Commit 9fc747f6 authored by Matthew Wilcox's avatar Matthew Wilcox

dax: Convert dax writeback to XArray

Use XArray iteration instead of a pagevec.
Signed-off-by: default avatarMatthew Wilcox <willy@infradead.org>
parent 07f2d89c
...@@ -997,11 +997,9 @@ static void dax_entry_mkclean(struct address_space *mapping, pgoff_t index, ...@@ -997,11 +997,9 @@ static void dax_entry_mkclean(struct address_space *mapping, pgoff_t index,
i_mmap_unlock_read(mapping); i_mmap_unlock_read(mapping);
} }
static int dax_writeback_one(struct dax_device *dax_dev, static int dax_writeback_one(struct xa_state *xas, struct dax_device *dax_dev,
struct address_space *mapping, pgoff_t index, void *entry) struct address_space *mapping, void *entry)
{ {
struct radix_tree_root *pages = &mapping->i_pages;
void *entry2, **slot;
unsigned long pfn; unsigned long pfn;
long ret = 0; long ret = 0;
size_t size; size_t size;
...@@ -1013,29 +1011,35 @@ static int dax_writeback_one(struct dax_device *dax_dev, ...@@ -1013,29 +1011,35 @@ static int dax_writeback_one(struct dax_device *dax_dev,
if (WARN_ON(!xa_is_value(entry))) if (WARN_ON(!xa_is_value(entry)))
return -EIO; return -EIO;
xa_lock_irq(pages); if (unlikely(dax_is_locked(entry))) {
entry2 = get_unlocked_mapping_entry(mapping, index, &slot); void *old_entry = entry;
/* Entry got punched out / reallocated? */
if (!entry2 || WARN_ON_ONCE(!xa_is_value(entry2))) entry = get_unlocked_entry(xas);
goto put_unlocked;
/* /* Entry got punched out / reallocated? */
* Entry got reallocated elsewhere? No need to writeback. We have to if (!entry || WARN_ON_ONCE(!xa_is_value(entry)))
* compare pfns as we must not bail out due to difference in lockbit goto put_unlocked;
* or entry type. /*
*/ * Entry got reallocated elsewhere? No need to writeback.
if (dax_to_pfn(entry2) != dax_to_pfn(entry)) * We have to compare pfns as we must not bail out due to
goto put_unlocked; * difference in lockbit or entry type.
if (WARN_ON_ONCE(dax_is_empty_entry(entry) || */
dax_is_zero_entry(entry))) { if (dax_to_pfn(old_entry) != dax_to_pfn(entry))
ret = -EIO; goto put_unlocked;
goto put_unlocked; if (WARN_ON_ONCE(dax_is_empty_entry(entry) ||
dax_is_zero_entry(entry))) {
ret = -EIO;
goto put_unlocked;
}
/* Another fsync thread may have already done this entry */
if (!xas_get_mark(xas, PAGECACHE_TAG_TOWRITE))
goto put_unlocked;
} }
/* Another fsync thread may have already written back this entry */
if (!radix_tree_tag_get(pages, index, PAGECACHE_TAG_TOWRITE))
goto put_unlocked;
/* Lock the entry to serialize with page faults */ /* Lock the entry to serialize with page faults */
entry = lock_slot(mapping, slot); dax_lock_entry(xas, entry);
/* /*
* We can clear the tag now but we have to be careful so that concurrent * We can clear the tag now but we have to be careful so that concurrent
* dax_writeback_one() calls for the same index cannot finish before we * dax_writeback_one() calls for the same index cannot finish before we
...@@ -1043,8 +1047,8 @@ static int dax_writeback_one(struct dax_device *dax_dev, ...@@ -1043,8 +1047,8 @@ static int dax_writeback_one(struct dax_device *dax_dev,
* at the entry only under the i_pages lock and once they do that * at the entry only under the i_pages lock and once they do that
* they will see the entry locked and wait for it to unlock. * they will see the entry locked and wait for it to unlock.
*/ */
radix_tree_tag_clear(pages, index, PAGECACHE_TAG_TOWRITE); xas_clear_mark(xas, PAGECACHE_TAG_TOWRITE);
xa_unlock_irq(pages); xas_unlock_irq(xas);
/* /*
* Even if dax_writeback_mapping_range() was given a wbc->range_start * Even if dax_writeback_mapping_range() was given a wbc->range_start
...@@ -1056,7 +1060,7 @@ static int dax_writeback_one(struct dax_device *dax_dev, ...@@ -1056,7 +1060,7 @@ static int dax_writeback_one(struct dax_device *dax_dev,
pfn = dax_to_pfn(entry); pfn = dax_to_pfn(entry);
size = PAGE_SIZE << dax_entry_order(entry); size = PAGE_SIZE << dax_entry_order(entry);
dax_entry_mkclean(mapping, index, pfn); dax_entry_mkclean(mapping, xas->xa_index, pfn);
dax_flush(dax_dev, page_address(pfn_to_page(pfn)), size); dax_flush(dax_dev, page_address(pfn_to_page(pfn)), size);
/* /*
* After we have flushed the cache, we can clear the dirty tag. There * After we have flushed the cache, we can clear the dirty tag. There
...@@ -1064,16 +1068,18 @@ static int dax_writeback_one(struct dax_device *dax_dev, ...@@ -1064,16 +1068,18 @@ static int dax_writeback_one(struct dax_device *dax_dev,
* the pfn mappings are writeprotected and fault waits for mapping * the pfn mappings are writeprotected and fault waits for mapping
* entry lock. * entry lock.
