Older blog entries for rml (starting at number 32)

kernel:

Still bitching over how to "solve" the interactivity problems. People seem pretty happy with Con's latest patch, but I still prefer a solution without the interactivity estimator, such as Nick Piggin's. Oh well.

64-bit dev_t support seems to be almost there, if not there. I smell 2.6.0 soonish.

procps:

procps 2.0.15 is out. Lots new -- this is a good release. New vmstat(8), with an O(1) algorithm, I/O wait support, unit selection, removal of the dumb "swapped" field, and oh-so-much cleaner code. Also new pmap(1), with new features, Solaris-like formatting, and a GPL license (replacing a BSD-like license).

The future for procps looks good. Next up is cleaning up libproc severely. I have a series of patches to turn it into a Real Library.

schedutils:

schedutils 1.3.3 is out. Small release: support for Alpha and PA-RISC and some cleanup.

Unleash the fury, Mitch (aka, my book):

I wrote a book, Linux Kernel Development. It is a fun and practical guide to the design of the new 2.6 kernel, with an eye toward allowing the reader to learn enough to get some real work done.

Look for it in fine book sellers everywhere, this coming Monday (8 Sept 2003).

It was a lot of fun, although more work than anyone should have to endure... but the end result was well worth it. I hope readers agree.

schedutils:

Got schedutils 1.3.1 and then 1.3.2 out. Also now in Red Hat Rawhide, which is cool.

procps:

Got procps 2.0.14 out. Not much but bug fixes, but the current CVS tree is looking great: finally merged the O(1) vmstat(8), a tons of cleanup, and some more bug fixes. Also, someone is working on a reworked, GPL'ed, pmap(1) and perhaps some other bits, too. Fun.

Kernel:

Kernel is coming along. Please test.

Busy. Little. Bee.

Also, good luck to everyone at Ximian during these exciting times.

In Ottawa for Kernel Summit and OLS.

Kernel Summit was interesting but nothing mind blowing. Linus said some stuff. 2.6 final in a few months. 2.7 starts in a few months. Not too sure what 2.7 needs, aside from the usual "fix the tty layer" bit. Still very fun -- I am glad we do these.

Gave talk at OLS on kernel performance on the desktop and in real-time.

Some might say I drank a little beer.

kernel

Fix scheduler bugs. One more fix pending. Poop.

Apple

Damn I love this Apple. Both the PowerBook and the underlying operating system are a pleasure.

So Apple finally unveiled the G5. 64-bit. Sick HyperTransport interconnect. Half clock speed system bus. Currently up to 2GHz. Runs dual. Amazing SPECfp and - more importantly - great SPECint. Way to go, Apple.

SPECint vs. SPECfp

Speaking of SPEC scores... I see a lot of people on slashdot and elsewhere exclaiming how amazing the SPECfp numbers are for their favorite architecture. Wild claims like SPECfp is all that matters. What crack are these people on? I do not deny for one moment that SPECfp is important, especially for scientific computationally-intensive applications. For example, if I were designing a cluster for physics research, I would pay a lot of attention to SPECfp numbers. I would look at chips like Alpha and POWER and even IA-64 with good SPECfp numbers. But the vast majority of users - and I am talking nearly all desktop, server, embedded, and even many computational powerhouses - are much more dependent on SPECint numbers.

Look at x86. It has awful floating point performance but (with its high clock speeds) excellent integer performance. And, guess what? It performs best at the stuff people care about. So I think some of us need a reality check. For most uses, SPECint is what matters.

schedutils

schedutils 1.3.0 was released last Thursday. Nothing too special. Mostly build fixes and cleanup, to help get schedutils working in Red Hat Rawhide (yay).

preemptive kernel

Despite 2.5 running so amazingly well, and coming out-of-the-box with a preemptive kernel and tons of sweet low latency enhancements, demand for a preempt-kernel patch for 2.4 is unfortunately high enough that I released a patch for 2.4.21 today.

It fixed an awful lot of bugs, both existing preempt-unsafe code and some new things that cropped up in the longish 2.4.21 development cycle. I recommend an upgrade.

Even better, of course, I recommend an upgrade to 2.5. ;)

Apple, PowerBooks, Mac OS X, and Lupus

So I have been contemplating a laptop upgrade for some time, and eyeing an IBM X31 (a nice lightweight Centrino machine). Screw all that.

I got an Apple PowerBook G4 15".

What can I say? It is an amazingly well designed machine and Mac OS X is (brace yourself) really damn nice. Of course, I run many of the usual Unix applications so I am not in a totally foreign land. It is really an amazingly well put together system, from the hardware up through the software. Stuff just works.

In the past, I have said a lot of good things about GNOME. I strongly believe the Windows interface is junk; it is confusing, ugly, and stupid. I really think GNOME has made a lot of advances and is a very powerful, well thought out, intelligently designed system. That said, there are a few things GNOME can learn from Mac OS X. I see jdub has mentioned this in recent postings. A lot of it is just refinement... but Aqua does impress me.

I have to say the battery life (~3.5 hours on average I would guess) and the suspend/wake up time (mere seconds; and the battery lasts days if not at least a week while suspended).

I have not used anything but Linux for many years now, so it is a nice change to have another machine to play with. Nothing demands my love like Linux, of course ;)

procps

Finally released procps 2.0.13. It includes Red Hat's NPTL enhancements plus other optimizations, bug fixes and cleanups. Please give it a whirl, especially if you are using 2.5.

2.6 Kernel

Looks like 2.6-pre is coming soon. Hooray. For those who have not yet tried it out, I highly recommend it. It has been fairly stable for some time, and we have never had any awful hard drive eating problems or any mess like that.

Falconer

Two weekends ago was the SNL season finale and it featured a rather funny Falconer.

Bay Area

So I am out in the Bay Area again for a few months. The weather sure beats Florida.

Rejoice

There was a new Falconer on Saturday Night Live last night.

schedutils

Released schedutils 1.2.0. Mostly minor bugfixes and cleanups, although also the demise of irqset(1) (what did that have to do with scheduling?) and lsrt(1) (all of the same features, and more, are now in procps).

So, what is in schedutils? taskset(1), a utility for manipulating task CPU affinity (setting it, retrieving it, and launching a new task with an initial given affinity mask) and chrt(1), a utility for manipulating both scheduling priority and policy (also setting them, retrieving them, and launching a new task with a given priority and policy).

Source tarball as well as source and i386 RPM packages built against Red Hat 9 are available.

Java

Man, Java.HasQuality(Quality.Taste.Gross). 'Nuff said.
Gimme c->quality = TASTEFUL anyday.

OLS

Ack! OLS paper is due soonish.

Centrino

I like the design of the chip. More specifically, I like the IBM X31. Maybe some day soon?