Linux 2.5 kernel summit

[M K Saravanan <mksarav@xxxxxxxxxxxxxxxxx>]

(http://www.lwn.net/2001/features/KernelSummit/):

=====
The Linux 2.5 kernel summit

The "Linux 2.5 Kernel Summit" is a two-day affair, held in San Jose,
California; the organizers are Usenix and OSDN, and Ted Ts'o in
particular. The purpose of this event was to get the core kernel
hackers together in one place to meet each other and to look toward
the imminent 2.5 development series. The attendee list shows 65
hackers, almost all of whom have names familiar to those who follow
kernel development. It's an impressive gathering; more than one
attendee has been heard to fret over what would happen to Linux if the
big Silicon Valley earthquake were to strike during this event.

[...]

High-performance database requirements

The first presentation was by Lance Larsh of Oracle who, essentially,
provided a laundry list of changes and features Oracle would like to
see in order to get better performance out of high-end, large database
servers.

[...]

SCTP

SCTP is the "Stream Control Transmission Protocol," defined by
RFC2960. It is intended to be a new, high-level protocol with many of
the advantages of both TCP and UDP, and with some additional features,
such as dynamic network failover for multihomed hosts. SCTP also
incorporates multiple streams of messages into a single connection,
which can bring performance benefits in situations where multiple
streams are used.

A beginning SCTP implementation was presented by La Monte Yarroll,
along with a description of the changes that are needed in the
kernel. Among other things, a new version of the bind() system call is
needed which can bind a socket to multiple addresses, so that the
failover mechanism can work.

Block layer redesign

It is generally accepted that the block I/O layer needs some major
changes in the 2.5 development series. The 2.4 implementation is
cleaner and faster than its predecessors, but a number of problems
remain. Stephen Tweedie presented a list of things he thought needed
to be addressed as a way of starting discussion on this topic.

[...]

Integrating high-performance filesystems

Steve Lord of SGI presented some of the advanced features of the XFS
filesystem, with the idea that, perhaps, some of them should be moved
into the Linux VFS layer.

[...]

The network driver API

Jamal Hadi Salim led a session describing changes to the network
driver interface. The stock Linux kernel performs poorly under very
heavy network loads - as the number of packets received goes up, the
number of packets actually processed begins to drop, until it
approaches zero in especially hostile situations. The desire, of
course, is to change that.

[...]

Hot plug devices

A common theme with modern hardware is the ability to attach and
detach devices while the system is running - hot plugging. PCMCIA
cards have worked that way for some time; now devices connected by
USB, Firewire, SCSI, and hot swap PCI all are hot pluggable. The
kernel needs to be able to deal with this kind of environment; USB
maintainer Johannes Erdfelt gave a presentation on how that is being
done.

[...]

Friday evening BOFs

Several birds-of-a-feather session were held after the reception on
Friday. One involved a presentation of the Subversion source control
system by Greg Stein. Subversion has the goal of replacing CVS over
the next few years; it is built on top of Apache and the WebDAV
protocol, and appears to have a number of nice features. The group is
pushing toward a 1.0 release later this year.

Linus attended the session and seemed impressed; he also commented
that what he really liked was the name of the project.

[...]

The kernel build system

A number of bleary-eyed kernel hackers noted a lack of enthusiasm for
a discussion on kernel makefiles first thing Saturday morning - it's
just not one of the sexier topics. The presentation by Keith Owens and
Eric Raymond was interesting, however. The complexity of the kernel is
such that even the code for putting it all together is difficult. The
kbuild system is often seen as a bit of a hack; starting in 2.5, it's
likely to be a far better system.

[...]

Virtual memory

Rik van Riel presented his thoughts on what needs to be done with
virtual memory management in Linux. This talk covered a number of
different areas where improvements could be made. Memory management is
a complicated task, and there will always be possible improvements.

[...]

NSA Linux

Peter Loscocco from the National Security Agency presented the design
of the mandatory access control system in its SE Linux
distribution. Much of it was an overview of the system itself; the
same information can be found on the NSA SE Linux site, so I'll not
repeat it here. The executive summary is that SE Linux has
implemented:

* A great many check points where authorization to perform a particular
task is controlled.
      
* A security manager process which implements the actual authorization
policy.

The separation of the checks and the policy mechanism is an important
aspect of the system - different sites can implement very different
access policies using the same system.

[...]

Asynchronous I/O

Ben LaHaise presented his asynchronous I/O implementation. True async
I/O in the kernel has been a wishlist item for a long time; it appears
that it will become real, finally, in 2.5.

[...]

Power management

Andy Grover presented the work that is being done on power management
in Linux. The old APM code, of course, is obsolete - the industry is
moving over to ACPI, which is a much more comprehensive standard. It
is also large, the specification for ACPI fills a full 500 pages.

Intel's ACPI implementation is out there now, and is being used by
FreeBSD. They are currently waiting for the 2.5 fork to submit it for
Linux.

[...]

Bitkeeper

Larry McVoy had been unable to attend the BOF on source control, so he
presented his Bitkeeper system in a special session at the end of
Saturday. He was able to give an interesting presentation and focus on
the features of Bitkeeper without getting into the licensing issues.

[...]
=====