Subject: Re: procfs broken on -current?
To: Kouichirou Hiratsuka <email@example.com>
From: Chavdar Ivanov <firstname.lastname@example.org>
Date: 02/12/2007 10:01:53
On 12/02/07, Kouichirou Hiratsuka <email@example.com> wrote:
> At Sun, 11 Feb 2007 17:19:14 +0000,
> Andrew Doran wrote:
> > On Sat, Feb 10, 2007 at 11:33:50PM +0900, Kouichirou Hiratsuka wrote:
> > > After accessing /proc/curproc/file, the process is hung up.
> > > Does anyone else seeing this?
> > procfs_lookup() was leaking a reference to the process when called against
> > curproc, so it would get stuck draining references and couldn't exit. This
> > should be fixed now:
> > spanners$ ls -la /proc/curproc/file
> > -r-xr-xr-x 1 root wheel 23633 Feb 10 12:32 /proc/curproc/file
> > spanners$
> It works fine for me too.
> However, I got a kernel panic with the following command.
I am getting exactly the same with
ktruss java -version
(well, the addresses are different, if that matters - I think not).
I can write down the output from bt if necessary - this machine does
not have a serial port to redirect the debugger unfortunately. I might
try the same under VMware and get the output there.
With a slightly earlier verson of the kernel it froze itself trying to
open an entry under /emul/linux/proc.
> % ls -la /proc/*/*
> Mutex error: mutex_vector_enter: locking against myself
> lock address : 0x00000000c0505e70
> current cpu : 0
> current lwp : 0x00000000cbc6f2e0
> owner field : 0x00000000cbc6f2e0 wait/spin: 0/0
> panic: lock error
> Stopped in pid 120.1 (ls) at netbsd:cpu_Debugger+0x4: leave
> best regards.
> Kouichirou Hiratsuka