Macbook Air 2013

frank604 · 2015-09-02 02:41:12

@step-2 I rarely hear my fans. Only time it ramps up is when I'm on bed and watching a video. Otherwise it stays cool as a pickle. Ok I lied, slightly warm to the touch. As a checklist can you confirm
1. Suppressed all gpe interrupts that are having an interrupt shitstorm? For me it is 66 and 4e.
2. Test watching webm or mp4 video in firefox or other non-chrom* browser. Link us to a video that reproduces it for you.
3. Detail your powersaving methods. Have you set anything up for intel video?
4. MBA on a desk or on your lap?

If this was a hardware issue then many, if not all, of us should experience this together. Since it is not happening here, it has to be something else? Maybe someone has a better idea.

Also, you can set min fan speed in macfanctl.conf although I never modified it before. The link you provided does the same for mac osx. It just allows you to increase from Apple's min fan speed but not going below it.

step-2 · 2015-09-02 09:48:34

@frank604 thanks for the reply ,

1. gpe4e i suppress every time on boot (although i see 942 interrupts from it), osx update took care of gpe66 with a firmware update (no more high cpu utilization).
2. even firefox have the same problem , temperature reach 95 c while watching youtube video via viewtube script in mp4 format (not flash not HTML5) :
http://isebaro.com/viewtube/?ln=en
i also have libva-intel-driver installed
3. i don't have any power saving atm , just acpi_osi= in kernel parameters (most of the time on power).
4. MBA is on a desk , it's my main machine .

it could be because i don't have any powersaving configs , that is only because when i first bought the MBA in late 2013 (around linux 3.10) everything ran fine without any configs , also my DE is gnome if that change anything .

the smcFanControl was a lsat desperate measure .

Last edited by step-2 (2015-09-02 10:08:28)

frank604 · 2015-09-02 18:35:36

I think I remember you or someone else saying a firmware update fixed gpe66. That fix never fixed mine. So I would double check. It also hasn't fixed many others' 66 either from recent postings. I'll test that script. Could be the cause?

edit: I just enabled that script via greasemonkey and I do not feel or hear my fans going crazy, nor is it overheating. I also tested on gnome.

Last edited by frank604 (2015-09-03 00:25:03)

step-2 · 2015-09-03 10:08:40

@frank604 i just checked , gpe66 has 17 interrupts (used to be thousands before the update) also fixed on windows too , gpe4E has 900 interrupts before i disable it ,

if you can't reproduce the issue , it could be that i don't have any power savings turned on then ,

as i said before i don't have them because everything used to work fine in the 3.10 days , i have even tried downgrading the kernel with no luck ,

if you go a couple of pages back in this thread you would see a couple of people with the same issue ,

am always trying new kernels & linux-ck , i will also give your powersavings config a try ,

out of curiosity , what are you using to monitor the temperature & the fan rpm ?

thanks again for your reply .

frank604 · 2015-09-04 00:26:40

I use sensors. I only go up a few degrees hovering 62-64C. Fan stays at min of around 2000 for this temp for me.

step-2 · 2015-09-08 04:35:45

wow 4.2 seem to be answering my prayers ,

at initial testing the fans go up to 2000 rpm and the temperature at 70C ( not ideal , but much better than 5000 rpm & 90C with previous kernels ) ,

also i made the switch to broadcom-wl-dkms from broadcom-wl , don't know if it's what made the difference but i think it's the kernel ,

anyone ?

Last edited by step-2 (2015-09-08 11:07:33)

hobarrera · 2015-09-08 13:11:41

step-2 wrote:

also i made the switch to broadcom-wl-dkms from broadcom-wl , don't know if it's what made the difference but i think it's the kernel

Most likely the kernel.

broadcom-wl-dkms is based on broadcom-wl, and the only difference is that the former auto-rebuilds on kernel upgrade, while the latter does not. The resulting binary driver itself is the same. (or should be unless the -dkms one is out-of-date)

step-2 · 2015-09-08 13:35:54

hobarrera wrote:

Most likely the kernel.
broadcom-wl-dkms is based on broadcom-wl, and the only difference is that the former auto-rebuilds on kernel upgrade, while the latter does not. The resulting binary driver itself is the same. (or should be unless the -dkms one is out-of-date)

of course , it download the same binary during build , i mentioned it for anyone having the same issue (heat & fan noise), so that they know all the variables ,

also @frank604 mentioned some dmesg errors regarding wl ,

4.2 seems to be the solution for the fan & heat issues for now , hope kernel updates doesn't produce any regressions .

frank604 · 2015-09-08 18:56:16

@step-2 my dmesg errors on wl were due to a script calling on iwconfig. wow, that was awhile ago. crazy you still remember that.

