Space Loss in JFFS on an EST-SBC8260

Space Loss in JFFS on an EST-SBC8260

[Scott Ikenaga]

We are using JFFS on EST-SBC8260. The problem we are facing now is that
the space can't be reclaimed after files being deleted. So we are
running out of space as shown by "du".

Questions:
(1)Is this a known problem? If so, is there a fix?
(2)Could this contribute to another problem where the system becomes
   very sluggish?

RE: Space Loss in JFFS on an EST-SBC8260

[Allen Curtis]

We have the exact same problem. (We are using the MTD/JFFS with the HHL 2.0
Professional distribution) I retrieved a copy of the latest code from CVS
and posted a message earlier today asking how to insert this code into a HHL
source tree. Do you have an idea? Have you tried the latest version of the
code with the same results?

Any help would be appreciated.

> -----Original Message-----
> From: linux-mtd-admin@lists.infradead.org
> [mailto:linux-mtd-admin@lists.infradead.org]On Behalf Of Scott Ikenaga
> Sent: Thursday, January 24, 2002 8:29 AM
> To: linux-mtd@lists.infradead.org
> Subject: Space Loss in JFFS on an EST-SBC8260
>
>
> We are using JFFS on EST-SBC8260. The problem we are facing now is that
> the space can't be reclaimed after files being deleted. So we are
> running out of space as shown by "du".
>
> Questions:
> (1)Is this a known problem? If so, is there a fix?
> (2)Could this contribute to another problem where the system becomes
>    very sluggish?
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: Space Loss in JFFS on an EST-SBC8260

[Andrew Dixon]

Scott Ikenaga wrote:
> 
> We are using JFFS on EST-SBC8260. The problem we are facing now is that
> the space can't be reclaimed after files being deleted.

I don't believe that space should be reclaimed right away.  You only get
space back after the garbage collector has been through and cleaned up
the flash.  (The preceding is mostly a guess.  You should get
independent conformation.)

 So we are
> running out of space as shown by "du".

What's your kernel configuration?

> 
> Questions:
> (1)Is this a known problem? If so, is there a fix?

I haven't seen it before.

> (2)Could this contribute to another problem where the system becomes
>    very sluggish?

It doesn't seem very likely to me but it's hard to tell without knowing
what the configuration of your system is.  You could try running off of
an NFS root filesystem (if that's an option) and see if you performance
is better.

-- 
Andrew Dixon
Software Engineer
Seranoa Networks
978.897.3434 x231

RE: Space Loss in JFFS on an EST-SBC8260

[Allen Curtis]

We are having the same problem. We are using HHL 2.0 which is Linux 2.4.2.
According to some other feedback we have received, the latest version of the
code does a better job of reporting the correct size. I have not completed
the testing to tell you about the performance issue.

Background on performance:
Since it does not appear that the JFFS space is reclaimed, when the flash
gets "full" and it is forced to reclaim space, the system gets very slow.
(unacceptably slow when this is your primary FS)

Note:
If you do a directory listing a few times and then look at "du", you should
see the available space change. (at least in our experience) Not sure what
triggers the garbage collection but that seems to force the issue if you are
looking for test cases.

> -----Original Message-----
> From: linux-mtd-admin@lists.infradead.org
> [mailto:linux-mtd-admin@lists.infradead.org]On Behalf Of Andrew Dixon
> Sent: Friday, February 01, 2002 9:46 AM
> To: linux-mtd@lists.infradead.org
> Subject: Re: Space Loss in JFFS on an EST-SBC8260
>
>
> Scott Ikenaga wrote:
> >
> > We are using JFFS on EST-SBC8260. The problem we are facing now is that
> > the space can't be reclaimed after files being deleted.
>
> I don't believe that space should be reclaimed right away.  You only get
> space back after the garbage collector has been through and cleaned up
> the flash.  (The preceding is mostly a guess.  You should get
> independent conformation.)
>
>  So we are
> > running out of space as shown by "du".
>
> What's your kernel configuration?
>
> >
> > Questions:
> > (1)Is this a known problem? If so, is there a fix?
>
> I haven't seen it before.
>
> > (2)Could this contribute to another problem where the system becomes
> >    very sluggish?
>
> It doesn't seem very likely to me but it's hard to tell without knowing
> what the configuration of your system is.  You could try running off of
> an NFS root filesystem (if that's an option) and see if you performance
> is better.
>
> --
> Andrew Dixon
> Software Engineer
> Seranoa Networks
> 978.897.3434 x231
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/