Day: 27 August 2012

As I noted last week, I purchased a MacBook Air to replace my ageing MacBook Pro. As part of the transition process, I’m trying to lose legacy, unnecessary files as much as possible. To make this process as easy as possible, I have turned on file sharing on my Pro to connect to it over WiFi for a la carte file transfers. Right now, it works great. But if you’d asked me a couple of weeks ago, I would have snapped.

After upgrading my Pro to Mountain Lion on July 25, I noticed that it dropped my otherwise perfectly stable WiFi connection every few hours, inconsistent with any previous task. Even worse, it wouldn’t automatically reconnect to the network. This is a behaviour I had not seen in any of the developer builds of Mountain Lion, nor any version of Mac OS X prior.

I kept track of each of the console messages for every incidence of this. Every single one looked similar to this1:


2012-07-25 5:24:59.000 PM kernel[0]: AirPort: Link Down on en1. Reason 255 (unknown reason).
2012-07-25 5:24:59.000 PM kernel[0]: en1::IO80211Interface::postMessage bssid changed
2012-07-25 5:25:00.402 PM configd[17]: network changed: v4(en1-:10.0.1.40) DNS- Proxy- SMB
2012-07-25 5:25:00.440 PM com.apple.usbmuxd[23]: _SendDetachNotification (thread 0x[REMOVED]): sending detach for device [REMOVED]:5caa._apple-mobdev._tcp.local.: _BrowseReplyReceivedCallback got bonjour removal.
2012-07-25 5:25:00.442 PM iTunes[1364]: _AMDDeviceDetached (thread 0x11708a000): Device 'AMDevice 0x[REMOVED] {UDID = [REMOVED], device ID = 21, FullServiceName = [REMOVED]:[REMOVED]._apple-mobdev._tcp.local.}' detached.
2012-07-25 5:25:01.419 PM awacsd[58]: InnerStore GetWakeInfoForZone: zero address for [REMOVED].members.btmm.icloud.com.

[TRUNCATED]

2012-07-25 5:25:36.000 PM kernel[0]: en1: BSSID changed to [REMOVED]
2012-07-25 5:25:36.000 PM kernel[0]: en1::IO80211Interface::postMessage bssid changed
2012-07-25 5:25:36.000 PM kernel[0]: AirPort: Link Up on en1
2012-07-25 5:25:36.000 PM kernel[0]: en1: BSSID changed to [REMOVED]
2012-07-25 5:25:36.000 PM kernel[0]: en1::IO80211Interface::postMessage bssid changed
2012-07-25 5:25:36.000 PM kernel[0]: AirPort: RSN handshake complete on en1
2012-07-25 5:25:36.612 PM configd[17]: network changed: v4(en1+:10.0.1.40) DNS+ Proxy+ SMB
2012-07-25 5:25:36.660 PM UserEventAgent[11]: Captive: en1: Not probing 'Free Public WiFi' (protected network)
2012-07-25 5:25:36.681 PM configd[17]: network changed: v4(en1!:10.0.1.40) DNS Proxy SMB

The first half of that block represents my Mac instantly dropping its WiFi connection, and iTunes disconnecting my WiFi-synced iPad. Note that the iPad WiFi connection did not drop—it was just ejected from iTunes since the host WiFi connection was lost.

The second half of that block is my manual reconnection of the signal2.

Since I intended to create a small local file server with my MacBook Pro, its WiFi connection needed to be as reliable as possible. I began with the basic troubleshooting steps: doing a quick Google search to see if anyone else saw the same message. I didn’t expect to find many results, but I thought there would be a few. I was shocked when I discovered that nobody else saw the same console readout for their dropped connection as I did (or, at least, nobody complained about it with that same “Reason 255” code)3.

Even though no other computers in my house were exhibiting the same behaviour, I thought it might be a good idea to examine the AirPort Extreme base station logs to see if there were any clues there. However, this was made more difficult than it needed to be because Apple removed much of the functionality of AirPort Utility with the more user-friendly version 6.0.

You would think that I would be able to download AirPort Utility 5.6 for Lion and install it on Mountain Lion, given that the latter is a newer OS. Alas, the installer for it has both a minimum and maximum OS version. I used the excellent Pacifist to pull the app out of the package and run it. To my dismay, there were no relevant log entries.

