From 1a1e4530eacca37e85a4d66a164273c7dba9110c Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Mon, 8 Sep 2014 08:04:00 +0900 Subject: [PATCH] bdi: explain the dirty list transferring in bdi_destroy() bdi_destroy() has code to transfer the remaining dirty inodes to the default_backing_dev_info; however, given the shutdown sequence, it isn't clear how such condition would happen. Also, it isn't a full solution as the transferred inodes stlil point to the bdi which is being destroyed. Operations on those inodes can end up accessing already released fields such as the percpu stat fields. Digging through the history, it seems that the code was added as a quick workaround for a bug report without fully root-causing the issue. We probably want to remove the code in time but for now let's add a comment noting that it is a quick workaround. Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- mm/backing-dev.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/mm/backing-dev.c b/mm/backing-dev.c index cb7c5e323814..b65fe93ad612 100644 --- a/mm/backing-dev.c +++ b/mm/backing-dev.c @@ -475,8 +475,17 @@ void bdi_destroy(struct backing_dev_info *bdi) int i; /* - * Splice our entries to the default_backing_dev_info, if this - * bdi disappears + * Splice our entries to the default_backing_dev_info. This + * condition shouldn't happen. @wb must be empty at this point and + * dirty inodes on it might cause other issues. This workaround is + * added by ce5f8e779519 ("writeback: splice dirty inode entries to + * default bdi on bdi_destroy()") without root-causing the issue. + * + * http://lkml.kernel.org/g/1253038617-30204-11-git-send-email-jens.axboe@oracle.com + * http://thread.gmane.org/gmane.linux.file-systems/35341/focus=35350 + * + * We should probably add WARN_ON() to find out whether it still + * happens and track it down if so. */ if (bdi_has_dirty_io(bdi)) { struct bdi_writeback *dst = &default_backing_dev_info.wb; -- 2.20.1