The Daves Collective

Fatwa of Unsuitability
2004-05-31 15:30

I hereby pronounce a Fatwa of Unsuitability (F.U.) upon Gallery. Its replacement will be JAlbum and its ExhibitPlus skin. Gallery was found guilty of PHP Abuse (register_globals, uncaught warnings), Recent Security Cockups, Wasting Too Much Of Dave's Time, and Pandering To Copyright Nazis (by adding watermark support).

Another F.U. was recently pronounced on fetchmail, now replaced by Getmail. Fetchmail was found guilty of Being Written By A Right Wing Gobshite and Too Many Security Advisories.

Back with a revamp
2004-05-29 15:18

It's draining van! Halleluia After a break of four months (for good but unstated reasons), things here are moving again. The site looks very much the same, but it's been refactored to take advantage of Zen's SSI support on their free homepages, enabling common content and presentation throughout the site without lots of duplicated code to maintain and get out of step.

This has also enabled the new content in the right column of most pages. On the front page, this tracks interesting links on a number of topics, with old links automatically expiring as new ones are added. Elsewhere, the right column contains links related to each leaf page's content; the links aggregate automatically at index pages. Each such link is actually an individual story of a parallel Blosxom flavour, generated as static pages and merged into the site using SSI. El Neato or what.

There's also some new content. The new system pages describe various nonbiological entities in the Collective. There are some new files in the Config Bistro, including an XF86Config for the Toshiba Portege 7200 and a Linux 2.4.26 kernel .config for the SGI Indy.

Lastly, the bloody drains blocked again. Fixed now. What a relief.

Search terms and answers
2004-01-11 01:21

The Googlebot found me a while ago, and I can see from the webstats that there's a few folk tripping in with interesting search terms. For example, there are lots of people looking for answers to Linux 2.6 problems that aren't fully explained in the Linux 2.6 article I posted last month. So I thought it would be nice to provide what people are looking for.

linux 2.6 modprobe slackware -and- modprobe.devfs 2.6.0 etc

In recent Linux 2.4 kernels the new module-init-tools package is an alternative for the modutils package. In Linux 2.6, module-init-tools is required and modutils cannot be used.

Before upgrading to 2.6, you should install module-init-tools and its new config file /etc/modprobe.conf. If you use devfs, you will also need /etc/modprobe.devfs.

To create /etc/modprobe.conf from your own /etc/modules.conf, just run the script /sbin/generate-modprobe.conf. The /etc/modprobe.devfs file is provided in the module-init-tools source package, but not in the Slackware package. So you should get the source package, untar it and just copy the modprobe.devfs file from the top level directory to /etc.

bdflush pdflush update init -and- obsolete bdflush system call etc

warning: process `update' used the obsolete bdflush system call
In Linux 2.6 there's a new kernel thread named 'pdflush' which replaces the old 'update' daemon. All you need to do is to remove the command (somewhere in your boot scripts) that starts the old 'update' daemon. The new 'pdflush' thread runs all the time and you do not need to start it.

syslogd is using obsolete setsockopt so_bsdcompat

process `syslogd' is using obsolete setsockopt SO_BSDCOMPAT
The sysklogd package uses a socket option which still works in 2.6 but is deprecated. This package hasn't been updated yet (you can check here). You can ignore the message, or install the alternative syslog-ng package.

/proc/bus/i2c 2.6.0 test11

Ah yes, lm_sensors under Linux 2.6. Easy! Use the in-kernel i2c instead of the i2c package. The file /proc/bus/i2c and the i2c-proc module do not exist in 2.6 - there's an entry in the new /sys filesystem, but to use it you need lm_sensors-2.8.2. To build lm_sensors-2.8.2 you should only do 'make user' and 'make user_install' (and then do 'ldconfig'). Remove any references to i2c-proc in your boot scripts. Finally, in Linux 2.6.1, the 'sensors' script doesn't work with the eeprom driver, it gives a couple of errors (there's a fix in lm_sensors CVS).

usbfs kernel 2.6.0 fstab

The usbdevfs filesystem has been renamed usbfs. In Linux 2.4, both names work. In Linux 2.6, only the name usbfs works. Just change "usbdevfs" to "usbfs" in /etc/fstab, this will also work with recent 2.4 kernels, the relevant line should read
/dev/usb /proc/bus/usb usbfs defaults 0 0
You should also change /etc/modprobe.conf, the relevant line should read
alias usbfs usbcore
(If the mount point /proc/bus/usb does not exist, you probably forgot to change /etc/modprobe.conf.)

alsa asound.state emu10k1

Ah, now I'm glad you asked me that, you want to see my new Config Bistro. For information about building newer versions of ALSA for Linux 2.6 kernels, see the the AlsaBuild2.6 Wiki page.

exim no reverse dns deny

Interesting, you want Exim to reject incoming mail if there's no reverse DNS? Well personally I use sa-exim so that SpamAssassin scores IP addresses in 'Received' headers - not just if there's no reverse DNS but also if the address is in various RBLs such as the SBL.

filez par2 -and- part1.rar part2.rar help open

Listen kids, try something less intellectually demanding than warez. Like body piercing.

sex pix -and- unwanted gropes pix (I'm not making this up!)

