MacBook Pro 2012 Retina

frigaut · 2012-08-01 14:25:16

BlueDragonX. Thanks for your help. I believe you might be right on with the link speed.
I have a dmesg with drm.debug=6 at http://maumae.net/retina/dmesg_intel_drm_debug if anyone is interested.
The mode that seems to be used is:
[ 9.856125] [drm:drm_mode_debug_printmodeline], Modeline 17:"2880x1800" 60 337750 2880 2928 2960 3040 1800 1803 1809 1852 0x48 0x9
which seems to me the correct one.
On top of that, I can use xrandr and switch resolution and I get effectively a different resolution but the same corruption, which would be consistent with your explanation. I thought for a while the i915 was not getting the correct EDID, so I dumped the screen EDID from the nvidia-settings (when using nvidia) and used that as drm_kms_helper.edid_firmware, but that didn't help the corruption (although it did change the resolution as reported by xrandr).
ok. Is the work you're doing for nouveau going to help the issue with the intel driver, if it's link related?

BlueDragonX · 2012-08-01 14:48:32

That's the exact modeline I get out of nouveau, so you're right on there. That's also the exact behavior I'm experiencing with nouveau. Regardless of the mode I always get the same screen corruption.

I'm not 100% convinced it has to do with the available DP bandwidth. I've tried hardcoding it but the card refuses to take the settings. I have to look into it further.

The work I do won't effect the Intel driver, though what I discover should be helpful when implementing the fixes in i915. Unfortunately the DisplayPort link training code is specific to each driver. I think that's something that should be pulled down into the DRM module rather than depending on each driver to implement their own, but oh well...

BlueDragonX · 2012-08-01 15:00:59

This line right here out of your dmesg:

DP link bw 0a lane count 4 clock 270000 bpp 24

That's part of the DP link training and it's configuring it the same way as nouveau does, four lanes at 2.7Gbps. So if it is an issue with the Retina exceeding that bandwidth it's consistent across both devices. The link is configured based on the mode and panel capabilities, so that makes sense.

I still need to try a couple of things to validate whether or not this is the situation. Real work keeps getting in the way =P

frigaut · 2012-08-02 13:35:27

BlueDragonX,
one more piece of information: the nvidia blob is also reporting:
Connection link: 4 lanes @ 2.70 Gbps
Signal: DisplayPort
and it's displaying ok in native resolution at 59.99Hz.
Doesn't seem to cut it with your theory.

mzanetti · 2012-08-03 08:20:46

Somewhat unrelated to the current discussion with the graphics chip, but maybe of use for some of you: Replacing xf86-input-synaptics with xf86-input-mtrack makes the touchpach work much better. I can finally click without accidentally moving the content under the mouse. In combination with xbindkeys I can use stuff like 3-finger swipe to change desktop or pinch zoom (that one not being really smooth though).

I also have upgraded to linux 3.5. final in the meantime and it seems that resuming from suspend is working fine now (in combinanation with the 304.22 nVidia driver).

mzanetti · 2012-08-03 15:20:58

Ok... I have started a wiki page for this thing. For now its pretty basic but feel free to extend/fix it: https://wiki.archlinux.org/index.php/MacBookPro_Retina

frigaut · 2012-08-13 01:28:18

Update:
Benoit Gschwind has found the bug in the i915 graphic driver that corrupted the video signal.
See http://lists.freedesktop.org/archives/i … 19522.html for a patch.

Unfortunately, there is still a problem about the backlight not being turned on after suspend, so for me the i915 is still not usable for my daily uses. This might be related to gmux, which is going to be addressed soon I understand.

Related or not to gmux, the full advantage of using the intel graphic driver (which for me is to save power by being able to turn the nvidia graphic card off when I need it), will only be fulfilled when gmux is fixed.

frigaut · 2012-08-14 04:59:06

Another update:
Things are converging, thanks to the work of community members & developers.
bfroemel, gschwind and BlueDragonX, plus kernel and intel gfx developers solved the vast majority (if not all) issues with the graphic cards control.
No more display graphic corruption, one can now suspend/resume with the i915 driver (don't know about nouveau), switching graphic cards apparently works (haven't tried it), and switching on/off the nvidia works.
All the info at http://ubuntuforums.org/showthread.php? … 75&page=10 , and in particular http://ubuntuforums.org/showthread.php? … 75&page=10 for a summary of what to do in practice.

mzanetti · 2012-08-16 17:30:35

frigaut, thanks a lot for those updates!