I rarely install beta software on my main partition, just in case something goes wrong. However, I decided to bite the bullet and install the then-latest build of 10.8.1, 12B17, just in case something was fixed. Comically, a minute into using it, it disconnected. Same reason. Obviously, it wasn’t going to be fixed in 10.8.1.

I decided that the best course of action would be to pull old copies of the drivers and compare them. From my Time Machine drive, I grabbed the IO80211Family kernel extension from a version of 10.7, an early developer preview of 10.8, the GM version of 10.8, and the 10.8.1 copy I had just installed. My hope is that something in the AirPortAtheros40 driver for my AirPort card would be relevant.

Jackpot.

Mac OS X 10.7.x displayed a bundle version of 504.64.2, and the bundle version for the 10.8 developer preview was 600.65.1. Neither of these copies were faulty. The connection dropping began with the gold master version of 10.8, and continued with the 10.8.1 preview. Not coincidentally, both of these operating systems showed an AirPortAtheros40 driver bundle version of 600.70.23.

I restored the kernel extension with a known working copy, and rebooted. My WiFi connection has been stable ever since.

As one should, I have filed two related bugs. Radar number 11960274 concerns the faulty driver, with a system configuration report and all referred-to kernel extensions enclosed, while radar number 12129328 notes that it is impossible to extract logs from an AirPort base station using AirPort Utility 6.0 or newer.

I posted this for a few reasons. The first is to gain at least one hit for this specific issue. As I mentioned, I haven’t seen it discussed anywhere else. The second reason is a reminder to keep a backup, just in case. Finally, I was surprised that this would continue to persist, despite filing that bug (as you’ll note by the console log above) the day Mountain Lion shipped. As far as I know, my radar hasn’t been looked at, and therefore this problem will not be fixed in 10.8.2. My old-ass MacBook Pro simply isn’t relevant enough any more.


  1. I have removed MAC addresses and identifying information, but it’s representative else wise. ↥︎

  2. Yes, my AirPort network is named “Free Public WiFi”. What of it? ↥︎

  3. It’s worth noting that my MacBook Pro is the single oldest model of Mac that Mountain Lion will run on. In other words, it’s unlikely that many people are still using that model, having upgraded since it was released in 2007. ↥︎

Apple PR:

Apple today announced that Craig Federighi, Apple’s vice president of Mac Software Engineering, and Dan Riccio, Apple’s vice president of Hardware Engineering, have been promoted to senior vice presidents. Federighi and Riccio will report to Apple CEO Tim Cook and serve on Apple’s executive management team.

Well-deserved promotions, from what I’ve seen of both of them. Riccio takes over the responsibilities of Bob Mansfield. Speaking of whom:

Apple also announced that Bob Mansfield, who announced his retirement in June, will remain at Apple. Mansfield will work on future products, reporting to Tim Cook.

They always bring them back for one last job, don’t they?

Paul Haddad:

As some of you may have already noticed the download link for the Tweetbot for Mac alpha no longer works. Twitter’s latest API Changes means now we have a large but finite limit on the number of user tokens we can get for Tweetbot for Mac. We’ve been working with Twitter over the last few days to try to work around this limit for the duration of the beta but have been unable to come up with a solution that was acceptable to them. Because of this we’ve decided its best for us to pull the alpha.

Remember: this “consistent experience” is definitely best for users.

Quentin Fottrell for MarketWatch:

Someone who owned 10,000 hardcover books and the same number of vinyl records could bequeath them to descendants, but legal experts say passing on iTunes and Kindle libraries would be much more complicated.

And one’s heirs stand to lose huge sums of money. “I find it hard to imagine a situation where a family would be OK with losing a collection of 10,000 books and songs,” says Evan Carroll, co-author of “Your Digital Afterlife.” “Legally dividing one account among several heirs would also be extremely difficult.”

These are good questions, but I think that one of the most notable ones regards the concept of ownership:

Part of the problem is that with digital content, one doesn’t have the same rights as with print books and CDs. Customers own a license to use the digital files—but they don’t actually own them.

I think this concept is more contrived than it needs to be. In physical media, one is also given a license to use the files. You are allowed to do many things with a CD, but the music on it is subject to certain rules (copyright law), with certain restrictions. You own the media, but you license the content. One cannot own “Abbey Road” because the ownership of that music belongs to the label.

Since most music sold online today is DRM-free, one can easily transfer that content to their loved ones (though perhaps not according to the license terms, strictly speaking).