From: Chuck Lever <chuck.lever@xxxxxxxxxx> [ Upstream commit 3f6d810665dfde0d33785420618ceb03fba0619d ] Liam and Matthew say that once the RCU read lock is released, xa_state is not safe to re-use for the next xas_find() call. But the RCU read lock must be released on each loop iteration so that dput(), which might_sleep(), can be called safely. Thus we are forced to walk the offset tree with fresh state for each directory entry. xa_find() can do this for us, though it might be a little less efficient than maintaining xa_state locally. We believe that in the current code base, inode->i_rwsem provides protection for the xa_state maintained in offset_iterate_dir(). However, there is no guarantee that will continue to be the case in the future. Since offset_iterate_dir() doesn't build xa_state locally any more, there's no longer a strong need for offset_find_next(). Clean up by rolling these two helpers together. Suggested-by: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx> Message-ID: <170785993027.11135.8830043889278631735.stgit@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> Signed-off-by: Chuck Lever <chuck.lever@xxxxxxxxxx> Link: https://lore.kernel.org/r/170820142021.6328.15047865406275957018.stgit@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx Reviewed-by: Jan Kara <jack@xxxxxxx> Signed-off-by: Christian Brauner <brauner@xxxxxxxxxx> Signed-off-by: Chuck Lever <chuck.lever@xxxxxxxxxx> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> --- fs/libfs.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) --- a/fs/libfs.c +++ b/fs/libfs.c @@ -401,12 +401,13 @@ static loff_t offset_dir_llseek(struct f return vfs_setpos(file, offset, U32_MAX); } -static struct dentry *offset_find_next(struct xa_state *xas) +static struct dentry *offset_find_next(struct offset_ctx *octx, loff_t offset) { struct dentry *child, *found = NULL; + XA_STATE(xas, &octx->xa, offset); rcu_read_lock(); - child = xas_next_entry(xas, U32_MAX); + child = xas_next_entry(&xas, U32_MAX); if (!child) goto out; spin_lock(&child->d_lock); @@ -429,12 +430,11 @@ static bool offset_dir_emit(struct dir_c static void *offset_iterate_dir(struct inode *inode, struct dir_context *ctx) { - struct offset_ctx *so_ctx = inode->i_op->get_offset_ctx(inode); - XA_STATE(xas, &so_ctx->xa, ctx->pos); + struct offset_ctx *octx = inode->i_op->get_offset_ctx(inode); struct dentry *dentry; while (true) { - dentry = offset_find_next(&xas); + dentry = offset_find_next(octx, ctx->pos); if (!dentry) return ERR_PTR(-ENOENT); @@ -443,8 +443,8 @@ static void *offset_iterate_dir(struct i break; } + ctx->pos = dentry2offset(dentry) + 1; dput(dentry); - ctx->pos = xas.xa_index + 1; } return NULL; } Patches currently in stable-queue which might be from cel@xxxxxxxxxx are queue-6.6/libfs-replace-simple_offset-end-of-directory-detection.patch queue-6.6/libfs-re-arrange-locking-in-offset_iterate_dir.patch queue-6.6/libfs-add-simple_offset_empty.patch queue-6.6/shmem-fix-shmem_rename2.patch queue-6.6/revert-libfs-add-simple_offset_empty.patch queue-6.6/libfs-use-d_children-list-to-iterate-simple_offset-directories.patch queue-6.6/libfs-define-a-minimum-directory-offset.patch queue-6.6/libfs-return-enospc-when-the-directory-offset-range-is-exhausted.patch queue-6.6/libfs-fix-simple_offset_rename_exchange.patch queue-6.6/libfs-add-simple_offset_rename-api.patch
