Date: Wed, 3 Mar 2004 12:09:30 +0100 From: Pawel Malachowski <pawmal-posting@freebsd.lublin.pl> To: Tim Robbins <tjr@freebsd.org> Cc: FreeBSD-gnats-submit@freebsd.org Subject: Re: Using read-only NULLFS leads to panic. gdb output included, easy to reproduce. Message-ID: <20040303110930.GA35449@shellma.zin.lublin.pl> In-Reply-To: <20040303002536.GA59500@cat.robbins.dropbear.id.au> References: <20040302213936.216CB5F103@shellma.zin.lublin.pl> <20040303002536.GA59500@cat.robbins.dropbear.id.au>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Mar 03, 2004 at 11:25:36AM +1100, Tim Robbins wrote: > There are known bugs in nullfs in all 4.x releases to date, and in 5.0. > If I have time, I may MFC the fixes some time before 4.10 is released. Sounds interesting. > Can you reproduce these problems on 5.1 or 5.2? Procedure: mount_null -o ro /usr/ports /mnt/1 find /usr/ports -type f -perm -u+s & find /mnt/1 -type f -perm -u+s & ... <about 30-40 processess of find(1) was enough> I was not able to reproduce panic on 5.1-RELEASE system. However, I've reproduced second problem described in this PR, which happened to me on 4.9-STABLE once. *Every* time (I've tried four times) 5.1-RELEASE becomes unusable and non-rebootable, all find(1) processes stuck into 'ufs' state, CPU is 100% idle, no disk activity, other processess, when trying to read from filesystem, become frozen. I'm able to switch beetween consoles and to observe top(1) (started earlier). Ctrl+Alt+Del does not work. One mount_null is sufficient, but more than one null mount seems (or it was an accident) to decrease the time we have to wait. -- Paweł Małachowski
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040303110930.GA35449>