On Tue, Jul 16, 2013 at 12:49:42PM -0400, Theodore Ts'o wrote: > On Tue, Jul 16, 2013 at 05:46:58PM +0200, Jan Kara wrote: > > and ext4_setattr() does (again under i_mutex): > > ext4_inode_block_unlocked_dio(inode); > > inode_dio_wait(inode); > > ext4_inode_resume_unlocked_dio(inode); > > Ah, I missed this; thanks for pointing this out. I took a closer look, and there are other code paths which call ext4_truncate() which don't call ext4_inode_block_unlocked_dio() and inode_dio_wait(). In particular, the orphan cleanup code and the SETFLAGS ioctl. I suspect the right answer is to move these these calls, and possibly also the truncate_pagecache() call. The other thing that we probably will want to do as a cleanup for the next merge window is to look at merging more of the code paths for ext4_punch_hole() and ext4_truncate(). Cheeers, - Ted -- To unsubscribe from this list: send the line "unsubscribe stable" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html