Funk Babblings

When trying to test and debug something to which you have limited access, always make sure you know how to undo anything you try, without having to wipe it and start over. And after verifying that what you did needs to be undone, don't reboot until you've undone it.

I have yet to figure out if all my iPod customizations will survive my inappropriate use of "launchctl remove com.apple.mDNSResponder". And I hope, if I do have to wipe it, that Apple isn't pushing firmware 2.0 just yet.


Update: It's fixed. The same Windows program I used to "jailbreak" the iPod lets me switch it from Recovery Mode to Normal Mode. And after that things seem to be working fine; "launchctl list" includes mDNSResponder in the list (unlike after I first did the remove).

I think I could've undone the "launchctl remove" with "launchctl load /System/Library/LaunchDaemons/com.apple.mDNSResponder". But I'm not eager to test that anytime soon. It would be nice to have a working Linux program that does what ZiPhone does, though.

Today there was a security alert released covering almost all DNS server software, and in response a bunch of them were updated to improve the way they randomize one aspect of the protocol.

But I noticed that there was as yet no word on one DNS server program I use on my networks, dnsmasq. So I checked a bit, and discovered that the author had just released a release-candidate before going on a two-week vacation. I grabbed the latest code, and did a search for "random" in the supporting documentation. The result of that search?

Upgraded random ID generation - patch from Rob Funk.

Oh yeah! I'd forgotten about sending in a patch way back when. I checked my own machine, and found that I still have a copy of that original patch in a backup folder, dated 2002. And when I looked deeper in the current code, I found the part I'd sent in, still unchanged. (Hey, those comments sound like something I'd write!)

Of course, that didn't answer my original question; my code improved the way the program gets random numbers, but other parts of the program determine how those numbers are used. And the ultimate answer seems to be that it relies on the underlying operating system for the randomness I was looking for.....

Update: I think my random code may finally need to be updated, as part of the mitigation for this whole problem. Needs more bits.