Is it just me? I had this error on building mba6x_bl for linux 4.2.0-3

Kernel preparation unnecessary for this kernel.  Skipping...

Building module:
cleaning build area....
make KERNELRELEASE=4.2.0-3-mba all KVERSION=4.2.0-3-mba....(bad exit status: 2)
Error! Bad return status for module build on kernel: 4.2.0-3-mba (x86_64)
Consult /var/lib/dkms/mba6x_bl/39.fc64de1/build/make.log for more information.

cat /var/lib/dkms/mba6x_bl/39.fc64de1/build/make.log

DKMS make.log for mba6x_bl-39.fc64de1 for kernel 4.2.0-3-mba (x86_64)
Tue Sep  8 11:52:27 PDT 2015
make -C /lib/modules/4.2.0-3-mba/build M=/var/lib/dkms/mba6x_bl/39.fc64de1/build modules
make[1]: Entering directory '/usr/lib/modules/4.2.0-3-mba/build'
  CC [M]  /var/lib/dkms/mba6x_bl/39.fc64de1/build/mba6x_bl.o
/var/lib/dkms/mba6x_bl/39.fc64de1/build/mba6x_bl.c: In function ‘platform_probe’:
/var/lib/dkms/mba6x_bl/39.fc64de1/build/mba6x_bl.c:336:2: error: implicit declaration of function ‘acpi_video_dmi_promote_vendor’ [-Werror=implicit-function-declaration]
  acpi_video_dmi_promote_vendor();
  ^
cc1: some warnings being treated as errors
scripts/Makefile.build:264: recipe for target '/var/lib/dkms/mba6x_bl/39.fc64de1/build/mba6x_bl.o' failed
make[2]: *** [/var/lib/dkms/mba6x_bl/39.fc64de1/build/mba6x_bl.o] Error 1
Makefile:1386: recipe for target '_module_/var/lib/dkms/mba6x_bl/39.fc64de1/build' failed
make[1]: *** [_module_/var/lib/dkms/mba6x_bl/39.fc64de1/build] Error 2
make[1]: Leaving directory '/usr/lib/modules/4.2.0-3-mba/build'
Makefile:8: recipe for target 'all' failed
make: *** [all] Error 2

commenting out the line is the work around. mba6x_bl is functioning still. just thought I'd ask you since you already jumped up to 4.2

step-2 · 2015-09-08 22:07:02

@frank604 crazy is strong word but ok, lol ,

i remember the wl issue cause it used to show up when switching tty , but it's resolved now ,

and sorry i don't use mba6x_bl .

johnk · 2015-09-12 23:57:12

Hi,

Firstly, thanks to everyone here for providing an excellent source of info for getting linux running well on these devices.

Now to my question. I have been seeing the same problem as below, with serious sata errors after connecting power and hibernating.

frank604 wrote:

I forgot to ask the group if you guys are experiencing ata errors only after a suspend/resume? Anyone? I've tried

libata.force=noncq
libata.force=1.5G
libata.force=1.5G,noncq
libata.force=3.0G,noncq

[  +0.000001] ata1.00: status: { DRDY }
[  +0.000003] ata1: hard resetting link
[  +0.719684] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 320)
[  +0.000465] ata1.00: unexpected _GTF length (8)
[  +0.000546] ata1.00: unexpected _GTF length (8)
[  +0.000073] ata1.00: configured for UDMA/133
[  +0.000110] ata1: EH complete
[  +4.384538] ata1.00: exception Emask 0x10 SAct 0x80000 SErr 0x4040000 action 0xe frozen
[  +0.000004] ata1.00: irq_stat 0x80000040, connection status changed

Edit: Nevermind, this issue was due to tlp. I solved it by commenting out one line in /usr/sbin/tlp

#set_sata_link_power $1

Following these steps and disabling the tlp sata link power stuff does help, but it has a terrible impact on battery life (from ~9 hours -> ~3 hours).

Has anybody got any further tips on how to configure power management to avoid these errors?

