https://bugzilla.kernel.org/show_bug.cgi?id=219306 --- Comment #4 from Theodore Tso (tytso@xxxxxxx) --- "Won't be called most of the time" is rather understating the case. The WARN_ON should never, *EVER* fire. If it does we have a kernel bug, and if the user has panic_on_warn set, the kernel will crash. Also, please note that the WARN_ON would triggers when we have a inode-specific inconsistency. In the extremely unlikely case that the WARN_ON triggers, the doom loop that you are worried about would only happen if (a) panic_on_warn is disabled, and (b) the inconsistent data state happens to be the journald's log file. If the file system has millions of inodes, we're talking about a probability of one in millions, even if the kernel is buggy and triggers at all. -- You may reply to this email to add a comment. You are receiving this mail because: You are watching the assignee of the bug.
