Macbook Pro

Message

je_fro · Post by **je_fro** » Sun May 06, 2007 11:30 pm

WPA is working fine with the new madwifi-ng build.
Now all that's left is to solve these ridiculous heat and power consumption issues...

tylerwylie · Post by **tylerwylie** » Mon May 07, 2007 5:58 am

je_fro wrote:WPA is working fine with the new madwifi-ng build.
Now all that's left is to solve these ridiculous heat and power consumption issues...

You have a really creepy avatar.

Pic0 · Post by **Pic0** » Mon May 07, 2007 10:16 am

je_fro wrote:Now all that's left is to solve these ridiculous heat and power consumption issues...

I think the only way to solve this issue is to convince AMD/ATI to fix their crappy drivers. POWERplay is working on MBPv1, why not on MBPv2 ?

genmich · Post by **genmich** » Mon May 07, 2007 1:01 pm

I have a problem with udev and the uvcvideo module. When I boot udev tries 3 times to load the uvcvideo module (in the init process). I tried to log it with RC_BOOTLOG="yes" in the rc.conf file but it didn't log these messages.

If I uncomment the following line in the udev rules (50):

Code: Select all

# Module autoloading

# Autoload modules that lack aliases but have them defined inutoload modules
ENV{MODALIAS}=="?*", ACTION=="add", RUN+="modprobe.sh $env{MODALIAS}"

it won't load them at all. So I think that this rule tries to load it more than once. It also loads the ath module which I don't want to be loaded all the time (only if I need it, not directly at boot).

How can I change the udev thing?

Another thing mentioned in the mactel mailing list some time ago: having a proper kernel config for every MB(P) [C1D/C2D]. I have mine here (MBP C1D / Kernel 2.6.21-mactel): http://migaweb.de/foren/.config
Perhaps someone can have a look and point out some enhancements or mistakes in it.

nss · Post by **nss** » Mon May 07, 2007 6:48 pm

With the new madwifi build, do you need to have the kernel built from source / .config file? I did an svn checkout of madwifi-hal-0.9.30.13. i am able to modprobe ath_pci and see networks but dhcpcd will not retrieve an ip. The kernel source / .config seems unrelated since I got as far as seeing networks but I can't see what I'm doing wrong.

dmvianna · Post by **dmvianna** » Mon May 07, 2007 8:51 pm

miga wrote:When I boot udev tries 3 times to load the uvcvideo module (in the init process).

Same here.

je_fro · Post by **je_fro** » Mon May 07, 2007 9:02 pm

I took uvcvideo out of /etc/modules.autoload.d/kernel-2.6 and no more firmware errors and udev only tries to load it once. I didn't make any changes to any config files regarding uvcvideo. It's broken on 2.6.21 for the time being, but under 2.6.20 everything was fine.
Also, I have ACCEPT_KEYWORDS="~amd64" in make.conf.

genmich · Post by **genmich** » Tue May 08, 2007 8:21 am

Hmm...i don't have it in my modules.autoload.d/kernel-2.6 file! I only have appletouch and usbhid in it.

john-boro · Post by **john-boro** » Sat May 12, 2007 6:49 pm

Hi,

I've just installed 2.6.21.1 and the keyboard backlight is no longer working. Pommed reports no errors, no does the applesmc tool. Anyone else had this?

Pic0 · Post by **Pic0** » Sat May 12, 2007 7:49 pm

john-boro wrote:I've just installed 2.6.21.1 and the keyboard backlight is no longer working. Pommed reports no errors, no does the applesmc tool. Anyone else had this?

Maybe this is due to the new behaviour of pommed ? You have to "switch on" keyboard backlight using F8, then you can adjust the blacklight level with F9 and F10. Before, you just had to press F10 to enable backlight.

Also are you using a custom kernel or the ebuild from my overlay ?

john-boro · Post by **john-boro** » Sat May 12, 2007 11:44 pm

pic0,

thanks for the swift reply.