Shocking picture of Chokotay groping Janeway Yes, I've had both these searches in the web logs. Apparently I'm second out of four results on a Google text search for "unwanted gropes pix". Blimey. Anyway, listen up idiots, if you want pix then do a Google image search. I did, and look what I found!

Announcing the Config Bistro
2004-01-09 17:37

Now open - the provisionally wonderful Config Bistro. This contains some of my config files, shared in the hope that they will be useful and informative.

Death of the Internet, part XXXXII
2004-01-08 22:06

Bing bong we regret to announce the Internet is broken due to the expiry of a Verisign root certificate. This jiggers yer Java and knackers yer Norton. Groooovy. It explains the repeated 20-sec hangs on starting the Norton firewall which have been bothering me all afternoon. If it wasn't for a couple of Windoze apps I wouldn't have this grief. Because the Norton stuff is Java based, upgrading the JRE fixed it.

Staggeringly, right-clicking in Windoze Explorer produces the same 20-sec hang! It would appear that the notoriously integrated MSIE gets upset over the expiry if you have "check for certificate revocation" checked (as you should) - even though you are not using any certificates that need checking. I'd be willing to bet this is an example of objecthead design, some bloody class constructor probably contains code it shouldn't.

So let's hear it for Sun, who could have been shipping JRE's with new certificates since December 2001, but who have apparently only done so since November 2003. Let's hear it for Symantec, whose software is not directly at fault but who chose to use Java for their product and who are too coy to admit to this problem on their webshite, or to recommend the JRE update as a workround, and whose LiveUpdate has insisted all day that everything is peachy. Let's hear it once again for Microsoft and the integration of MSIE. And let's hear it for Verislime, not the Net's favourite company, who reacted to the collapse of crl.verisign.com by null-routeing it in DNS, thereby temporarily abdicating their position of trust and (for now) making the Net just a little bit less safe.

Oh yes, not to forget, the new certs expire in 2011. Here's to the next time.

Linux 2.6.0 (boast, boast)
2003-12-18 21:53

Released today, installed today! Yes, we are at home to Mister Smug-face. It's quite usable, perhaps I'll stick with it for now since there's a good-enough workround for each of the issues from when I tried -test11. (Fond memories of when 2.4.0 was released, within 24 hours I had that up and running on a 486DX33 with an ESDI disk, aaah)

d$ cat /proc/version
Linux version 2.6.0 (d@dsds) (gcc version 3.2.3) #1 Thu Dec 18 18:19:29 GMT 2003
d$

Further evidence of Global Conspiracy
2003-12-13 12:43

System Tools menu There's an exposé in Wired entitled Microsoft Loses the Swastika. The gist of it is that the naughty squiggle was inadvertently included in an Office 2003 font named Bookshelf Symbol 7 (their excuse being it "was derived from a Japanese font set"; that's certainly plausible to anyone who's seen Bridge On The River Kwai). Yes indeedy, there it is at 0x7E. But the really scary thing is that Character Map is right next to the scientologists' Diskeeper in the Win2K System Tools menu! Tools of the System! Yes, it makes sense now! I wonder if System Information is fnord channeling the fnord fnord Illuminati?

Character map Führer Gates

Update: They're watching me!

There's more. Webdings font, 0x85. Now I'm really scared. Gonna leave the light on tonight.

Firmware upgrade on Nokia 221T
2003-12-11 15:47

Exciting! Today they're broadcasting an upgrade for my Nokia 221T DTTV box.
Full writeup in the picture album.

Nokia 221T Receiver upgrade
The worst thing about this box is that it looks like a handbag.

Defragging and scientology, DEC and Microsoft
2003-12-10 21:51

Why is the Win2K defragmenter such a pile of poo? Why does it say it needs >15% free to do the biz, even when the biggest single file on the partition is much smaller than the available free space? And why, at 17% free, does it still moan that it only has 8% "available" to it? Why is the fs still fragged to buggery when it's finished defragging, but then if you run it again straight away it defrags a lot more - so why doesn't the sodding thing iterate until it's done a proper job? Why, when you run it on a partition with no reported fragments at all, does it sometimes actually produce a couple of fragments?

