *pages[],
unsigned long num_pages)
{
struct page *page, *next;
struct svga_guest_mem_descriptor *page_virtual = NULL;
struct svga_guest_mem_descriptor *desc_virtual = NULL;
unsigned int desc_per_page;
unsigned long prev_pfn;
unsigned long pfn;
int ret;
desc_per_page = PAGE_SIZE /
sizeof(struct svga_guest_mem_descriptor) - 1;
while (likely(num_pages != 0)) {
page = alloc_page(__GFP_HIGHMEM);
if (unlikely(page == NULL)) {
ret = -ENOMEM;
goto out_err;
}
list_add_tail(&page->lru, desc_pages);
/*
* Point previous page terminating descriptor to this
* page before unmapping it.
*/
if (likely(page_virtual != NULL)) {
desc_virtual->ppn = page_to_pfn(page);
kunmap_atomic(page_virtual, KM_USER0);
}
page_virtual = kmap_atomic(page, KM_USER0);
desc_virtual = page_virtual - 1;
prev_pfn = ~(0UL);
while (likely(num_pages != 0)) {
pfn = page_to_pfn(*pages);
if (pfn != prev_pfn + 1) {
if (desc_virtual - page_virtual ==
desc_per_page - 1)
break;
(++desc_virtual)->ppn = cpu_to_le32(pfn);
desc_virtual->num_pages = cpu_to_le32(1);
} else {
uint32_t tmp =
le32_to_cpu(desc_virtual->num_pages);
desc_virtual->num_pages = cpu_to_le32(tmp + 1);
}
prev_pfn = pfn;
--num_pages;
++pages;
}
(++desc_virtual)->ppn = cpu_to_le32(0);
desc_virtual->num_pages = cpu_to_le32(0);
}
if (likely(page_virtual != NULL))
kunmap_atomic(page_virtual, KM_USER0);
return 0;
out_err:
list_for_each_entry_safe(page, next, desc_pages, lru) {
list_del_init(&page->lru);
__free_page(page);
}
/**
* @brief initialize host fram list
* @param host struct uhci_host
*/
int
init_hframelist(struct uhci_host *host)
{
struct usb_request_block *urb;
u32 frid;
phys32_t *frame_p;
virt_t framelist_virt;
phys_t framelist_phys;
int n_skels;
/* allocate a page for frame list */
alloc_page((void *)&framelist_virt, &framelist_phys);
if (!framelist_phys)
return -1;
host->hframelist = framelist_phys;
host->hframelist_virt = (phys32_t *)framelist_virt;
spinlock_lock(&host->lock_hfl);
/* create a TD for termination */
host->term_tdm = uhci_new_td_meta(host, NULL);
if (!host->term_tdm)
return -1;
host->term_tdm->td->link = UHCI_TD_LINK_TE;
host->term_tdm->td->status = 0U;
host->term_tdm->td->token =
UHCI_TD_TOKEN_DEVADDRESS(0x7f) | UHCI_TD_TOKEN_ENDPOINT(0) |
UHCI_TD_TOKEN_PID_IN | uhci_td_explen(0);
host->term_tdm->td->buffer = 0U;
/* create skelton QHs */
for (n_skels = 0; n_skels<UHCI_NUM_SKELTYPES; n_skels++) {
urb = create_urb(host);
if (!urb)
break;
urb->address = URB_ADDRESS_SKELTON;
URB_UHCI(urb)->qh =
uhci_alloc_qh(host, &URB_UHCI(urb)->qh_phys);
if (!URB_UHCI(urb)->qh)
break;
if (n_skels == 0) {
URB_UHCI(urb)->tdm_head = host->term_tdm;
URB_UHCI(urb)->qh->element = (phys32_t)
URB_UHCI(urb)->tdm_head->td_phys;
URB_UHCI(urb)->qh->link = UHCI_QH_LINK_TE;
} else {
URB_UHCI(urb)->qh->element = UHCI_QH_LINK_TE;
URB_UHCI(urb)->qh->link = (phys32_t)
URB_UHCI(host->host_skelton
[n_skels - 1])->qh_phys |
UHCI_QH_LINK_QH;
urb->link_next = host->host_skelton[n_skels - 1];
}
host->host_skelton[n_skels] = urb;
}
/* make link to a QH in each frame list entry
according to intervals */
for (frid = 0U; frid < UHCI_NUM_FRAMES; frid++) {
frame_p = (phys32_t *)
(framelist_virt + frid * sizeof(phys32_t));
n_skels = __ffs((frid + 1) |
(1 << (UHCI_NUM_SKELTYPES - 1)));
*frame_p = (phys32_t)
URB_UHCI(host->host_skelton[n_skels])->qh_phys |
UHCI_FRAME_LINK_QH;
}
for (n_skels = 0; n_skels < 2; n_skels++)
host->tailurb[n_skels] = host->host_skelton[0];
spinlock_unlock(&host->lock_hfl);
return 0;
}
static void
check_mm_shm_swap(void) {
size_t nr_free_pages_store = nr_free_pages();
size_t slab_allocated_store = slab_allocated();
int ret, i;
for (i = 0; i < max_swap_offset; i ++) {
assert(mem_map[i] == SWAP_UNUSED);
}
extern struct mm_struct *check_mm_struct;
assert(check_mm_struct == NULL);
struct mm_struct *mm0 = mm_create(), *mm1;
assert(mm0 != NULL && list_empty(&(mm0->mmap_list)));
struct Page *page = alloc_page();
assert(page != NULL);
pgd_t *pgdir = page2kva(page);
memcpy(pgdir, boot_pgdir, PGSIZE);
pgdir[PGX(VPT)] = PADDR(pgdir) | PTE_P | PTE_W;
mm0->pgdir = pgdir;
check_mm_struct = mm0;
lcr3(PADDR(mm0->pgdir));
uint32_t vm_flags = VM_WRITE | VM_READ;
uintptr_t addr0, addr1;
addr0 = 0;
do {
if ((ret = mm_map(mm0, addr0, PTSIZE * 4, vm_flags, NULL)) == 0) {
break;
}
addr0 += PTSIZE;
} while (addr0 != 0);
assert(ret == 0 && addr0 != 0 && mm0->map_count == 1);
ret = mm_unmap(mm0, addr0, PTSIZE * 4);
assert(ret == 0 && mm0->map_count == 0);
struct shmem_struct *shmem = shmem_create(PTSIZE * 2);
assert(shmem != NULL && shmem_ref(shmem) == 0);
// step1: check share memory
struct vma_struct *vma;
addr1 = addr0 + PTSIZE * 2;
ret = mm_map_shmem(mm0, addr0, vm_flags, shmem, &vma);
assert(ret == 0);
assert((vma->vm_flags & VM_SHARE) && vma->shmem == shmem && shmem_ref(shmem) == 1);
ret = mm_map_shmem(mm0, addr1, vm_flags, shmem, &vma);
assert(ret == 0);
assert((vma->vm_flags & VM_SHARE) && vma->shmem == shmem && shmem_ref(shmem) == 2);
// page fault
for (i = 0; i < 4; i ++) {
*(char *)(addr0 + i * PGSIZE) = (char)(i * i);
}
for (i = 0; i < 4; i ++) {
assert(*(char *)(addr1 + i * PGSIZE) == (char)(i * i));
}
for (i = 0; i < 4; i ++) {
*(char *)(addr1 + i * PGSIZE) = (char)(- i * i);
}
for (i = 0; i < 4; i ++) {
assert(*(char *)(addr1 + i * PGSIZE) == (char)(- i * i));
}
// check swap
ret = swap_out_mm(mm0, 8) + swap_out_mm(mm0, 8);
assert(ret == 8 && nr_active_pages == 4 && nr_inactive_pages == 0);
refill_inactive_scan();
assert(nr_active_pages == 0 && nr_inactive_pages == 4);
// write & read again
memset((void *)addr0, 0x77, PGSIZE);
for (i = 0; i < PGSIZE; i ++) {
assert(*(char *)(addr1 + i) == (char)0x77);
}
// check unmap
ret = mm_unmap(mm0, addr1, PGSIZE * 4);
assert(ret == 0);
addr0 += 4 * PGSIZE, addr1 += 4 * PGSIZE;
*(char *)(addr0) = (char)(0xDC);
assert(*(char *)(addr1) == (char)(0xDC));
*(char *)(addr1 + PTSIZE) = (char)(0xDC);
assert(*(char *)(addr0 + PTSIZE) == (char)(0xDC));
cprintf("check_mm_shm_swap: step1, share memory ok.\n");
// setup mm1
mm1 = mm_create();
assert(mm1 != NULL);
page = alloc_page();
assert(page != NULL);
pgdir = page2kva(page);
memcpy(pgdir, boot_pgdir, PGSIZE);
pgdir[PGX(VPT)] = PADDR(pgdir) | PTE_P | PTE_W;
mm1->pgdir = pgdir;
ret = dup_mmap(mm1, mm0);
assert(ret == 0 && shmem_ref(shmem) == 4);
// switch to mm1
check_mm_struct = mm1;
lcr3(PADDR(mm1->pgdir));
for (i = 0; i < 4; i ++) {
*(char *)(addr0 + i * PGSIZE) = (char)(0x57 + i);