EDIT: This is either a bit confusing or just very embarassing. I took all the LED trigger support from my kernel, and tried again, still no luck. Then I realised the applesmc module was not getting loaded on startup and modprobed it, then everything worked. Weird because even when I modprobed applesmc before, it would still not work. I can only guess the LED trigger stuff in the kernel may have something to do with it. I recommend compiling applesmc in.

regards

john

je_fro · Post by **je_fro** » Mon May 14, 2007 7:06 pm

Well, that new kernel (v2.6.21.1-mactel-r2 with the AHCI and CFS patches) is a dud... Using it my system seemed a little slower and became unresponsive when mounting CIFS shares....I couldn't unmount them, but then again I couldn't tell if they were mounted either...odd.
For example... /home/me/mountpoint is where I'd mount a remote CIFS share. Using bash tab completion I can just do cd mo<tab> and the name "mountpoint/" autocompletes. Not now. "mount mountpoint/" (it's in fstab so this works) seems to work but "ls mountpoint/" hangs that bash shell. So does "cd mountpoint/" or anything else...
I'll try searching and see if it's a problem with CFS or AHCI....I don't know what the culprit is...

Pic0 · Post by **Pic0** » Wed May 16, 2007 3:20 pm

je_fro wrote:Well, that new kernel (v2.6.21.1-mactel-r2 with the AHCI and CFS patches) is a dud... Using it my system seemed a little slower and became unresponsive when mounting CIFS shares....I couldn't unmount them, but then again I couldn't tell if they were mounted either...odd.
For example... /home/me/mountpoint is where I'd mount a remote CIFS share. Using bash tab completion I can just do cd mo<tab> and the name "mountpoint/" autocompletes. Not now. "mount mountpoint/" (it's in fstab so this works) seems to work but "ls mountpoint/" hangs that bash shell. So does "cd mountpoint/" or anything else...
I'll try searching and see if it's a problem with CFS or AHCI....I don't know what the culprit is...

It is most probably related with CFS. Please see with Ingo Molnar on the Linux Kernel ML. I can't give you any advice excepted that you should not use if it cause problems. Please note that I have enabled it here since v8, and I did not notice any problem (but I can't see a significant improvement either).

I have just commited mactel-sources-2.6.21.1-r4. Suspend2 is now 2.2.9.16; with USE=powersaving, it will also use the patch by Matthew Garrett to reduce wakeups caused by appletouch ( http://www.linuxpowertop.org/known.php#appletouch ).

You may also have noticed the SVN ebuild for powertop. With USE=powersaving, I get between 450 and 650 wakeups-from-idle per second on a full KDE desktop (that was about 1100/s without). see linuxpowertop.org

Regards,

je_fro · Post by **je_fro** » Wed May 16, 2007 10:04 pm

That's great news about the powersaving...I have to use OSX most of the time because of the heat and the battery usage under linux.
Also, the AHCI driver seems to be an improvement:

Code: Select all

WITH AHCI
# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   5630 MB in  2.00 seconds = 2820.61 MB/sec
 Timing buffered disk reads:  116 MB in  3.04 seconds =  38.11 MB/sec

WITHOUT AHCI
# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   6562 MB in  2.00 seconds = 3288.07 MB/sec
 Timing buffered disk reads:  100 MB in  3.03 seconds =  32.95 MB/sec

Pic0 · Post by **Pic0** » Thu May 17, 2007 10:53 am

je_fro wrote:That's great news about the powersaving...I have to use OSX most of the time because of the heat and the battery usage under linux.

The GPU is still what drains the most power on the MBPv2. Even with all the open source drivers fixed, don't hope for more than 2 hour of battery life.

je_fro wrote: Also, the AHCI driver seems to be an improvement:

Code: Select all

WITH AHCI
# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   5630 MB in  2.00 seconds = 2820.61 MB/sec
 Timing buffered disk reads:  116 MB in  3.04 seconds =  38.11 MB/sec

WITHOUT AHCI
# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   6562 MB in  2.00 seconds = 3288.07 MB/sec
 Timing buffered disk reads:  100 MB in  3.03 seconds =  32.95 MB/sec

Yes, I forgot to report about it earlier, but I gain about 8-9MB/sec with AHCI. However the hard drive I'm using is not the standard one in the MBPv2. It is a SAMSUNG HM160JI (160GB), which is a bit faster.