(my current workaround of only charging the device after hibernating it first is not a great long term solution)

Thanks,
John

johnk · 2015-09-13 01:16:44

Here is what I have for my trackpad config for my Macbook Air 6,2 13" which works well for me based on combining suggestions from a number of sources:

Section "InputClass"
            MatchIsTouchpad "on"
            Identifier      "Touchpads"
            Driver          "mtrack"
            Option          "Sensitivity" "0.65"
            Option          "FingerHigh" "12"
            Option          "FingerLow" "4"
            Option          "IgnoreThumb" "true"
            Option          "IgnorePalm" "true"
            Option          "TapButton1" "0"
            Option          "TapButton2" "0"
            Option          "TapButton3" "0"
            Option          "TapButton4" "0"
            Option          "ClickFinger1" "1"
            Option          "ClickFinger2" "3"
            Option          "ClickFinger3" "3"
            Option          "ButtonMoveEmulate" "true"
            Option          "ButtonIntegrated" "true"
            Option          "ClickTime" "25"
            Option          "BottomEdge" "40"
            Option          "SwipeLeftButton" "8"
            Option          "SwipeRightButton" "9"
            Option          "SwipeUpButton" "0"
            Option          "SwipeDownButton" "0"
            Option          "ScrollDistance" "75"
            Option          "ScrollUpButton" "4"
            Option          "ScrollDownButton" "5"
            Option          "ThumbSize" "35"
            Option          "PalmSize" "55"
            Option          "DisableOnThumb" "false"
            Option          "DisableOnPalm" "true"
            Option          "SwipeDistance" "1000"
            Option          "ScrollLeftButton" "7"
            Option          "ScrollRightButton" "6"
            Option          "AccelerationProfile" "2"
            Option          "ConstantDeceleration" "2.0"  # Decelerate endspeed
            Option          "AdaptiveDeceleration" "2.0"  # Decelerate slow movements
EndSection

Pitmairen · 2015-09-13 06:05:17

johnk wrote:

Following these steps and disabling the tlp sata link power stuff does help, but it has a terrible impact on battery life (from ~9 hours -> ~3 hours).
Has anybody got any further tips on how to configure power management to avoid these errors?
(my current workaround of only charging the device after hibernating it first is not a great long term solution)
Thanks,
John

You can try to edit /etc/default/tlp and chage SATA_LINKPWR_ON_BAT from min_power to medium_power

Read more here: http://linrunner.de/en/tlp/docs/tlp-con … .html#alpm

You will probably have to re-enable set_sata_link_power in the /usr/sbin/tlp for this to work.

johnk · 2015-09-14 03:52:25

Pitmairen - Thanks. I'll try this for a week or so and report back on my findings.

Pitmairen · 2015-09-14 11:58:34

I just noticed that I still get sata errors with SATA_LINKPWR_ON_BAT set to medium_power so it may have to be set to max_performance

frank604 · 2015-09-14 20:24:42

Allowing tlp to set sata link power does cause ata errors. Where the bug is I do not know nor have the capabilities to find it. What does powertop tunables say for enable sata link power management for you? If it says bad, try this service which auto-tunes this.

[Unit]
Description=Powertop tunings

[Service]
Type=oneshot
ExecStart=/usr/bin/powertop --auto-tune

[Install]
WantedBy=multi-user.target

and enable/start it. Does your battery life improve?

johnk · 2015-09-15 08:25:28

What does powertop tunables say for enable sata link power management for you?

After a clean boot it comes up as Bad

So I went back into tlp and commented out the sata power line:

 #set_sata_link_power $1

and I added your systemd powertop service to /etc/systemd/system and things are looking good :-)

I am getting ~5.5W consumption under low load which is good enough for me and so far no sata errors.
(I was previously able to produce errors by hibernating and un-hibernating with the power connected)

I'll check back in a week or two to confirm it is still working well.

Thanks!

(I put a link to this thread in the Arch wiki to help other people who have this problem)

Last edited by johnk (2015-09-15 10:30:10)

MartiniMoe · 2015-11-03 17:34:51

Hi everyone.
I tried Arch on my MBA 6,2 a few months ago with kernel 4.0 or 4.1 and I had problems with supending (MBA woke up after ~2 seconds after suspending for the second time).
I just wanted to know, if anyone still has this bug? Is arch able to serve as a daily driver on our MBA? Then I would probably give it another try.

Thanks!

