MacBookAir4,1 & MacBookAir4,2 (MBA 2011) support

[jonny]

power usage. Is it me or is Oneiric running like a hog? powertop says my laptop's battery is discharging at > 19W/hour when the system is idling. Previously on Natty it was at 7-10W/hour depending on settings. What gives? A solution to this issue might also help reduce the fan noise.

No, it's not your imagination. Battery life under Linux is around half that on OS X, and powertop gives similar results for me. I've made some basic attempts towards tuning power usage using powertop, but only managed to shave a couple of watts at the cost of reducing network throughput by a factor of 10: not a great tradeoff.

There have been some pretty nasty power regressions in Linux recently, and they're probably at the heart of the problem. If your interested, Phoronix has had several articles exploring the matter in depth. There are a few suggested kernel patches drifting around the internet, but I'll just stay close to a power outlet for now - three hours battery life is generally sufficient for me.

[jonny]

I haven't yet tried it myself (I'm not at home), but you can read about a possible partial fix for the excessive power consumption here. There are at least two other power regression issues that aren't addressed by this fix, but this is alleged to be the big one, accounting for much more than half of the excessive power draw compared with earlier kernels.

First person to try it, please post before and after results with notes of any other issues - eg system hanging.

[dfacto]

Just checking in to say that Oneiric seems to be running well on my air too. Not surprising, the 1440x900 hack works with the Samsung display in my laptop. The post-install script made configuration a breeze. Thanks to all of you for all the support!

There's still a few lingering issues I'd like to fix/change and wonder if anyone else has suggestions:

power button. it shuts down the machine, even though the power settings are configured to ask before doing anything when the power button is pressed. this can be really problematic if i fat-finger the delete key.

tap&drag someone mentioned this earlier. i goog'd for x11 settings but didn't find anything that worked.

power usage. Is it me or is Oneiric running like a hog? powertop says my laptop's battery is discharging at > 19W/hour when the system is idling. Previously on Natty it was at 7-10W/hour depending on settings. What gives? A solution to this issue might also help reduce the fan noise.

My power button brings a pop-up to choose one of, restart, shutdown, hibernate, suspend. Perhaps my .Xmodmap fixes this (I don't remeber but I'm fairly certain it does NOT).

The trackpad settings seem to be better with just the defaults (see earlier post). Still I don't think tap&drag works at least not for title bars. I can tap-to-select--which is close--but this is based on a double-click not a single-click.

Power is known issue--there are several phoronix articles on the topic [1] [2]. Koadman, your fans shouldn't be running all the time. Make sure all of the macfanctld related steps of the script were successfully applied. The default macfanctld settings work great for me.

Powertop gives me 15.2W as I am writing this post and 16.8W when I just ran update-grub (basically the same as you report). I'll try the "pcie_aspm=force" boot flag right now.

After reboot: I do not notice any difference when using the "pcie_aspm=force" flag nor do I see system hangs. Perhaps this flag is ineffective because I am using legacy mode rather than pure EFI.

Oh, and I forgot to mention some time back. It takes me about ~20 seconds to to a full reboot (ballpark). The refit Tux logo takes 8-9 seconds (this measurement was precise) and the rest is the standard linux stuff (grub+startup sequence). If you have anything more than 10 sec at the Tux logo, then something is wrong.

Update: Someone try this and see what happens (here are some more tweaks). In general that thread seems to be the best source of information, particularly since steevven1 has seemingly tried every boot arg under the sun. He has also filed bug 834037 which I encourage you to check-out and click the "Does this bug affect you" link.

[happicow]

Wow so much to reply to. To any 11" users out there: get me your lsusb output if you want your trackpad, fn keys, keyboard backlight to work.

Hi,

I don't know if you are still looking for the lsusb output from an 11 inch air, but here it is. I tried to update your script myself but I am not sure where to look. Thanks for putting it out there though!

- Chris

Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 001 Device 002: ID 0424:2513 Standard Microsystems Corp.
Bus 001 Device 003: ID 05ac:850a Apple, Inc.
Bus 002 Device 002: ID 0424:2513 Standard Microsystems Corp.
Bus 001 Device 004: ID 0a5c:4500 Broadcom Corp. BCM2046B1 USB 2.0 Hub (part of BCM2046 Bluetooth)
Bus 001 Device 005: ID 05ac:0249 Apple, Inc.
Bus 001 Device 006: ID 05ac:820a Apple, Inc.
Bus 001 Device 007: ID 05ac:820b Apple, Inc.
Bus 001 Device 008: ID 05ac:821f Apple, Inc.

[koadman]

Thanks for the tips, power usage comes down to 9-10 watts by adding i915.i915_enable_rc6=1 to the kernel command-line. Haven't tried the sparse irq kernel config. The phoronix article suggests other tweaks but their effect is small compared to i915.i915_enable_rc6=1. I haven't observed any graphics rendering problems so far, but I haven't tried much 3D either.