}
for (i = 0; i < 4; i ++) {
assert(*(char *)(addr1 + i * PGSIZE) == (char)(0x57 + i));
}
check_mm_struct = mm0;
lcr3(PADDR(mm0->pgdir));
for (i = 0; i < 4; i ++) {
assert(*(char *)(addr0 + i * PGSIZE) == (char)(0x57 + i));
assert(*(char *)(addr1 + i * PGSIZE) == (char)(0x57 + i));
}
swap_out_mm(mm1, 4);
exit_mmap(mm1);
free_page(kva2page(mm1->pgdir));
mm_destroy(mm1);
assert(shmem_ref(shmem) == 2);
cprintf("check_mm_shm_swap: step2, dup_mmap ok.\n");
// free memory
check_mm_struct = NULL;
lcr3(boot_cr3);
exit_mmap(mm0);
free_page(kva2page(mm0->pgdir));
mm_destroy(mm0);
refill_inactive_scan();
page_launder();
for (i = 0; i < max_swap_offset; i ++) {
assert(mem_map[i] == SWAP_UNUSED);
}
assert(nr_free_pages_store == nr_free_pages());
assert(slab_allocated_store == slab_allocated());
cprintf("check_mm_shm_swap() succeeded.\n");
}
static void skd_request_fn(struct request_queue *q)
{
struct skd_device *skdev = q->queuedata;
struct skd_fitmsg_context *skmsg = NULL;
struct fit_msg_hdr *fmh = NULL;
struct skd_request_context *skreq;
struct request *req = NULL;
struct skd_scsi_request *scsi_req;
struct page *page;
unsigned long io_flags;
int error;
u32 lba;
u32 count;
int data_dir;
u32 be_lba;
u32 be_count;
u64 be_dmaa;
u64 cmdctxt;
u32 timo_slot;
void *cmd_ptr;
int flush, fua;
if (skdev->state != SKD_DRVR_STATE_ONLINE) {
skd_request_fn_not_online(q);
return;
}
if (blk_queue_stopped(skdev->queue)) {
if (skdev->skmsg_free_list == NULL ||
skdev->skreq_free_list == NULL ||
skdev->in_flight >= skdev->queue_low_water_mark)
/* There is still some kind of shortage */
return;
queue_flag_clear(QUEUE_FLAG_STOPPED, skdev->queue);
}
/*
* Stop conditions:
* - There are no more native requests
* - There are already the maximum number of requests in progress
* - There are no more skd_request_context entries
* - There are no more FIT msg buffers
*/
for (;; ) {
flush = fua = 0;
req = blk_peek_request(q);
/* Are there any native requests to start? */
if (req == NULL)
break;
lba = (u32)blk_rq_pos(req);
count = blk_rq_sectors(req);
data_dir = rq_data_dir(req);
io_flags = req->cmd_flags;
if (io_flags & REQ_FLUSH)
flush++;
if (io_flags & REQ_FUA)
fua++;
pr_debug("%s:%s:%d new req=%p lba=%u(0x%x) "
"count=%u(0x%x) dir=%d\n",
skdev->name, __func__, __LINE__,
req, lba, lba, count, count, data_dir);
/* At this point we know there is a request */
/* Are too many requets already in progress? */
if (skdev->in_flight >= skdev->cur_max_queue_depth) {
pr_debug("%s:%s:%d qdepth %d, limit %d\n",
skdev->name, __func__, __LINE__,
skdev->in_flight, skdev->cur_max_queue_depth);
break;
}
/* Is a skd_request_context available? */
skreq = skdev->skreq_free_list;
if (skreq == NULL) {
pr_debug("%s:%s:%d Out of req=%p\n",
skdev->name, __func__, __LINE__, q);
break;
}
SKD_ASSERT(skreq->state == SKD_REQ_STATE_IDLE);
SKD_ASSERT((skreq->id & SKD_ID_INCR) == 0);
/* Now we check to see if we can get a fit msg */
if (skmsg == NULL) {
if (skdev->skmsg_free_list == NULL) {
pr_debug("%s:%s:%d Out of msg\n",
skdev->name, __func__, __LINE__);
break;
}
}
skreq->flush_cmd = 0;
skreq->n_sg = 0;
skreq->sg_byte_count = 0;
skreq->discard_page = 0;
/*
* OK to now dequeue request from q.
