On 5/5/22 9:25 PM, Jeff Layton wrote:
On Thu, 2022-05-05 at 20:47 +0800, Xiubo Li wrote:
Use the VM_BUG_ON_FOLIO macro to get more infomation when we hit
the BUG_ON, and continue the loop when seeing the incorrect none
write opcode in writepages_finish().
URL: https://tracker.ceph.com/issues/55421
Signed-off-by: Xiubo Li <xiubli@xxxxxxxxxx>
---
fs/ceph/addr.c | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index e52b62407b10..d4bcef1d9549 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -122,7 +122,7 @@ static bool ceph_dirty_folio(struct address_space *mapping, struct folio *folio)
* Reference snap context in folio->private. Also set
* PagePrivate so that we get invalidate_folio callback.
*/
- BUG_ON(folio_get_private(folio));
+ VM_BUG_ON_FOLIO(folio_get_private(folio), folio);
folio_attach_private(folio, snapc);
return ceph_fscache_dirty_folio(mapping, folio);
@@ -733,8 +733,11 @@ static void writepages_finish(struct ceph_osd_request *req)
/* clean all pages */
for (i = 0; i < req->r_num_ops; i++) {
- if (req->r_ops[i].op != CEPH_OSD_OP_WRITE)
- break;
+ if (req->r_ops[i].op != CEPH_OSD_OP_WRITE) {
+ pr_warn("%s incorrect op %d req %p index %d tid %llu\n",
+ __func__, req->r_ops[i].op, req, i, req->r_tid);
+ continue;
A break here is probably fine. We don't expect to see any non OP_WRITE
ops in here, so if we do I don't see the point in continuing.
I am afraid if the non OP_WRITE op is in middle of the r_ops[], for the
following OP_WRITE ops we may miss some pages with the dirty bit cleared
but with the private data kept ?
For this check I didn't in which case will the non OP_WRITE ops will
happen unless the request is corrupted.
-- Xiubo
+ }
osd_data = osd_req_op_extent_osd_data(req, i);
BUG_ON(osd_data->type != CEPH_OSD_DATA_TYPE_PAGES);
Otherwise looks fine.
