Re: [RFC] Breaking data compatibility with userspace bzlib

David S. Miller (davem@redhat.com)
Fri, 20 Jun 2003 13:23:14 -0700 (PDT)


From: Jörn Engel <joern@wohnheim.fh-wedel.de>
Date: Fri, 20 Jun 2003 21:56:58 +0200

On Fri, 20 June 2003 12:45:10 -0700, David S. Miller wrote:
> Therefore we couldn't make the compress cryptolib interface
> take scatterlists elements either, which is a huge problem.

Is there a reason to use the zlib and nothing but the zlib for the
cryptolib? RFCs 1950 - 1952? Or would any form of compression do, in
principle at least?

Because it's not very sensible to have 2 pieces of code that
do exactly the same thing? :-)

In the worst case, I consider it not too hard to add a wrapper
interface to the zlib to do the scatter-gather handling.

It's not that simple.

Is the wrapper called in IRQ or BH or user context? This matters
to determine what kind of kmap() operations you must use in the
wrapper.

That's merely the beginning of the problems with this kind of idea.

Don't let my statements stall your work, it was just a FYI and
something we should take care of in the 2.7.x timeframe not now.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/