Greg's diary--April 2002

Monday, 1 April 2002	Echunga
		Top of page
		next day
		last day

Up late again this morning. Spent a lot of time messing around, and even more trying to make a cassoulet. There's more to it than the average recipe tells you. One of these days I should start writing these things down. I did look the recipes up on the web, but didn't find anything too useful there either. At any rate, it ended up tasting surprisingly authentic.

Tuesday, 2 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

I was right when I decided not to plan anything for today. All sorts of things piled up, like they always do over a long weekend, and I seemed to get nothing done, though I note that I am being bombarded with mail again:

On 2 April 2002 you received 2078 mail messages and sent 55 mail messages.

It's just surprising that these things happen without being noticed. As a result, I didn't get much work done.

Wednesday, 3 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Teleconference day, which typically means that I don't get much work done. Today started off differently: we got through the teleconference in record time, and I was on my way to getting something done when Yana came and told me that her laptop power wasn't working.

Took a look at that: she had managed to bend the connector enough to break off one of the contacts. It's one of those silly non-standard connectors which are almost impossible to fix, so I took the machine apart to see if I could make a direct connection inside, but that's as good as impossible.

In the process, managed to dislodge a couple of keys from the keyboard, a thing warned against in the workshop manual, with good reason: it's very difficult to get them back in, but finally I managed.

In the meantime the dishwasher repairman came and discovered the reason why the machine kept stopping in mid-cycle: the drain connection was constricted. The leaks had nothing to do with that: a mouse had apparently chewed through the drain hose. I wonder what it thought it was doing.

After that investigated the alternatives for Yana's machine, and set up tweedledee.lemis.com, a Sun 3/50 running NetBSD on her desk. That was very frustrating. It's really out of the question to use such old machines any more, and I couldn't find a way to get rid of the console field at the bottom. All this took up the entire afternoon. A frustrating day all round.

Thursday, 4 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Yvonne in early to tell me that we had had a power failure. Turned on the emergency generator, which has made itself useful over the years, and tried to keep things running. It worked fairly well until the power came back and I turned back to it. Despite having shut down sat-gw, a Linux machine, it came back with a badly demolished root file system. Everything I did made it worse, so in the end I gave up and decided to reinstall the system.

If yesterday seemed bad, this was much worse. Just about everything seemed to go wrong. This was the machine I installed on 5 January, when the installation went well, but today it failed point blank. Firstly the machine wasn't able to boot from CD-ROM, so I tried installing on another machine (pizza), but the same CD (Mandrake 8.1) that I used in January refused to install, running into problems reading data from CD. I confirmed that the CD itself is OK, and changed the drive, all to no avail. Finally I gave up and tried to install Red Hat 7.0, the latest Red Hat version I have. That installed quite nicely, but when I tried to load my 2.4.8 kernel, it failed to recognize the host adaptor: it seems that the configuration is so specific that a change in host adaptor will stop the machine from booting.

Tried to build a new kernel, and discovered I hadn't installed any development software. Tried to install it afterwards and discovered that I would have to install a maze of twisty little packages, all different and all selected by mouse clicks. Gave up and reinstalled from scratch, including the development software. Tried to put it back in the old machine and discovered it wouldn't boot, so gave up and accepted using pizza.

Finally got a kernel built and discovered that it didn't have support for iptables, so had to start again. Even the new kernel didn't have it, though. It was getting late, so gave up for the day, after discovering that the machine wasn't forwarding packets, and that I couldn't find a way to configure it with the linuxconf utility. Finally traced it down in the initialization scripts: it's the value of /proc/sys/net/ipv4/ip_forward: 0 means no forwarding, 1 means forward. I wonder how people are supposed to find out these things.

In the evening a bit of light entertainment: Yana saw a mouse in the dining room. Yvonne went and looked for Lilac, who took one look at the mouse and caught it within two seconds. She really makes herself useful. Maybe we should lock her in with the dish washer.

Friday, 5 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

After the last couple of days' frustration, I had intended to take today off and do something else, but of course I wasn't finished by yesterday evening, so I first needed to build a new kernel after discovering that there were a plethora of kernel options that I hadn't seen first time round. And of course there were over 2000 mail messages which had piled up over the last day and a half, so I didn't get through all that until the early afternoon.

Then off to Mount Barker for the first time in weeks. Didn't do much of interest, but it brings back to me just how much work I've had lately, and that there's a real world out there.