*
* At this point we are comitted to either start or reject
* the native request. Note that skd_request_context is
* available but is still at the head of the free list.
*/
blk_start_request(req);
skreq->req = req;
skreq->fitmsg_id = 0;
/* Either a FIT msg is in progress or we have to start one. */
if (skmsg == NULL) {
/* Are there any FIT msg buffers available? */
skmsg = skdev->skmsg_free_list;
if (skmsg == NULL) {
pr_debug("%s:%s:%d Out of msg skdev=%p\n",
skdev->name, __func__, __LINE__,
skdev);
break;
}
SKD_ASSERT(skmsg->state == SKD_MSG_STATE_IDLE);
SKD_ASSERT((skmsg->id & SKD_ID_INCR) == 0);
skdev->skmsg_free_list = skmsg->next;
skmsg->state = SKD_MSG_STATE_BUSY;
skmsg->id += SKD_ID_INCR;
/* Initialize the FIT msg header */
fmh = (struct fit_msg_hdr *)skmsg->msg_buf;
memset(fmh, 0, sizeof(*fmh));
fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT;
skmsg->length = sizeof(*fmh);
}
skreq->fitmsg_id = skmsg->id;
/*
* Note that a FIT msg may have just been started
* but contains no SoFIT requests yet.
*/
/*
* Transcode the request, checking as we go. The outcome of
* the transcoding is represented by the error variable.
*/
cmd_ptr = &skmsg->msg_buf[skmsg->length];
memset(cmd_ptr, 0, 32);
be_lba = cpu_to_be32(lba);
be_count = cpu_to_be32(count);
be_dmaa = cpu_to_be64((u64)skreq->sksg_dma_address);
cmdctxt = skreq->id + SKD_ID_INCR;
scsi_req = cmd_ptr;
scsi_req->hdr.tag = cmdctxt;
scsi_req->hdr.sg_list_dma_address = be_dmaa;
if (data_dir == READ)
skreq->sg_data_dir = SKD_DATA_DIR_CARD_TO_HOST;
else
skreq->sg_data_dir = SKD_DATA_DIR_HOST_TO_CARD;
if (io_flags & REQ_DISCARD) {
page = alloc_page(GFP_ATOMIC | __GFP_ZERO);
if (!page) {
pr_err("request_fn:Page allocation failed.\n");
skd_end_request(skdev, skreq, -ENOMEM);
break;
}
skreq->discard_page = 1;
req->completion_data = page;
skd_prep_discard_cdb(scsi_req, skreq, page, lba, count);
} else if (flush == SKD_FLUSH_ZERO_SIZE_FIRST) {
skd_prep_zerosize_flush_cdb(scsi_req, skreq);
SKD_ASSERT(skreq->flush_cmd == 1);
} else {
skd_prep_rw_cdb(scsi_req, data_dir, lba, count);
}
if (fua)
scsi_req->cdb[1] |= SKD_FUA_NV;
if (!req->bio)
goto skip_sg;
error = skd_preop_sg_list(skdev, skreq);
if (error != 0) {
/*
* Complete the native request with error.
* Note that the request context is still at the
* head of the free list, and that the SoFIT request
* was encoded into the FIT msg buffer but the FIT
* msg length has not been updated. In short, the
* only resource that has been allocated but might
* not be used is that the FIT msg could be empty.
*/
pr_debug("%s:%s:%d error Out\n",
skdev->name, __func__, __LINE__);
skd_end_request(skdev, skreq, error);
continue;
}
skip_sg:
scsi_req->hdr.sg_list_len_bytes =
cpu_to_be32(skreq->sg_byte_count);
/* Complete resource allocations. */
skdev->skreq_free_list = skreq->next;
skreq->state = SKD_REQ_STATE_BUSY;
skreq->id += SKD_ID_INCR;
skmsg->length += sizeof(struct skd_scsi_request);
fmh->num_protocol_cmds_coalesced++;
/*
* Update the active request counts.
* Capture the timeout timestamp.