Pitmairen · 2015-11-03 20:12:43

I still have the suspend problem, but I have fixed this by just disabling everything in /proc/acpi/wakeup. I have also disabled suspend when closing the lid, and instead suspend by using a keyboard shortcut.
This works great for me, and I have not had any unexpected wake ups after doing this. The only problem I currently have is that it gets a little bit hot when the power plug is connected, but this is not a big problem for me.

step-2 · 2015-11-28 05:16:39

for the wake up after sleep issue , this should help :

https://wiki.archlinux.org/index.php/Ma … _Hibernate

also disabling gpe4E disable suspend via closing the lid ,

hope it's not doing something damaging to the ssd , as it sets high interrupts when not disabled .

Last edited by step-2 (2015-12-03 16:47:50)

brainwave · 2016-03-25 16:06:33

Hi guys, been long since I posted here. I have an issue with power consumption.

Last time I tried arch on my MBA, default installation after configuring power saving settings took around 5-5.5 Watt. Now same setup is taking 6.8 Watts minimum. With just web browser, its taking around 7.6-8.5 Watts.

I have installed tlp and powertop. Powertop shows all tunables as good. In the running processes page, it shows my WiFI is taking around 4.9 Watts constantly. What could be the issue behind this? I have written this service to enable WiFi power saving by creating a script 70-wifi-powersave-rules in /etc/udev/rules.d

Any advice?

Last edited by brainwave (2016-03-25 16:11:34)

Excubia · 2016-05-07 16:37:44

Broadcom-wl has forced me into upgrading the kernel 3.17 ----> 4.5

Heat is back again… What can I do?

ste · 2016-07-21 15:32:50

anyone have experience running arch on the 7,1 from 2015?

ugis · 2016-08-12 10:10:17

Hi everyone... I suspect that most of the things are working for people with MBA 2013 now, since this thread has been relatively quiet lately. However, I have an issue which I can not find solution for.

The problem is that after exitting suspend-to-RAM, I can log in, but it seems that there is no access to my SSD whatsoever. Any terminal command spits out "input/output error" and, when I shutdown, the normal shutdown procedure complains about lost write to sda and that's it. Then I have to do hard-power-off. I have tried passing kernel parameter libata.force=1:noncq found in wiki, however that did not change anything. I could not find any other solution. I thought about trying to prevent SSD entering sleep mode at all, but from my googling results it does not seem to be possible without tampering with firmware?

The thing is that I installed OWC 1TB SSD in my laptop, maybe it is not very friendly with linux...

Any comments or suggestions?

hobarrera · 2016-08-27 03:55:41

After the latest OS update (pacman -Syu), the wireless network has stopped working for me. Has anyone else been having issues? I still haven't pinpointed which package broke stuff, and don't have many leads.

Arch Linux

#351 2015-09-02 02:41:12

Re: Macbook Air 2013

#352 2015-09-02 09:48:34

Re: Macbook Air 2013

#353 2015-09-02 18:35:36

Re: Macbook Air 2013

#354 2015-09-03 10:08:40

Re: Macbook Air 2013

#355 2015-09-04 00:26:40

Re: Macbook Air 2013

#356 2015-09-08 04:35:45

Re: Macbook Air 2013

#357 2015-09-08 13:11:41

Re: Macbook Air 2013

#358 2015-09-08 13:35:54

Re: Macbook Air 2013

#359 2015-09-08 18:56:16

Re: Macbook Air 2013

#360 2015-09-08 22:07:02

Re: Macbook Air 2013

#361 2015-09-12 23:57:12

Re: Macbook Air 2013

#362 2015-09-13 01:16:44

Re: Macbook Air 2013

#363 2015-09-13 06:05:17

Re: Macbook Air 2013

#364 2015-09-14 03:52:25

Re: Macbook Air 2013

#365 2015-09-14 11:58:34

Re: Macbook Air 2013

#366 2015-09-14 20:24:42

Re: Macbook Air 2013

#367 2015-09-15 08:25:28

Re: Macbook Air 2013

#368 2015-11-03 17:34:51

Re: Macbook Air 2013

#369 2015-11-03 20:12:43

Re: Macbook Air 2013

#370 2015-11-28 05:16:39

Re: Macbook Air 2013

#371 2016-03-25 16:06:33

Re: Macbook Air 2013

#372 2016-05-07 16:37:44

Re: Macbook Air 2013

#373 2016-07-21 15:32:50

Re: Macbook Air 2013

#374 2016-08-12 10:10:17

Re: Macbook Air 2013

#375 2016-08-27 03:55:41

Re: Macbook Air 2013

Board footer