Saturday, 6 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Not much work today. Finally went out riding in the forest for the first time in nearly 4 months. Darah went very well, but unfortunately Pebbles (whom Yvonne was riding) didn't. She's 25 years old now, and her old wound has never really healed up, and now the joint is getting arthritic. That in itself wouldn't be an issue, but Pebbles refuses to acknowledge it, and Yvonne spent most of her time trying to hold her back. If Pebbles had her own way, she'd probably go until she either dropped dead or broke a joint. Neither seems a good idea, so we decided that that was her last ride ever and went back home pretty quickly.

Sunday, 7 April 2002	Echunga	Images for 7 April 2002
		Top of page
		previous day
		next day
		last day

Finally got some time to work on my book... and the system hung on me, possibly a problem with missing disk interrupts. The disk activity light stays on, but the system doesn't hang immediately, and it continues running and accessing the disk until everything piles up waiting on some lock or another. Unfortunately, I can't dump this machine, since I made the swap partition exactly the same size as memory.

Decided to do a two-pronged approach: change the disk and upgrade the system, which was running a 5-CURRENT of 31 December last year. Moving the disk went without problem, and in the process I made a new swap partition slightly more than double the size of memory (in case I double memory). The upgrade was a different matter: there have been so many changes to the configuration lately that it just wouldn't start half the services. Spent all day trying, and finally reverted to the 31 December version. We really need a better way to update /etc than mergemaster.

Monday, 8 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

After all the fun at the weekend, found plenty had piled up, and spent most of the day cleaning that up. I wish I could find time to do some real work.

As if the annoyances of the past couple of days hadn't been enough, Yvonne told me should couldn't open any windows on her display, battunga:0. We hadn't changed anything on battunga, though it had been hit by the power failure on Thursday. After quite a bit of frustrating investigation, discovered that the startx script contained the line:

listen_tcp="-nolisten tcp"

Tracking this down, this seems to have been put in the XFree86 port in the Ports Collection. I'm all for adequate security, but this is ridiculous.

A couple of days later I wrote a more detailed description of this problem in a separate file called security-über-alles, and then forgot about it. Nearly 11 years later I found it again when a program nvidia-settings tripped over the name (with a ü in ISO 8859-1). Here's the content:

In the last few months, I've discovered it's becoming increasingly difficult to install a system which works the way I want. Last night I spent a very frustrating half hour trying to find why I couldn't open any windows on a particular display. Finally I found the reason: without any notice, the option had been disabled. /usr/ports/x11/XFree86-4-libraries/files/patch-startx contains:

+listen_tcp="-nolisten tcp" while [ x"$1" != x ]; do case "$1" in + -listen_tcp) + listen_tcp="" + ;; # '' required to prevent cpp from treating "/*" as a C comment. /''*|\./''*) if [ "$whoseargs" = "client" ]; then @@ -164,7 +168,7 @@ done #endif -xinit $client $clientargs -- $server $display $serverargs +xinit $client $clientargs -- $server $display $serverargs $listen_tcp

Yes, there's also corresponding info in the man page:

+The +.RB '-listen_tcp' +option of startx enables the TCP/IP transport type which is needed for remote X +displays. This is disabled by default for security reasons.

I think this is missing the point. Anybody who moves to FreeBSD from another operating system will have difficulty with this. It took me 30 minutes to find; I suspect most people would just give up. After all, there's a perfectly good (well, usable) way to do this, with xinit, which is also more granular. By adding this kind of feature we're just making the system less usable.

This isn't the only thing I can't get to work. rsh stopped working a long time ago. I can't replace it with ssh, because some systems on my network are either too gutless to run ssh, or it's just not there on that platform (think AIX). This is not a security issue, since the network contains only trusted systems and is firewalled from the Internet.

Tuesday, 9 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

More frustration today. Much mail, and I didn't get through it until after midday. After that, had an attack of dizziness which gave the feeling it might have something to do with heart problems, so called an ambulance. The paramedics couldn't find anything wrong with me, so they carted me off to the Flinders Medical Centre for a more thorough examination, which also revealed nothing. Probably just overwork, given the way things have been going lately, but a nuisance in view of all the work I have waiting already.

Wednesday, 10 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Conference call day again today, and this time I didn't get away so lightly: two calls, 90 minutes each. Not quite half the day gone on that, but along with mail it meant I didn't get much JFS 1 work done, though I did more than for a while.

In the evening to the AUUG SA meeting this evening, again at IBM. To my surprise, we had more attendees than last time, 25. Considering that we only have 24 members, that's quite remarkable. I wonder how long we'll keep it up.

Thursday, 11 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

After all the action of the last week and a bit, finally had a quiet day, and got a bit of work done. Also went to the barber's in Mount Barker. He does a reasonable job, and he's not too expensive, but he takes forever. I wonder how he makes ends meet.