Code: Select all

# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   5370 MB in  2.00 seconds = 2690.24 MB/sec
 Timing buffered disk reads:  128 MB in  3.02 seconds =  42.42 MB/sec

je_fro · Post by **je_fro** » Thu May 17, 2007 11:28 am

While we're on the subject....
Has anyone been able to get "/opt/bin/aticonfig --lsp" to work properly on a MBP core2duo?

john-boro · Post by **john-boro** » Mon May 21, 2007 6:49 pm

I know most of us who are l33t enough to install gentoo on an mbp probably don't care about beryl, but I think it's..well.. nice. So I've installed xgl from the overlay and beryl from portage, and have it working dandily. One thing I'm not sure about is how to get vsync working. I have sync to vblank checked in beryl's settings, but any time I've used xgl on a computer I've never been able to make it stop tearing.

anyone solved this?

cheers

john

PS If anyone remembers, I used to use debian on my mbp. I still have it, but I'm now dual booting gentoo since a few days ago, and things are looking pretty damn good

je_fro · Post by **je_fro** » Mon May 21, 2007 7:03 pm

john-boro wrote:I know most of us who are l33t enough to install gentoo on an mbp probably don't care about beryl, but I think it's..well.. nice.

HaHaHa....I'm so fscking 1337 that I _DO_ run beryl on my MBP, but I do so using the xeffects overlay. I also have Sync to VBlank checked and haven't noticed anything odd. Perhaps you should give xeffects a try?

je_fro · Post by **je_fro** » Mon May 21, 2007 7:08 pm

Pic0's mactel-sources has a "powersaving" USE flag that enabled a patch altering the touchpad code. With USE="powersaving" my mouse cursor would be real "wiggly" and the buttons wouldn't function properly after an X restart. Disabling that USE flag fixed the issue. It didn't appear to have any noticeable effect of battery life either.

Pic0 · Post by **Pic0** » Tue May 22, 2007 8:13 am

je_fro wrote:Pic0's mactel-sources has a "powersaving" USE flag that enabled a patch altering the touchpad code. With USE="powersaving" my mouse cursor would be real "wiggly" and the buttons wouldn't function properly after an X restart. Disabling that USE flag fixed the issue. It didn't appear to have any noticeable effect of battery life either.

Yes this is a commun issue with this patch. see http://lkml.org/lkml/2007/5/15/259
When it happens you can remove/reload the "appletouch" module.
And regarding battery life, it may save maybe 0,5W, but it is unnoticeable when we have a GPU drainning ~10W at its side... :/

je_fro · Post by **je_fro** » Wed May 23, 2007 11:29 am

Pic0 - Looks like the madwifi folks chagned the tree on ya...

Pic0 · Post by **Pic0** » Wed May 23, 2007 11:56 am

je_fro wrote:Pic0 - Looks like the madwifi folks chagned the tree on ya...

You are right. The good news is that our branch has become trunk ! So the next official version of madwifi will support the AR5008 chipset natively !

I will fix the ebuild ASAP.

dmvianna · Post by **dmvianna** » Sat May 26, 2007 2:30 am

The speakers stopped working with 2.6.21-r2 + mactel patches here. Anyone got the same?

Line out works, though.

john-boro · Post by **john-boro** » Sat May 26, 2007 9:28 am

I'm using 2.6.21.1 with mactel patches and the speakers work fine.

dmvianna · Post by **dmvianna** » Sat May 26, 2007 9:47 am

Ah, forgot to tell. I use gentoo-sources. So it's 2.6.21-gentoo-r2 + mactel.

And it's the old MacBook Pro (Core Duo). No speakers again...

Macbook Pro

success!!

Re: success!!

Re: success!!

udev uvcvideo / perfect kernel

Re: udev uvcvideo / perfect kernel

Keyboard Backlight in 2.6.21

Re: Keyboard Backlight in 2.6.21

Beryl, xgl, tearing

Re: Beryl, xgl, tearing