This patch extends the rcu regions, so that lookup followed by a read or
write of a page is done inside rcu read lock. This si be needed for the
following patch that enables discard.
Note that we also replace "BUG_ON(!page);" with "if (page) ..." in
copy_to_brd - the page may be NULL if write races with discard. In this
situation, the result is undefined, so we can actually skip the write
operation at all.
Signed-off-by: Mikulas Patocka <mpato...@redhat.com>
---
drivers/block/brd.c | 59 +++++++++++++++++++++++-----------------------------
1 file changed, 27 insertions(+), 32 deletions(-)
Index: linux-2.6/drivers/block/brd.c
===================================================================
--- linux-2.6.orig/drivers/block/brd.c
+++ linux-2.6/drivers/block/brd.c
@@ -50,31 +50,12 @@ struct brd_device {
/*
* Look up and return a brd's page for a given sector.
+ * This must be called with the rcu lock held.
*/
static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector)
{
- pgoff_t idx;
- struct page *page;
-
- /*
- * The page lifetime is protected by the fact that we have opened the
- * device node -- brd pages will never be deleted under us, so we
- * don't need any further locking or refcounting.
- *
- * This is strictly true for the radix-tree nodes as well (ie. we
- * don't actually need the rcu_read_lock()), however that is not a
- * documented feature of the radix-tree API so it is better to be
- * safe here (we don't have total exclusion from radix tree updates
- * here, only deletes).
- */
- rcu_read_lock();
- idx = sector >> PAGE_SECTORS_SHIFT; /* sector to page index */
- page = radix_tree_lookup(&brd->brd_pages, idx);
- rcu_read_unlock();
-
- BUG_ON(page && page->index != idx);
-
- return page;
+ pgoff_t idx = sector >> PAGE_SECTORS_SHIFT; /* sector to page index */
+ return radix_tree_lookup(&brd->brd_pages, idx);
}
/*
@@ -88,7 +69,9 @@ static bool brd_insert_page(struct brd_d
struct page *page;
gfp_t gfp_flags;
+ rcu_read_lock();
page = brd_lookup_page(brd, sector);
+ rcu_read_unlock();
if (page)
return true;
@@ -198,23 +181,29 @@ static void copy_to_brd(struct brd_devic
size_t copy;
copy = min_t(size_t, n, PAGE_SIZE - offset);
- page = brd_lookup_page(brd, sector);
- BUG_ON(!page);
- dst = kmap_atomic(page);
- memcpy(dst + offset, src, copy);
- kunmap_atomic(dst);
+ rcu_read_lock();
+ page = brd_lookup_page(brd, sector);
+ if (page) {
+ dst = kmap_atomic(page);
+ memcpy(dst + offset, src, copy);
+ kunmap_atomic(dst);
+ }
+ rcu_read_unlock();
if (copy < n) {
src += copy;
sector += copy >> SECTOR_SHIFT;
copy = n - copy;
- page = brd_lookup_page(brd, sector);
- BUG_ON(!page);
- dst = kmap_atomic(page);
- memcpy(dst, src, copy);
- kunmap_atomic(dst);
+ rcu_read_lock();
+ page = brd_lookup_page(brd, sector);
+ if (page) {
+ dst = kmap_atomic(page);
+ memcpy(dst, src, copy);
+ kunmap_atomic(dst);
+ }
+ rcu_read_unlock();
}
}
@@ -230,6 +219,8 @@ static void copy_from_brd(void *dst, str
size_t copy;
copy = min_t(size_t, n, PAGE_SIZE - offset);
+
+ rcu_read_lock();
page = brd_lookup_page(brd, sector);
if (page) {
src = kmap_atomic(page);
@@ -237,11 +228,14 @@ static void copy_from_brd(void *dst, str
kunmap_atomic(src);
} else
memset(dst, 0, copy);
+ rcu_read_unlock();
if (copy < n) {
dst += copy;
sector += copy >> SECTOR_SHIFT;
copy = n - copy;
+
+ rcu_read_lock();
page = brd_lookup_page(brd, sector);
if (page) {
src = kmap_atomic(page);
@@ -249,6 +243,7 @@ static void copy_from_brd(void *dst, str
kunmap_atomic(src);
} else
memset(dst, 0, copy);
+ rcu_read_unlock();
}
}
--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel
