On Tue, Aug 08, 2023 at 09:15:52AM -0700, Christoph Hellwig wrote: > XFS never uses the block device mapping, so there is no point in calling > invalidate_bdev which invalidates said mapping. > > Signed-off-by: Christoph Hellwig <hch@xxxxxx> This is a revert of commit 032e160305f68 ("xfs: invalidate block device page cache during unmount"). How do you propose dealing with the block device mapping being incoherent due to unmount write races so you don't re-break my test system? --D > --- > fs/xfs/xfs_buf.c | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/fs/xfs/xfs_buf.c b/fs/xfs/xfs_buf.c > index 15d1e5a7c2d340..83b8702030f71d 100644 > --- a/fs/xfs/xfs_buf.c > +++ b/fs/xfs/xfs_buf.c > @@ -1944,7 +1944,6 @@ xfs_free_buftarg( > list_lru_destroy(&btp->bt_lru); > > blkdev_issue_flush(btp->bt_bdev); > - invalidate_bdev(btp->bt_bdev); > fs_put_dax(btp->bt_daxdev, btp->bt_mount); > > kmem_free(btp); > -- > 2.39.2 >