This is the mail archive of the ecos-discuss@sources.redhat.com mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: JFFS2 "eating memory" when creating files


ons, 10.03.2004 kl. 17.25 skrev Thomas Koeller:
> Øyvind Harboe wrote:
> > > Of course, you will have to estimate the maximum number of nodes
> > > you will ever use.
> >
> > I shudder at the thought of estimating a fixed limit for something that
> > I don't understand.
> 
> I have dealt with JFFS2 for quite some time now, and I have reached the
> conclusion that making proper use of it in a resource-constrained environment
> positively requires a thorough understanding of its inner workings.

Thats my experience as well.

However, I claim that a number of issues for resource constrained
environments(like the ability to turn off compression) have been put to
bed. Why stop here?

If JFFS2 RAM usage is at a minimum when freshly mounted, it appears to
me that it should be possible to have JFFS2 return to the "freshly
mounted state" if all files are closed.

Effectively doing an unmount+mount when filehandles/locks=0. 

I seem to remember that there was some talk about current directory
unmount+mount problems, but I don't need or want a current directory so
it should be possible to configure eCos to build without a current
directory concept.

In fact, unmount checks if it is possible to unmount before it does so,
so I may be able to simply add the unmount+mount to the tail of jffs2
close.

Øyvind



-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]