On Mon, Dec 06, 2004 at 02:22:51PM -0600, Frederick Bruckman wrote:
> On Mon, 6 Dec 2004, Jaromir Dolecek wrote:
> 
> >Frederick Bruckman wrote:
> >
> >>the intuitive thing would be to flush the buffer cache *after* the
> >>filesystem is marked "rw", if that's possible.
> 
> Actually, I meant to say "ro" there. I guess you got that.
> 
> >Yeah, the intuitive thing would be to flush the buffer cache
> >after filesystem is marked ro. Is there any reason we don't
> >do that? AFAICS ffs only flushes file system data on update to 'ro',
> >but not metadata.
> 
> Perhaps that code path simply hasn't been adapted to soft updates. 
> There didn't used to be any metadata in the buffer cache, ever.

Is there an easy fix or workaround for this?

I'm afraid my Soekris boxes are badly affected by this bug.  I have an
app on Soekris boards that does mount -u -w /, writes some configuration
files, and does mount -u -r /.  Sometimes the app reboots the box directly
after.  The last file created is oftentimes truncated to 0 after reboot.

Other data points:

        1 sometimes the box is unresponsive for several seconds
          after i type 'reboot'; sometimes it prints 'syncing disks 3
          3 3 3 3 3 3 3 ...' before the BIOS boot starts

        2 i have read in another thread that it can take up to 20 seconds
          (!!) for a CF card to synchronize its data buffers with the
          non-volatile Flash---could that explain (1) ?

        3 i am using null mounts; i write the config files through the
          null mounts.

Dave

-- 
David Young             OJC Technologies
dyoung@ojctech.com      Urbana, IL * (217) 278-3933