*/
skreq->timeout_stamp = skdev->timeout_stamp;
timo_slot = skreq->timeout_stamp & SKD_TIMEOUT_SLOT_MASK;
skdev->timeout_slot[timo_slot]++;
skdev->in_flight++;
pr_debug("%s:%s:%d req=0x%x busy=%d\n",
skdev->name, __func__, __LINE__,
skreq->id, skdev->in_flight);
/*
* If the FIT msg buffer is full send it.
*/
if (skmsg->length >= SKD_N_FITMSG_BYTES ||
fmh->num_protocol_cmds_coalesced >= skd_max_req_per_msg) {
skd_send_fitmsg(skdev, skmsg);
skmsg = NULL;
fmh = NULL;
}
}
/*
* Is a FIT msg in progress? If it is empty put the buffer back
* on the free list. If it is non-empty send what we got.
* This minimizes latency when there are fewer requests than
* what fits in a FIT msg.
*/
if (skmsg != NULL) {
/* Bigger than just a FIT msg header? */
if (skmsg->length > sizeof(struct fit_msg_hdr)) {
pr_debug("%s:%s:%d sending msg=%p, len %d\n",
skdev->name, __func__, __LINE__,
skmsg, skmsg->length);
skd_send_fitmsg(skdev, skmsg);
} else {
/*
* The FIT msg is empty. It means we got started
* on the msg, but the requests were rejected.
*/
skmsg->state = SKD_MSG_STATE_IDLE;
skmsg->id += SKD_ID_INCR;
skmsg->next = skdev->skmsg_free_list;
skdev->skmsg_free_list = skmsg;
}
skmsg = NULL;
fmh = NULL;
}
/*
* If req is non-NULL it means there is something to do but
* we are out of a resource.
*/
if (req)
blk_stop_queue(skdev->queue);
}
static void
check_mm_swap(void) {
size_t nr_free_pages_store = nr_free_pages();
size_t slab_allocated_store = slab_allocated();
int ret, i, j;
for (i = 0; i < max_swap_offset; i ++) {
assert(mem_map[i] == SWAP_UNUSED);
}
extern struct mm_struct *check_mm_struct;
assert(check_mm_struct == NULL);
// step1: check mm_map
struct mm_struct *mm0 = mm_create(), *mm1;
assert(mm0 != NULL && list_empty(&(mm0->mmap_list)));
uintptr_t addr0, addr1;
addr0 = 0;
do {
ret = mm_map(mm0, addr0, PUSIZE, 0, NULL);
assert(ret == (USER_ACCESS(addr0, addr0 + PUSIZE)) ? 0 : -E_INVAL);
addr0 += PUSIZE;
} while (!((addr0 >> 48) & 0xFFFF));
addr0 = 0;
for (i = 0; i < 1024; i ++, addr0 += PUSIZE) {
ret = mm_map(mm0, addr0, PGSIZE, 0, NULL);
assert(ret == -E_INVAL);
}
mm_destroy(mm0);
mm0 = mm_create();
assert(mm0 != NULL && list_empty(&(mm0->mmap_list)));
addr0 = 0, i = 0;
do {
ret = mm_map(mm0, addr0, PUSIZE - PGSIZE, 0, NULL);
assert(ret == (USER_ACCESS(addr0, addr0 + PUSIZE)) ? 0 : -E_INVAL);
if (ret == 0) {
i ++;
}
addr0 += PUSIZE;
} while (!((addr0 >> 48) & 0xFFFF));
addr0 = 0, j = 0;
do {
addr0 += PUSIZE - PGSIZE;
ret = mm_map(mm0, addr0, PGSIZE, 0, NULL);
assert(ret == (USER_ACCESS(addr0, addr0 + PGSIZE)) ? 0 : -E_INVAL);
if (ret == 0) {
j ++;
}
addr0 += PGSIZE;
} while (!((addr0 >> 48) & 0xFFFF));
assert(j + 1 >= i);
mm_destroy(mm0);
assert(nr_free_pages_store == nr_free_pages());
assert(slab_allocated_store == slab_allocated());
cprintf("check_mm_swap: step1, mm_map ok.\n");
// step2: check page fault
mm0 = mm_create();
assert(mm0 != NULL && list_empty(&(mm0->mmap_list)));
// setup page table
struct Page *page = alloc_page();
assert(page != NULL);
pgd_t *pgdir = page2kva(page);
