Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 17 Jun 2004 04:14:46 -0400
From:      Brian Fundakowski Feldman <green@FreeBSD.org>
To:        current@FreeBSD.org
Cc:        sos@FreeBSD.org
Subject:   Re: ATAPICAM hangs with cdparanoia
Message-ID:  <20040617081446.GA15918@green.homeunix.org>
In-Reply-To: <20040615055940.GA1016@green.homeunix.org>
References:  <20040615055940.GA1016@green.homeunix.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Jun 15, 2004 at 01:59:40AM -0400, Brian Feldman wrote:
> Trying to rip a brand new CD with cdparanoia today, I've run into hangs
> during the rip (read/seek within a track) process.  Sometimes it's a
> "cbwait" hang, but it seems I can also get it to lock hard at the console.
> Does anyone have experience debugging ATAPICAM/ATA/CAM that will come in
> handy trying to figure out why this happens?
> 
> The most information I have right now is that it was a generic IO (read)
> CCB, and when I looked at the union ccb inside the process doing the
> xpt read request, "entries" was 0. Doing a camcontrol reset all locked
> things up at that point.

Alright, so I figured out parts of the problem.  After upgrading my
Pioneer DVR-107's firmware *, I was able to get FreeBSD to boot most
of the time (instead of hang/error out most of the time), with a CD
in the drive.  Next, I found out that the timeouts were 1000 times too
long with atapicam, so fixing that certainly helped.

There's still a lock-up issue somewhere, and it's hard to pin down.
Cdparanoia eventually falls back to trying to read where it's erroring
out one sector at a time, then after that finally fails it ends up
closing the device or resetting it or something (obtuse code, that
program is), and this is causing the REAL hang.

I can't get any further because even though I have CAMDEBUG in the
kernel, it does me a fat lot of good when -CURRENT can't even output
printf() to the console anymore.  I'd like to see the request that was
made right before the system hangs fully, but I never see bold output
on ttyv0 anymore on any of my systems and I REALLY NEED THIS UNBROKEN :(
Is this related to kern.log_console_output, which is undocumented?

Any help would once again be much appreciated.

* I ended up having to take mtools, mformat a new 2.88MB disk with
  an MS-DOS boot disk boot sector and set of system files, mcopy over
  and set up HIMEM64.EXE, TDSK.EXE, and the firmware flash stuff,
  and mkisofs -b it into a bootable CD image....

-- 
Brian Fundakowski Feldman                           \'[ FreeBSD ]''''''''''\
  <> green@FreeBSD.org                               \  The Power to Serve! \
 Opinions expressed are my own.                       \,,,,,,,,,,,,,,,,,,,,,,\



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040617081446.GA15918>