From: Goldwyn Rodrigues <rgoldwyn@xxxxxxxx>
We replace the existing entry to the newly allocated one
in case of CoW. Also, we mark the entry as PAGECACHE_TAG_TOWRITE
so writeback marks this entry as writeprotected. This
helps us snapshots so new write pagefaults after snapshots
trigger a CoW.
btrfs does not support hugepages so we don't handle PMD.
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@xxxxxxxx>
---
fs/dax.c | 40 ++++++++++++++++++++++++++++++----------
1 file changed, 30 insertions(+), 10 deletions(-)
diff --git a/fs/dax.c b/fs/dax.c
index 949a40cf1fe7..e6174c0c669a 100644
--- a/fs/dax.c
+++ b/fs/dax.c
@@ -705,6 +705,9 @@ static int copy_user_dax(struct block_device *bdev, struct dax_device *dax_dev,
return 0;
}
+#define DAX_IF_DIRTY (1ULL << 0)
+#define DAX_IF_COW (1ULL << 1)
+
/*
* By this point grab_mapping_entry() has ensured that we have a locked entry
* of the appropriate size so we don't have to worry about downgrading PMDs to
@@ -714,14 +717,17 @@ static int copy_user_dax(struct block_device *bdev, struct dax_device *dax_dev,
*/
static void *dax_insert_entry(struct xa_state *xas,
struct address_space *mapping, struct vm_fault *vmf,
- void *entry, pfn_t pfn, unsigned long flags, bool dirty)
+ void *entry, pfn_t pfn, unsigned long flags,
+ unsigned int insert_flags)
{
void *new_entry = dax_make_entry(pfn, flags);
+ bool dirty = insert_flags & DAX_IF_DIRTY;
+ bool cow = insert_flags & DAX_IF_COW;
if (dirty)
__mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
- if (dax_is_zero_entry(entry) && !(flags & DAX_ZERO_PAGE)) {
+ if (cow || (dax_is_zero_entry(entry) && !(flags & DAX_ZERO_PAGE))) {
unsigned long index = xas->xa_index;
/* we are replacing a zero page with block mapping */
if (dax_is_pmd_entry(entry))
@@ -733,7 +739,7 @@ static void *dax_insert_entry(struct xa_state *xas,
xas_reset(xas);
xas_lock_irq(xas);
- if (dax_is_zero_entry(entry) || dax_is_empty_entry(entry)) {
+ if (cow || dax_is_zero_entry(entry) || dax_is_empty_entry(entry)) {
void *old;
dax_disassociate_entry(entry, mapping, false);
@@ -757,6 +763,9 @@ static void *dax_insert_entry(struct xa_state *xas,
if (dirty)
xas_set_mark(xas, PAGECACHE_TAG_DIRTY);
+ if (cow)
+ xas_set_mark(xas, PAGECACHE_TAG_TOWRITE);
+
xas_unlock_irq(xas);
return entry;
}
@@ -1044,7 +1053,7 @@ static vm_fault_t dax_load_hole(struct xa_state *xas,
vm_fault_t ret;
*entry = dax_insert_entry(xas, mapping, vmf, *entry, pfn,
- DAX_ZERO_PAGE, false);
+ DAX_ZERO_PAGE, 0);
ret = vmf_insert_mixed(vmf->vma, vaddr, pfn);
trace_dax_load_hole(inode, vmf, ret);
@@ -1313,6 +1322,7 @@ static vm_fault_t dax_iomap_pte_fault(struct vm_fault *vmf, pfn_t *pfnp,
void *entry;
pfn_t pfn;
void *kaddr;
+ unsigned long insert_flags = 0;
trace_dax_pte_fault(inode, vmf, ret);
/*
@@ -1405,8 +1415,10 @@ static vm_fault_t dax_iomap_pte_fault(struct vm_fault *vmf, pfn_t *pfnp,
if (error < 0)
goto error_finish_iomap;
- entry = dax_insert_entry(&xas, mapping, vmf, entry, pfn,
- 0, write && !sync);
+ if (write && !sync)
+ insert_flags |= DAX_IF_DIRTY;
+ entry = dax_insert_entry(&xas, mapping, vmf, entry, pfn, 0,
+ insert_flags);
if (srcmap.type != IOMAP_HOLE) {
error = dax_copy_edges(pos, PAGE_SIZE, &srcmap, kaddr,
@@ -1438,8 +1450,10 @@ static vm_fault_t dax_iomap_pte_fault(struct vm_fault *vmf, pfn_t *pfnp,
goto finish_iomap;
case IOMAP_UNWRITTEN:
- if (srcmap.type != IOMAP_HOLE)
+ if (srcmap.type != IOMAP_HOLE) {
+ insert_flags |= DAX_IF_COW;
goto cow;
+ }
/*FALLTHRU*/
case IOMAP_HOLE:
if (!write) {
@@ -1497,7 +1511,7 @@ static vm_fault_t dax_pmd_load_hole(struct xa_state *xas, struct vm_fault *vmf,
pfn = page_to_pfn_t(zero_page);
*entry = dax_insert_entry(xas, mapping, vmf, *entry, pfn,
- DAX_PMD | DAX_ZERO_PAGE, false);
+ DAX_PMD | DAX_ZERO_PAGE, 0);
if (arch_needs_pgtable_deposit()) {
pgtable = pte_alloc_one(vma->vm_mm);
@@ -1549,6 +1563,7 @@ static vm_fault_t dax_iomap_pmd_fault(struct vm_fault *vmf, pfn_t *pfnp,
int error;
pfn_t pfn;
void *kaddr;
+ unsigned long insert_flags = 0;
/*
* Check whether offset isn't beyond end of file now. Caller is
@@ -1636,8 +1651,11 @@ static vm_fault_t dax_iomap_pmd_fault(struct vm_fault *vmf, pfn_t *pfnp,
if (error < 0)
goto finish_iomap;
+ if (write && !sync)
+ insert_flags |= DAX_IF_DIRTY;
+
entry = dax_insert_entry(&xas, mapping, vmf, entry, pfn,
- DAX_PMD, write && !sync);
+ DAX_PMD, insert_flags);
if (srcmap.type != IOMAP_HOLE) {
error = dax_copy_edges(pos, PMD_SIZE, &srcmap, kaddr,
@@ -1664,8 +1682,10 @@ static vm_fault_t dax_iomap_pmd_fault(struct vm_fault *vmf, pfn_t *pfnp,
result = vmf_insert_pfn_pmd(vmf, pfn, write);
break;
case IOMAP_UNWRITTEN:
- if (srcmap.type != IOMAP_HOLE)
+ if (srcmap.type != IOMAP_HOLE) {
+ insert_flags |= DAX_IF_COW;
goto cow;
+ }
/*FALLTHRU*/
case IOMAP_HOLE:
if (WARN_ON_ONCE(write))
--
2.23.0