*/ */
xa_lock_irq(pages); xas_reset(xas);
radix_tree_tag_clear(pages, index, PAGECACHE_TAG_DIRTY); xas_lock_irq(xas);
xa_unlock_irq(pages); xas_store(xas, entry);
trace_dax_writeback_one(mapping->host, index, size >> PAGE_SHIFT); xas_clear_mark(xas, PAGECACHE_TAG_DIRTY);
put_locked_mapping_entry(mapping, index); dax_wake_entry(xas, entry, false);
trace_dax_writeback_one(mapping->host, xas->xa_index,
size >> PAGE_SHIFT);
return ret; return ret;
put_unlocked: put_unlocked:
put_unlocked_mapping_entry(mapping, index, entry2); put_unlocked_entry(xas, entry);
xa_unlock_irq(pages);
return ret; return ret;
} }
...@@ -1085,13 +1091,13 @@ static int dax_writeback_one(struct dax_device *dax_dev, ...@@ -1085,13 +1091,13 @@ static int dax_writeback_one(struct dax_device *dax_dev,
int dax_writeback_mapping_range(struct address_space *mapping, int dax_writeback_mapping_range(struct address_space *mapping,
struct block_device *bdev, struct writeback_control *wbc) struct block_device *bdev, struct writeback_control *wbc)
{ {
XA_STATE(xas, &mapping->i_pages, wbc->range_start >> PAGE_SHIFT);
struct inode *inode = mapping->host; struct inode *inode = mapping->host;
pgoff_t start_index, end_index; pgoff_t end_index = wbc->range_end >> PAGE_SHIFT;
pgoff_t indices[PAGEVEC_SIZE];
struct dax_device *dax_dev; struct dax_device *dax_dev;
struct pagevec pvec; void *entry;
bool done = false; int ret = 0;
int i, ret = 0; unsigned int scanned = 0;
if (WARN_ON_ONCE(inode->i_blkbits != PAGE_SHIFT)) if (WARN_ON_ONCE(inode->i_blkbits != PAGE_SHIFT))
return -EIO; return -EIO;
...@@ -1103,41 +1109,29 @@ int dax_writeback_mapping_range(struct address_space *mapping, ...@@ -1103,41 +1109,29 @@ int dax_writeback_mapping_range(struct address_space *mapping,
if (!dax_dev) if (!dax_dev)
return -EIO; return -EIO;
start_index = wbc->range_start >> PAGE_SHIFT; trace_dax_writeback_range(inode, xas.xa_index, end_index);
end_index = wbc->range_end >> PAGE_SHIFT;
trace_dax_writeback_range(inode, start_index, end_index);
tag_pages_for_writeback(mapping, start_index, end_index);
pagevec_init(&pvec); tag_pages_for_writeback(mapping, xas.xa_index, end_index);
while (!done) {
pvec.nr = find_get_entries_tag(mapping, start_index,
PAGECACHE_TAG_TOWRITE, PAGEVEC_SIZE,
pvec.pages, indices);
if (pvec.nr == 0) xas_lock_irq(&xas);
xas_for_each_marked(&xas, entry, end_index, PAGECACHE_TAG_TOWRITE) {
ret = dax_writeback_one(&xas, dax_dev, mapping, entry);
if (ret < 0) {
mapping_set_error(mapping, ret);
break; break;
for (i = 0; i < pvec.nr; i++) {
if (indices[i] > end_index) {
done = true;
break;
}
ret = dax_writeback_one(dax_dev, mapping, indices[i],
pvec.pages[i]);
if (ret < 0) {
mapping_set_error(mapping, ret);
goto out;
}
} }
start_index = indices[pvec.nr - 1] + 1; if (++scanned % XA_CHECK_SCHED)
continue;
xas_pause(&xas);
xas_unlock_irq(&xas);
cond_resched();
xas_lock_irq(&xas);
} }
out: xas_unlock_irq(&xas);
put_dax(dax_dev); put_dax(dax_dev);
trace_dax_writeback_range_done(inode, start_index, end_index); trace_dax_writeback_range_done(inode, xas.xa_index, end_index);
return (ret < 0 ? ret : 0); return ret;
} }
EXPORT_SYMBOL_GPL(dax_writeback_mapping_range); EXPORT_SYMBOL_GPL(dax_writeback_mapping_range);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment