From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Theodore Y. Ts'o" Subject: Re: [PATCH 5/7] ext4: pass -ESHUTDOWN code to jbd2 layer Date: Thu, 22 Mar 2018 11:26:45 -0400 Message-ID: <20180322152644.GE2852@thunk.org> References: <20180220023038.19883-1-tytso@mit.edu> <20180220023038.19883-6-tytso@mit.edu> <20180306171041.mz7qlpeikih6mtlg@quack2.suse.cz> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Ext4 Developers List , stable@vger.kernel.org To: Jan Kara Return-path: Content-Disposition: inline In-Reply-To: <20180306171041.mz7qlpeikih6mtlg@quack2.suse.cz> Sender: stable-owner@vger.kernel.org List-Id: linux-ext4.vger.kernel.org On Tue, Mar 06, 2018 at 06:10:41PM +0100, Jan Kara wrote: > Is it really correct that once the filesystem gets shutdown you clear the > previous error from the journal? Because if we hit some real fs corruption, > the journal gets aborted, and then someone calls ext4_shutdown(), we'd > clear that error which looks like a bug to me because that shutdown hardly > fixes the fs corruption... That's not what the code does. If journal->j_errno is set, then we won't clear it, for precisely what concern you've articulated. - Ted