memcpy(pgdir, boot_pgdir, PGSIZE);
pgdir[PGX(VPT)] = PADDR(pgdir) | PTE_P | PTE_W;
// prepare for page fault
mm0->pgdir = pgdir;
check_mm_struct = mm0;
lcr3(PADDR(mm0->pgdir));
uint32_t vm_flags = VM_WRITE | VM_READ;
struct vma_struct *vma;
addr0 = 0;
do {
if ((ret = mm_map(mm0, addr0, PTSIZE, vm_flags, &vma)) == 0) {
break;
}
addr0 += PUSIZE;
} while (!((addr0 >> 48) & 0xFFFF));
assert(ret == 0 && addr0 != 0 && mm0->map_count == 1);
assert(vma->vm_start == addr0 && vma->vm_end == addr0 + PTSIZE);
// check pte entry
pte_t *ptep;
for (addr1 = addr0; addr1 < addr0 + PTSIZE; addr1 += PGSIZE) {
ptep = get_pte(pgdir, addr1, 0);
assert(ptep == NULL);
}
memset((void *)addr0, 0xEF, PGSIZE * 2);
ptep = get_pte(pgdir, addr0, 0);
assert(ptep != NULL && (*ptep & PTE_P));
ptep = get_pte(pgdir, addr0 + PGSIZE, 0);
assert(ptep != NULL && (*ptep & PTE_P));
ret = mm_unmap(mm0, - PTSIZE, PTSIZE);
assert(ret == -E_INVAL);
ret = mm_unmap(mm0, addr0 + PTSIZE, PGSIZE);
assert(ret == 0);
addr1 = addr0 + PTSIZE / 2;
ret = mm_unmap(mm0, addr1, PGSIZE);
assert(ret == 0 && mm0->map_count == 2);
ret = mm_unmap(mm0, addr1 + 2 * PGSIZE, PGSIZE * 4);
assert(ret == 0 && mm0->map_count == 3);
ret = mm_map(mm0, addr1, PGSIZE * 6, 0, NULL);
assert(ret == -E_INVAL);
ret = mm_map(mm0, addr1, PGSIZE, 0, NULL);
assert(ret == 0 && mm0->map_count == 4);
ret = mm_map(mm0, addr1 + 2 * PGSIZE, PGSIZE * 4, 0, NULL);
assert(ret == 0 && mm0->map_count == 5);
ret = mm_unmap(mm0, addr1 + PGSIZE / 2, PTSIZE / 2 - PGSIZE);
assert(ret == 0 && mm0->map_count == 1);
addr1 = addr0 + PGSIZE;
for (i = 0; i < PGSIZE; i ++) {
assert(*(char *)(addr1 + i) == (char)0xEF);
}
ret = mm_unmap(mm0, addr1 + PGSIZE / 2, PGSIZE / 4);
assert(ret == 0 && mm0->map_count == 2);
ptep = get_pte(pgdir, addr0, 0);
assert(ptep != NULL && (*ptep & PTE_P));
ptep = get_pte(pgdir, addr0 + PGSIZE, 0);
assert(ptep != NULL && *ptep == 0);
ret = mm_map(mm0, addr1, PGSIZE, vm_flags, NULL);
memset((void *)addr1, 0x88, PGSIZE);
assert(*(char *)addr1 == (char)0x88 && mm0->map_count == 3);
for (i = 1; i < 16; i += 2) {
ret = mm_unmap(mm0, addr0 + PGSIZE * i, PGSIZE);
assert(ret == 0);
if (i < 8) {
ret = mm_map(mm0, addr0 + PGSIZE * i, PGSIZE, 0, NULL);
assert(ret == 0);
}
}
assert(mm0->map_count == 13);
ret = mm_unmap(mm0, addr0 + PGSIZE * 2, PTSIZE - PGSIZE * 2);
assert(ret == 0 && mm0->map_count == 2);
ret = mm_unmap(mm0, addr0, PGSIZE * 2);
assert(ret == 0 && mm0->map_count == 0);
cprintf("check_mm_swap: step2, mm_unmap ok.\n");
// step3: check exit_mmap
ret = mm_map(mm0, addr0, PTSIZE, vm_flags, NULL);
assert(ret == 0);
for (i = 0, addr1 = addr0; i < 4; i ++, addr1 += PGSIZE) {
*(char *)addr1 = (char)0xFF;
}
exit_mmap(mm0);
for (i = 0; i < PGX(KERNBASE); i ++) {
assert(pgdir[i] == 0);
}
cprintf("check_mm_swap: step3, exit_mmap ok.\n");
// step4: check dup_mmap
for (i = 0; i < max_swap_offset; i ++) {
assert(mem_map[i] == SWAP_UNUSED);
}
ret = mm_map(mm0, addr0, PTSIZE, vm_flags, NULL);
assert(ret != 0);
addr1 = addr0;
for (i = 0; i < 4; i ++, addr1 += PGSIZE) {
