[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Jffs2 compression and paging
I apologize if this is a duplicate post of a question.
A developer I am working with attempted to send this
question but he was having problems subscribing to the
list for some reason. I am on the list so I will
repost his question.
Here is the original message:
I have an obscure question about jffs2...
Though I couldn't find any documentation indicating
that it does, jffs2
appears to implement some kind of default compression.
If I take a
collection of files on my ext3 filesystem and build a
jffs2 image from
them using mkfs.jffs2, the resulting image is
substantially smaller than
the collection of files by themselves. Is there some
compression going on or is it just an artifact of the
difference in the way the
filesystems store the data?
If there is no compression going on, then my next set
of questions are
moot, but if there is...
If there is some kind of compression, how does it
interact with the
demand paging system? When the pager requests the 3rd
page of a file,
does the entire file (or some chunk of it larger than
just that page) have
to be uncompressed before the page can be delivered?
If so, are the
uncompressed pages just left in ram? What happens if
there isn't enough
ram to uncompress into?
The reason I'm asking is because I've observed
behaviour that is
inconsistent with my understanding of demand paging in
linux. When I load a
program, the entire program and all libraries that it
appear to get loaded instead of just the initial page
or two. The memory
usage of the process appears consistent with this
determination, and I'm
certain that at the point the code starts waiting for
input that it has
not attempted to execute code in every page of the
library (or itself).
I don't have a good explanation for it and I
ultimately became curious
if there's some special interaction with jffs2 that
was causing it.
Anyone have any ideas?
Do you Yahoo!?
Faith Hill - Exclusive Performances, Videos & More
To unsubscribe from this list: send the line "unsubscribe jffs-dev" in
the body of a message to email@example.com