Friday, 12 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

More work on JFS today, more than slightly hampered by the fact that my mouse seems to be wearing out. After replacing it with a cord mouse, it got out of sync with the driver; I'm sure we need to look at mouse protocols more carefully. It's really a pain that peripheral vendors in the Microsoft space all create their own protocols and then supply Microsoft-only drivers for them. The worn-out mouse was a Logitech wireless mouse, which I liked because the cord doesn't get in the way, but I was never able to enable the side key due to lack of documentation.

The mouse was connected to echunga, my main gateway machine, which had been up for 97 days. Spent some time trying to get the mouse in sync. Found the following code in /usr/src/sys/isa/psm.c:

        if (sc->config & PSM_CONFIG_SYNCHACK) {
        /*
         * XXX: this is a grotesque hack to get us out of
         * dreaded "out of sync" error.
         */
        log(LOG_DEBUG, "psmintr: re-enable the mouse.\n");
        disable_aux_dev(sc->kbdc);
        enable_aux_dev(sc->kbdc);
        }

PSM_CONFIG_SYNCHACK wasn't set, of course. Messed around with gdb on the running kernel to patch the softc, in the process learning a new option to gdb: in order to write to the kernel you have to write:

# gdb -k --wcore kernel.debug /dev/mem

Finally got it modified, but all that happened was:

> Apr 12 17:13:54 echunga /kernel: psmintr: out of sync (0000 != 0008).
> Apr 12 17:13:54 echunga /kernel: psmintr: re-enable the mouse.
> Apr 12 17:13:54 echunga /kernel: psmintr: out of sync (0000 != 0008).
> Apr 12 17:13:54 echunga /kernel: psmintr: re-enable the mouse.

In other words, it didn't work. I had to reboot a machine that had been up for over 3 months just because the mouse got disconnected. sigh

Saturday, 13 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

More work on the book today. Somehow it's difficult to get back into the swing. Maybe I should stop reading mail every day and spend a whole day doing nothing except my book or my coding.

Sunday, 14 April 2002	Echunga	Images for 14 April 2002
		Top of page
		previous day
		next day
		last day

More work on the book today. Got three chapters to review status. It's amazing how much has changed in the last three years.

Monday, 15 April 2002	Echunga → Canberra
		Top of page
		previous day
		next day
		last day

Off to Canberra at a civilized time this morning; it's a lot more comfortable leaving at 11 am than early in the morning.

Not much to do in Canberra; most of the people I wanted to talk to today were on holiday. Talked to Martin Schwenke, who decided I should be installing Wine on my laptop so that I wouldn't have to run Microsoft at all. The problem was, it wasn't just any kind of Wine, it had to be a version that he had on his system, packed up as an RPM and ready to install on Linux. Unpacked it and tried installing, but didn't get very far. There's too much manual stuff required, and that needs to be reproducible.

Tuesday, 16 April 2002	Canberra
		Top of page
		previous day
		next day
		last day

Talked to Hugh and decided that kernel debugging on the pSeries is becoming interesting again. I still need to get the JFS1 stuff working, but that looks like a good thing to do after that.

While doing some work, the disk on my laptop started making the old “clunk” noises that I know from the Dell laptop, so I mentioned it to Hugh, who had a replacement here within hours. That's impressive. Put it into an external adaptor and had the entire disk (30 GB) copied in 50 minutes. Possibly the most painless disk change I've ever had.

Wednesday, 17 April 2002	Canberra	Images for 17 April 2002
		Top of page
		previous day
		next day
		last day

This morning it occurred to me that the disk change yesterday went painlessly and fast, but it wasn't what I wanted to do, since the partitions were all messed up. Started repartitioning and copying, which took rather more effort.

People gradually started trickling back to work today. Had our weekly meeting, for a change not a teleconference, and then out into the park on the shores of the lake to take a photo session, since for the first time we were all here.

It wasn't until nearly 20 years later that I scanned in my copy:

Image title: Ozlabs group photo Dimensions: 3181 x 2392, 2637 kB

Make a single page with this image Hide this image

Make this image a thumbnail Make thumbnails of all images on this page

Make this image small again Display small version of all images on this page

All images taken on Wednesday, 17 April 2002, thumbnails All images taken on Wednesday, 17 April 2002, small

Diary entry for Wednesday, 17 April 2002 Complete exposure details

From rear to front, left to right:

Martin Schwenke, Greg Lehey, Greg Rodgers, Stephen Rothwell

Rusty Russell, Hugh Blemings, Alan Modra

Anton Blanchard, David Gibson, Paul McKerras

Tracy Whatman, Chris Yeoh

Then off to lunch, our first anniversary get-together.