*(char *)addr1 = (char)(i * i);
}
ret = 0;
ret += swap_out_mm(mm0, 10);
ret += swap_out_mm(mm0, 10);
assert(ret == 4);
for (; i < 8; i ++, addr1 += PGSIZE) {
*(char *)addr1 = (char)(i * i);
}
// setup mm1
mm1 = mm_create();
assert(mm1 != NULL);
page = alloc_page();
assert(page != NULL);
pgdir = page2kva(page);
memcpy(pgdir, boot_pgdir, PGSIZE);
pgdir[PGX(VPT)] = PADDR(pgdir) | PTE_P | PTE_W;
mm1->pgdir = pgdir;
ret = dup_mmap(mm1, mm0);
assert(ret == 0);
// switch to mm1
check_mm_struct = mm1;
lcr3(PADDR(mm1->pgdir));
addr1 = addr0;
for (i = 0; i < 8; i ++, addr1 += PGSIZE) {
assert(*(char *)addr1 == (char)(i * i));
*(char *)addr1 = (char)0x88;
}
// switch to mm0
check_mm_struct = mm0;
lcr3(PADDR(mm0->pgdir));
addr1 = addr0;
for (i = 0; i < 8; i ++, addr1 += PGSIZE) {
assert(*(char *)addr1 == (char)(i * i));
}
// switch to boot_cr3
check_mm_struct = NULL;
lcr3(boot_cr3);
// free memory
exit_mmap(mm0);
exit_mmap(mm1);
free_page(kva2page(mm0->pgdir));
mm_destroy(mm0);
free_page(kva2page(mm1->pgdir));
mm_destroy(mm1);
cprintf("check_mm_swap: step4, dup_mmap ok.\n");
refill_inactive_scan();
page_launder();
for (i = 0; i < max_swap_offset; i ++) {
assert(mem_map[i] == SWAP_UNUSED);
}
assert(nr_free_pages_store == nr_free_pages());
assert(slab_allocated_store == slab_allocated());
cprintf("check_mm_swap() succeeded.\n");
}
struct page *_raid_page_alloc(void)
{
return alloc_page(GFP_KERNEL);
}
int __init efi_setup_page_tables(unsigned long pa_memmap, unsigned num_pages)
{
unsigned long pfn, text, pf;
struct page *page;
unsigned npages;
pgd_t *pgd = efi_mm.pgd;
if (efi_enabled(EFI_OLD_MEMMAP))
return 0;
/*
* It can happen that the physical address of new_memmap lands in memory
* which is not mapped in the EFI page table. Therefore we need to go
* and ident-map those pages containing the map before calling
* phys_efi_set_virtual_address_map().
*/
pfn = pa_memmap >> PAGE_SHIFT;
pf = _PAGE_NX | _PAGE_RW | _PAGE_ENC;
if (kernel_map_pages_in_pgd(pgd, pfn, pa_memmap, num_pages, pf)) {
pr_err("Error ident-mapping new memmap (0x%lx)!\n", pa_memmap);
return 1;
}
/*
* Certain firmware versions are way too sentimential and still believe
* they are exclusive and unquestionable owners of the first physical page,
* even though they explicitly mark it as EFI_CONVENTIONAL_MEMORY
* (but then write-access it later during SetVirtualAddressMap()).
*
* Create a 1:1 mapping for this page, to avoid triple faults during early
* boot with such firmware. We are free to hand this page to the BIOS,
* as trim_bios_range() will reserve the first page and isolate it away
* from memory allocators anyway.
*/
pf = _PAGE_RW;
if (sev_active())
pf |= _PAGE_ENC;
if (kernel_map_pages_in_pgd(pgd, 0x0, 0x0, 1, pf)) {
pr_err("Failed to create 1:1 mapping for the first page!\n");
return 1;
}
/*
* When making calls to the firmware everything needs to be 1:1
* mapped and addressable with 32-bit pointers. Map the kernel
* text and allocate a new stack because we can't rely on the
* stack pointer being < 4GB.