[dfacto]

@koadman Looks like post-install-oneiric.sh was not installing macfanctld since this is not in the Mactel ppa. The problem is fixed now.

[dfacto]

To whoever posted the script on git hub: do you mind removing it? I don't think its smart to have old copies floating around (unless you keep updating it). For the time being I prefer to keep editing it on my own machine because this is simpler for me to maintain. Once we all agree its stable, then we can share its maintenance (or ideally someone else can carry the torch). In general, it seems to me the Mactel community is quite fractured (for God's sake look at the Ubuntu Mactel commuity wiki!) so let's at least try to keep the MBA4 stuff in one place.

Oh, on the issue of "stability": I just reinstalled Oneiric last night and re-ran the script. Aside from discovering the macfanctld issue it seems to work quite nicely. Next it would be nice to /var/log all the build messages so it is more clear what changes are being made to the system. (I put in echo's all through-out but at run-time you can hardly tell they're there.)

[kiffykroker]

Appreciate all the effort you guys have put in to this. I have an 11" Air running 11.04 / 3.1.0-0301rc4-generic, the resolution is correct thanks to dfactos fix-i915.sh(*) but I've encountered the following issues so far:

- the fans are going all the time
- trackpad only recognize movement and primary clicks
- fn key doesn't work

Browsing through the thread it seems that some of you have experienced at least some of these issues, anyone solved them? Perhaps dfactos post install script fixed the last two for the 13"?

Also, I noticed that dfacto asked for lsusb from an 11", here's mine

$ lsusb
Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 002 Device 002: ID 0424:2513 Standard Microsystems Corp. 
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 001 Device 008: ID 05ac:821f Apple, Inc. 
Bus 001 Device 005: ID 05ac:024a Apple, Inc. 
Bus 001 Device 004: ID 0a5c:4500 Broadcom Corp. BCM2046B1 USB 2.0 Hub (part of BCM2046 Bluetooth)
Bus 001 Device 003: ID 05ac:850a Apple, Inc. 
Bus 001 Device 002: ID 0424:2513 Standard Microsystems Corp. 
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

(*) = I had to change $(uname -r) on line 77 in fix-i915.sh to a hard coded folder

# patch
cd linux-3.0.0/drivers/gpu/drm/i915/

since the kernel source the script grabbed was linux-3.0.0 (or atleast the folder name) and not $uname -r

[dfacto]

Appreciate all the effort you guys have put in to this. I have an 11" Air running 11.04 / 3.1.0-0301rc4-generic (*), the resolution is correct thanks to dfactos script but I've encountered the following issues so far:
- the fans are going all the time

Mactel doesn't have macfanctld in the ppa for Oneiric. See what the revised script does.

- trackpad only recognize movement and primary clicks
- fn key doesn't work

Both of these are because the script is for 13" and I didn't have the product codes until recently. I'll have to make a patch when I get time but anyone should be able to figure it what I did from the post-install-natty.sh script. (The oneiric version didn't do it because I submitted those patches to mainline so they are already in 3.0.0.)

There is extremely high probability that I won't have time to do this--I graduate in two weeks!

Browsing through the thread it seems that some of you have experienced at least some of these issues, anyone solved them? Perhaps dfactos script fixed the last two for the 13"?

Also, I noticed that dfacto asked for lsusb from an 11", here's mine

$ lsusb
Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 002 Device 002: ID 0424:2513 Standard Microsystems Corp. 
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 001 Device 008: ID 05ac:821f Apple, Inc. 
Bus 001 Device 005: ID 05ac:024a Apple, Inc. 
Bus 001 Device 004: ID 0a5c:4500 Broadcom Corp. BCM2046B1 USB 2.0 Hub (part of BCM2046 Bluetooth)
Bus 001 Device 003: ID 05ac:850a Apple, Inc. 
Bus 001 Device 002: ID 0424:2513 Standard Microsystems Corp. 
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

Thanks; this is necessary information to make the patch. Also need the bluetooth product id.

(*) = I had to change $(uname -r) on line 77 in fix-i915.sh to a hard coded folder

# patch
cd linux-3.0.0/drivers/gpu/drm/i915/

since the kernel source the script grabbed was linux-3.0.0 (or atleast the folder name) and not $uname -r

You must have a really old fix-i915.sh? My version has a `cut` to lop off the excess from uname -r. Make sure you are actually downloading a fresh copy and not getting your browsers cache (perhaps use wget).

[jonny]

Thanks for the tips, power usage comes down to 9-10 watts by adding i915.i915_enable_rc6=1 to the kernel command-line. Haven't tried the sparse irq kernel config. The phoronix article suggests other tweaks but their effect is small compared to i915.i915_enable_rc6=1. I haven't observed any graphics rendering problems so far, but I haven't tried much 3D either.

I can confirm that I see very similar results on my 13" Air, too. Huge power saving and no obvious adverse effects.