In the afternoon, finished repartitioning the laptop, and decided to install Linux. It's certainly easier with a few Linux gurus hanging around, but it's still a lot more work-intensive than with FreeBSD. Got everything installed, but there's still at least X configuration to do.

In the evening met Terry McGee, and gave him two old flutes for repair, including the Siccama, which he identified as having been built by Chapell.

This was no particular surprise, since it's written on the instrument:

Image title: siccama 10 detail

Complete exposure details

Dimensions: 553 x 655, 78 kB

Dimensions of original: 553 x 655, 78 kB

Display this image:
thumbnail    hidden   alone on page

Display all images on this page as:
thumbnails    this size

Show for Sunday, 23 December 2001:
thumbnails    small images    diary entry

This seems to have been the second time I missed this; I also mentioned it on 1 April 1970, when Philip Bate looked at it and made the same observation.

Talked with him at some length about acoustics; seems there's a fair amount of research going on here at the ANU and UNSW. I can see myself getting involved again, though I should really recognize that I don't have enough time.

Thursday, 18 April 2002	Canberra
		Top of page
		previous day
		next day
		last day

It seems I don't need to be at home to find plenty of distractions. Took another look at the kdb sources, which I'm planning to port to PPC64. I thought it would make more sense to pick up the SPARC-64 sources (strangely, since the project is hosted at SGI, they didn't have a MIPS port), but noted after picking it up that I really needed an Intel port first, so ended up with both patch sets. Strangely, the SPARC patches were twice the size of the Intel patches. I also was not able to find any documentation. sigh

Also spent some time going over the JFS1 code with Rusty. Fortunately, he didn't find anything too much wrong with it.

In the evening to Hugh's place for dinner. It's a surprising distance out of Canberra. Rusty took me back in the new Mazda RX-5 which he bought as a surprise for Ali a few weeks back. Nice car, but what do you do with a really fast car in Australia?

Friday, 19 April 2002	Canberra → Echunga
		Top of page
		previous day
		next day
		last day

Spent most of the day collecting together stuff to take back home with me. There's something to be said for a large disk on a laptop; I ended up with several gigabytes of data.

The journey home was uneventful, though Qantas has plumbed new depths in the quality of their on-board food. These cardboard boxes are terrible; I returned mine untouched and was surprised to discover that they didn't even serve coffee.

Saturday, 20 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Another day with little of note. Spent most of the day catching up with things I wasn't able to do during the week and preparing for Yvonne's departure to Lima tomorrow. Didn't even get any work done on the book.

Gradually our front gate is taking shape. Today the people came to deliver the gate; we still need to wait for the people to install the mechanics, which will possibly happen on Monday.

Sunday, 21 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Up early this morning to take Yvonne to the airport on her way to Peru. Managed to get her an upgrade to Business Class despite Qantas' bureaucracy.

The flight from Melbourne to Lima was first class, but their frequent flyer system had only been able to give her economy from Adelaide to Melbourne.

Back home, tried to get some work done on my book, but somehow didn't get very far. Instead pondered the amount of network traffic I've been receiving lately; it's been round the 1.2 GB per month for the last few months since I moved to the satellite service, but on Thursday last week I somehow managed 339 MB, which would quickly become expensive. Spent some time playing around with the satellite receiver software as a result, and got it to log statistics.

Yvonne called from Sydney: they had some mechanical problems with the plane, and as a result left 2½ hours late, enough to ensure that she would miss her connection at Buenos Aires. She called later from Auckland to confirm that yes, indeed, she would miss her connection. Grrr.

Monday, 22 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Got through my mail pretty quickly this morning and tried to boot sphinx, my pSeries machine. I failed: somehow I managed to make a mess of the open firmware environment, or maybe it was a delayed reaction to the power failure we had on a couple of weeks ago. Called up the blokes in Canberra, and between the three of us we managed to turn a problem in a catastrophe. By the end of it, we couldn't even get into the Open Firmware. So much for our hacking expertise.

In the afternoon, back to JFS1 with renewed vigour and strongly reduced time. I really want to get this stuff working soon.

Tuesday, 23 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

More work trying to get sphinx up today. Learnt a lot about how the boot process and the diagnostic process worked, with the result that by mid-afternoon I knew beyond reasonable doubt that I had a failed processor board, though it's not clear whether that was due to fried microcode or something more sinister. Called up the service people, who were really responsive; I even ended up getting two different calls on the subject at the same time. Didn't get much else done as a result.