*/
if (!IS_ENABLED(CONFIG_EFI_MIXED) || efi_is_native())
return 0;
page = alloc_page(GFP_KERNEL|__GFP_DMA32);
if (!page)
panic("Unable to allocate EFI runtime stack < 4GB\n");
efi_scratch.phys_stack = virt_to_phys(page_address(page));
efi_scratch.phys_stack += PAGE_SIZE; /* stack grows down */
npages = (_etext - _text) >> PAGE_SHIFT;
text = __pa(_text);
pfn = text >> PAGE_SHIFT;
pf = _PAGE_RW | _PAGE_ENC;
if (kernel_map_pages_in_pgd(pgd, pfn, text, npages, pf)) {
pr_err("Failed to map kernel text 1:1\n");
return 1;
}
return 0;
}
static void
check_swap(void)
{
//backup mem env
int ret, count = 0, total = 0, i;
list_entry_t *le = &free_list;
while ((le = list_next(le)) != &free_list) {
struct Page *p = le2page(le, page_link);
assert(PageProperty(p));
count ++, total += p->property;
}
assert(total == nr_free_pages());
cprintf("BEGIN check_swap: count %d, total %d\n",count,total);
//now we set the phy pages env
struct mm_struct *mm = mm_create();
assert(mm != NULL);
extern struct mm_struct *check_mm_struct;
assert(check_mm_struct == NULL);
check_mm_struct = mm;
pde_t *pgdir = mm->pgdir = boot_pgdir;
assert(pgdir[0] == 0);
struct vma_struct *vma = vma_create(BEING_CHECK_VALID_VADDR, CHECK_VALID_VADDR, VM_WRITE | VM_READ);
assert(vma != NULL);
insert_vma_struct(mm, vma);
//setup the temp Page Table vaddr 0~4MB
cprintf("setup Page Table for vaddr 0X1000, so alloc a page\n");
pte_t *temp_ptep=NULL;
temp_ptep = get_pte(mm->pgdir, BEING_CHECK_VALID_VADDR, 1);
assert(temp_ptep!= NULL);
cprintf("setup Page Table vaddr 0~4MB OVER!\n");
for (i=0;i<CHECK_VALID_PHY_PAGE_NUM;i++) {
check_rp[i] = alloc_page();
assert(check_rp[i] != NULL );
assert(!PageProperty(check_rp[i]));
}
list_entry_t free_list_store = free_list;
list_init(&free_list);
assert(list_empty(&free_list));
//assert(alloc_page() == NULL);
unsigned int nr_free_store = nr_free;
nr_free = 0;
for (i=0;i<CHECK_VALID_PHY_PAGE_NUM;i++) {
free_pages(check_rp[i],1);
}
assert(nr_free==CHECK_VALID_PHY_PAGE_NUM);
cprintf("set up init env for check_swap begin!\n");
//setup initial vir_page<->phy_page environment for page relpacement algorithm
pgfault_num=0;
check_content_set();
assert( nr_free == 0);
for(i = 0; i<MAX_SEQ_NO ; i++)
swap_out_seq_no[i]=swap_in_seq_no[i]=-1;
for (i= 0;i<CHECK_VALID_PHY_PAGE_NUM;i++) {
check_ptep[i]=0;
check_ptep[i] = get_pte(pgdir, (i+1)*0x1000, 0);
//cprintf("i %d, check_ptep addr %x, value %x\n", i, check_ptep[i], *check_ptep[i]);
assert(check_ptep[i] != NULL);
assert(pte2page(*check_ptep[i]) == check_rp[i]);
assert((*check_ptep[i] & PTE_P));
}
cprintf("set up init env for check_swap over!\n");
// now access the virt pages to test page relpacement algorithm
ret=check_content_access();
assert(ret==0);
//restore kernel mem env
for (i=0;i<CHECK_VALID_PHY_PAGE_NUM;i++) {
free_pages(check_rp[i],1);
}
//free_page(pte2page(*temp_ptep));
free_page(pa2page(pgdir[0]));
pgdir[0] = 0;
mm->pgdir = NULL;
mm_destroy(mm);
check_mm_struct = NULL;
nr_free = nr_free_store;
free_list = free_list_store;
le = &free_list;
while ((le = list_next(le)) != &free_list) {
struct Page *p = le2page(le, page_link);
count --, total -= p->property;
}
cprintf("count is %d, total is %d\n",count,total);
//assert(count == 0);
cprintf("check_swap() succeeded!\n");
}
