On Tue, May 29, 2018 at 12:17:42PM -0700, Omar Sandoval wrote: > On Mon, May 28, 2018 at 06:57:59PM +0200, David Sterba wrote: > > On Tue, May 22, 2018 at 01:47:22PM -0400, Josef Bacik wrote: > > > From: Josef Bacik <jbacik@xxxxxx> > > > > > > We don't actually need this. It used to be in place for O_SYNC writes, > > > but we've used the normal fsync() path for that for years now. The > > > other case we hit this is through sync(), which will commit the > > > transaction anyway. All this does is make us commit the transaction a > > > bunch for no reason, and it could deadlock with delayed iput's. > > > > In what way does it deadlock with delayed iput? > > Here's an example stack trace: Aha, so that's an actual bugfix. The changelog should have been stated the other way around: there's a deadlock scenario and can be fixed by removing the whole function because there's another mechanism to achieve the same O_SYNC behaviour. Please update and resend. -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