Because it was written by "fully trained scientologists, computer skills desirable but not a prerequisite". That's why.

The Win2K defragmenter is a deliberately impaired version of Diskeeper, licensed by MS from a firm called Executive Software who are so intertwined with the scientology racket that the German government threw a wobbly about Win2K until MS grudgingly coughed up a KB article (in German) on removing the defragger. As far as I can tell there are no free-gratis or free-libre alternatives.

If, dear reader, you might indulge the reminiscences of an old fart, ten or fifteen years ago Executive Software was notorious to VMS sysadmins for their, ummm, tenacious sales pitch. This was back when Diskeeper was a defragger not for NTFS but for VMS (in competition with a product called Rabbit-4, or was it Rabbit-7, anyway they eventually renamed it Perfectdisk).

And that leads me to ponder whether NTFS is, fundamentally, a knockoff of an outdated version of the VMS ODS-2 filesystem. Apparently I'm not the only one to wonder. Here's my evidence: (1) both Diskeeper and Perfectdisk were repurposed from VMS to NTFS with apparent ease; (2) the placement of the green NTFS "system files", as shown in the Win2K defragger, is a dead giveaway; (3) both FSes fragment early, and then the performance of each does the same bellyflop. (Incidentally I have similar suspicions about the so-called AdvFS in OSF/1, err Dec Unix, err Doomed64; it's funny how it has the same tunables as VMS... trust me guys, don't be fooled by its name, use UFS instead.)

Certainly the NT kernel is just a knockoff of a primordial version of the VMS kernel. It is a matter of record that VMS progenitor Dave Cutler flit from DEC to MS and there didst begat NT in the image of VMS. But lo, primordial VMS had a dire VM that paged too soon and thrashed the pagefile even unto death at the first sign of memory pressure, and sure enough, open one too many apps on my token Win2K box and the IDE disk makes EXACTLY the same sort of rattling rolling rhythms as the old RA80s used to do ten minutes before the valediction on the console printer. DEC did manage (post Cutler) to make the VM more tunable. Perhaps MS should license a deliberately impaired AUTOGEN from HP? Is Carly a scientologist?

By the way, lots of kewl VAX photos here.

A first go at Linux 2.6
2003-12-08 12:00

Trying out 2.6.0-test11 using the following references - The post-halloween document and HowTo Upgrade To The 2.6 Kernel. Well it mostly worked: here's the dmesg from booting 2.6 and here's my .config. Visible changes were/are as follows -

Mouse acceleration within X much much faster, need to set it much slower
Stinking nvidia binary-only module not yet rewritten for 2.6 - an unofficial patch is at http://www.minion.de/ but easier to revert to 'nv' driver
Peculiar errors launching X (having reverted to nv driver - wtf??)
Migration modutils to module-init-tools - cos I use devfs, need /etc/modprobe.devfs, but not in Slackware 9.1 module-init-tools package - had to get source tarball, though since modprobe.devfs is just a static file in there I didn't need to build from source
add new sysfs to fstab
/sys /sys sysfs defaults 0 0
devfs no longer provides devpts filesystem at /dev/pts, now needs the old explicit fstab entry as if devfs not present
/dev/pts /dev/pts devpts gid=4,mode=620 0 0
usbdevfs is now renamed usbfs (although presumably there should be some replacement in /sys)
/dev/usb /proc/bus/usb usbfs defaults 0 0
Migration bdflush to pdflush - "warning: process `update' used the obsolete bdflush system call" - just needed to remove 'update' from rc_sysinit
syslogd: "process `syslogd' is using obsolete setsockopt SO_BSDCOMPAT" - no sign of a new or patched version of sysklogd, 1.4.1 now two and a half years old, could try migrating to syslog-ng but it's got previous and the "current stable version" is tagged -rc4. Hmmmmm.
2.6 bzImage is much much bigger than 2.4 for similar config - why?
integrated version of i2c replaces /proc/bus/i2c with the new sysfs interface for which new userspace tools lm_sensors2 (in cvs) are needed. (Note the lm_sensors mailing list is here.)
the alsa emu10k1 driver in 2.6.0-test11 is quite some way behind recent alsa-driver releases (I upgraded 0.9.6 to 1.0-rc1 last week but the kernel seems to be older than 0.9.6), so it will mean manually reverting all the usual ghastly undocumented unversioned manual changes to the asound.state mixer settings, and then at some time in the future it'll all have to be painfully re-done yet again back to where I am now.

Note that the various fstab changes, the emu10k1 mixer settings and the X mouse acceleration are the only things I've noticed that make dual boot between 2.4 and 2.6 a bit grotty. But the main reason for not permanently going to 2.6 is still the criminally irresponsible great kernel header cockup.

Update - Prime Number Shitting Bear