neitzels irrelevant irregular intarweb notes

Studied NetBSD's driver(9) and pci(9) kernel man pages while looking at sys/dev/pic/if_ral_pci.c.

Very well done man pages, things start to fall into place.

Somehow I stumbled over Joerg Sonnenberger's fossil repository clone of the NetBSD src tree (wich is originally a CVS repo): http://netbsd.sonnenberger.org/

Very handy.

I had fooled around with fossil some years ago, so I looked at the current state of the project (still very active) and decided to give one of my old fossil installations from 4 years ago an update:

1. Use the old fossil binary to clone the current fossil source repository from http://www.fossil-scm.org/fossil.
2. Create a new fossil binary and install it over the old one, without any safety belt.
3. Tried a fossil status and fossil info, got errors, the information that the database format had changed, and the advice to do a fossil rebuild.
4. Which went very quickly and smooth, and most importantly: made everything work again. Not just for the current fossil repository but also for my own tiny repository from back when.

I like software wich takes care of itself, even across four years.

I located Intel's datasheet for its Mobile Intel 945 Express Chipset Family and spent eight hours reading in it. (These 482 pages offer a lot of material.)

I found some details which may be relevant for the buggy screen blanking happening with NetBSD-current on the Asus Eeepc: PCI device numbers on specific chipset functions and interactions with LVDS devices, including DDC and EDID information.

And before I forget it again: the i945GSE variant is designed for the Intel Atom N270; it is almost (no Analog TV MacroVision) compatible with the i945GMS which in turn is almost compatible with the i945GM/GME, except for:

• no External Graphics via PCIe
• 2GB memory max (of only certain kinds/topologies/speeds)
• only one SDVO port, no SDVO slot reversal

Did I ever wanted to read about "Legacy DOS/Video/BIOS Areas" again? Not really, but it appears they are still fully supported by current chipsets. Wow.

Even more "Wow!" is what 2D/3D graphics capabilities are implemented in these chipsets.

When we install a new server in our data center, there are six to ten labels to be printed, usually indicating the hostname and IP address:

1. server chassis, front and back
2. network cables (often two of them), labeled at each end
3. power cables (often two of them), labeled at each end

For the last twenty years, it always irked me that I wasted more laminated tape space for gaps between those labels (5cm) than for the actual labels themselves (3cm). Today was a slow day at work (nudged between a holiday and the weekend), and a fine occasion to study the P-Touch PT-310 manual. Result:

Pardon the crappy cell phone photo, but you get the picture... Hurray - no more gaps!

The manual features lots of exaggerated compound nouns the German language is so notorious for but which only occur in texts translated from other languages. Speak along with me:

Bandauslaufschlitz, Unterstreichen-Funktionstaste, Druckwiederholfunktion, Bandabschneidehebel, Aufkleberlaengenfunktion

Luckily, I just had to disable the automatische Bandvorschub.

• Finished reading the the "network" chapter of "4.4BSD Design & Implementation", done as background for the ral(4) work. By coincidence, parts were was also a bit relevant for the network admin lab that I am currently doing.

• Updated my NetBSD-current kernel to 7.99.15; found out that I will get video output when I attached my FullHD monitor to the external VGA port of the EeePC. My current impression is that i915drmkms(4) might start to work when I manage to fix the EDID recognition of the 1024x600 LVDS.

Three days continued reading on JunOS basics: CLI usage (with every little detail), and basic configuration.

Unfortunately, I cannot do live packet captures from op-mode. As of now, I am stuck with setting up filters to drive byte/packet counters. Quite tedious.

Usually working with Cisco gear, I have now my first login with a Juniper box, one of these:

The JunOS CLI and its organization appears to be not too different from the Cisco CLI (as are HP's or Dell's), but it never hurts to RTFM.

http://www.juniper.net/techpubs/en_US/junos12.1/information-products/pathway-pages/junos-cli/junos-cli.html#overview will is on the schedule for tonight.

Plenty more reading (mostly on basic routig and debugging) to come, the full documentation shelf is http://www.juniper.net/techpubs/content-applications/content-explorer/#cat=Configure&family=SRX%20Series&rel=Junos%2012.1x44.

So much to read, so little time...