Wednesday, 24 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Conference call this morning, shorter than normal; none of us seem to have done much work. I certainly was still kept busy by the failure of sphinx. It's interesting to note how much time goes into this sort of problem. The parts hadn't arrived today, and since tomorrow is a public holiday, we'll have to wait until Friday. Spent some time looking for replacement equipment, after which I had no motivation to work on the JFS 1 stuff, so did some AUUG organizational stuff instead. The elections are due to be held in a couple of weeks.

We have a surprising number of birds around the house at the moment. This morning I saw about forty welcome swallows, which may be migrating, though it's not clear how far: in some places they do, in others they don't. In addition, on the pond we had a heron, a moorhen, some unidentified duck (normally we have Australian wood ducks here, but they were not there today, and I don't know what this one was) and what I think is a baby little pied cormorant. Judging by the fact that it doesn't appear to be able to fly, it must have been born here. Today an adult little pied cormorant was also there. In addition, there were two different kinds of birds which I can't identify in the weeping willow on the island, and galahs, magpies, magpie larks and a crow near the bank.

In fact, it turned out that the bird was really a fully-grown grebe.

Thursday, 25 April 2002	Echunga	Images for 25 April 2002
		Top of page
		previous day
		next day
		last day

ANZAC day holiday today, so no work. Still, Barry Engel, the electrician, got me out of bed to wire up the front gate, which is now functional except for the intercom unit, which I'm planning to make out of an old laptop and a wireless card, so we can have the gate on the Internet. That'll take a while, though.

Spent most of the day working on the book, and actually made some progress. If things carry at this rate, it'll only be a few more days before the book's complete.

Today the Australian wood ducks came back in force, seven of them. This is what I saw from my office window:

https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20020427/big/ducks-1.jpeg

Image title:	ducks 1	Dimensions:	2048 x 1536, 656 kB
Make a single page with this image		Hide this image
Make this image a thumbnail		Make thumbnails of all images on this page
Make this image small again		Display small version of all images on this page
All images taken on Saturday, 27 April 2002, thumbnails		All images taken on Saturday, 27 April 2002, small
Diary entry for Saturday, 27 April 2002		Complete exposure details

The baby cormorant looks like it's lonely: it started swimming after the ducks, but wasn't fast enough:

https://lemis.nyc3.digitaloceanspaces.com/grog/Photos/20020427/big/ducks-2.jpeg

Image title:	ducks 2	Dimensions:	2048 x 1536, 672 kB
Make a single page with this image		Hide this image
Make this image a thumbnail		Make thumbnails of all images on this page
Make this image small again		Display small version of all images on this page
All images taken on Saturday, 27 April 2002, thumbnails		All images taken on Saturday, 27 April 2002, small
Diary entry for Saturday, 27 April 2002		Complete exposure details

In fact, we later discovered that it was a grebe.

Friday, 26 April 2002	Echunga	Images for 26 April 2002
		Top of page
		previous day
		next day
		last day

Gradually things are settling down again. I've got my mail folders more or less under control, and I found time to work on JFS1 again. I suppose it's time for another conference. By the evening I had most of the JFS1 directory lookup code ready; I suspect there isn't much more code to do, though there are still plenty of loose ends.

Lothar Dummin, the IBM tech, came in early this morning with replacement parts for sphinx. Rather to our mutual surprise, they didn't solve the problem. Rather than have him come out here again, let him take the box with him. I'm still a little surprised.

Saturday, 27 April 2002	Echunga	Images for 27 April 2002
		Top of page
		previous day
		next day
		last day

More progress today, and got quite a way with the book. Nearly completed stuff on wireless networking and PCMCIA, but I was left with the concern that the information is spread across too many chapters. It's really difficult to keep a good structure when there are so many interdependencies.

Sunday, 28 April 2002	Echunga	Images for 28 April 2002
		Top of page
		previous day
		next day
		last day

After the progress I made yesterday on the book, I was all set to continue, but it occurred to me that the end of the month was fast approaching, and that I had both the notice of the AUUG elections and an article for Daemon news to write, and spent most of the day doing that.

In the evening, Yana was gradually showing the strain of running the household for a week, with the result that I ended up doing most of it. Maybe as a result, surprisingly tired and early to bed.

Monday, 29 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

More hard work today on JFS 1. I now have the code ready and compiling, but no hardware. I really must do something about repairing the stuff that's down.

Tuesday, 30 April 2002	Echunga
		Top of page
		previous day
		next day
		last day

Down to 120 mail messages! I'm trying as hard as I can to get down to a single screenful (86 messages), but it's not easy. Still, I'm getting there.

Discovered a lot of missing stuff in my JFS1 code, including block read. Found corresponding code in the ext2 and minix file systems, but it's really strange. It took me some time to realize that I had already written the code, but it looked completely different. Now I'm left wondering whether I have missed something.