I have seen you have updated the wiki page and added some informations about how to boot the kernel directly from EFI (without grub). I tried to follow it but I just can't get it to boot. I can make the kernel showing up by putting it on a separate HFS+ partition and blessing it, but it doesn't boot (The Apple boot menu just stays there frozen).

I had even less luck with copying the kernel to the existing EFI partition. I couldn't even get to show it up in the menu.

Could you please extend the instructions with some more details how it worked for you?

frigaut · 2012-08-16 23:11:01

I believe EFISTUB (direct EFI booting on a kernel without passing through a EFI bootloader and still be able to pass some boot parameter with a linux.conf file located in the same directory, see https://wiki.archlinux.org/index.php/UE … up_EFISTUB ) was supposed to land on 3.6 but is not currently fully functional (I could be wrong on that, there has been a lot of dev activity lately). So you need an EFI bootloader (I think the apple default bootloader is not going to show the linux kernels, and if it does it will not allow you to pass boot parameters to it, which may explain your hang). I'm using refind (a fork of refit), which works well. Instructions:

1) Put your kernel where it'll be found:

1.1) put your kernel and initramfs in your disk efi partition (sda1), somewhere under the EFI directory, alongside the APPLE directory, for instance in EFI/linux. Your kernel has to have the .efi extension (just add it to the regular vmlinuz -> vmlinuz.efi)

1.2) create or copy from the refind install (see below) the refind_linux.conf, e.g.:
"Boot with default" "root=/dev/sda4 nointremap add_efi_memmap"
"Boot intel powersave" "root=/dev/sda4 nointremap i915.modeset=1 i915.i915_enable_rc6=1 i915.i915_enable_fbc=1 i915.lvds_downclock=1 drm.vblankoffdelay=1 i915.i915_enable_ppgtt=0 add_efi_memmap"
[...]

2) Install the efi bootloader:

2.1) Download and install refind on your osx partition (simple, basically putting refind on a /efi partition on your MacHD, and then bless the refind...x64.efi file, see instructions on refind site). At reboot, refind will find your kernel. select it and press "F2" if you want to select another set of boot parameters in your refind_linux.conf list defined above.

what else?

* There are actually more detailled instructions at https://wiki.archlinux.org/index.php/UE … ing_rEFInd

* There may be a need to sync the GPT and MBR partitions, which I believe you can do within the refind menu (I haven't had to do it with this machine and refind version though)

* Do not forget to copy over the vmlinuz and initramfs after every upgrade of the kernel. This would make you boot with the old kernel, using the new /usr/lib/modules, and it may not end up well.

* you can have as many linux/arch directories as you want in the sda1 EFI/ partition. This way, you can test new stuff, and/or leave there a stable version you know you can come back to if you have problems with a new setup.

I believe that's all there is to it. Let me know if this doesn't work.
Soon, when EFISTUB works well, we should not need refind anymore.

litemotiv · 2012-08-17 09:41:56

This might be a tip for you guys: lately i've been mounting my EFI system partition as /boot, allowing pacman to directly place the kernel and initramfs in the ESP's root directory. When combined with the "scan_all_linux_kernels" parameter in refind.conf, this creates a seamless integration without having to manually copy/rename the files on kernel upgrades. This seems to work really well so far!

89c51 · 2012-08-17 18:19:21

frigaut wrote:

I believe that's all there is to it. Let me know if this doesn't work.
Soon, when EFISTUB works well, we should not need refind anymore.

Efistub with efibootmgr works well (at least on my non apple HW) but up until they implement support for linux.conf passing parameters to the kernel is a bit "cumbersome".

On topic. You might want to check this. http://mjg59.dreamwidth.org/15948.html

frigaut · 2012-08-18 12:25:11

@litemotiv

yes, there is also the trick with adding a hook to mkinitramfs. I like manual things for now though...

frigaut · 2012-09-18 00:23:44

3.6-rc6 is running smoothly for me (no additional patch).
vga_switcheroo enabled (although I'm using only the intel graphics), good power management (11-13W), bluetooth listed in rfkill (haven't tried it), basically everything that used to work with the heavily patched 3.6rc1 is now working with the vanilla kernel. Thanks and good job to all the devs that pushed all of these patches so effectively to linus.
Note: Microphone still not working.

mzanetti · 2012-09-18 08:59:20

Nice. Thanks for the update. I'm currently still running 3.6-rc5 with only the one needed gmux patch. I still have to use Mac OS's gfxCardStatus to force usage of the intel card. Is that one fixed too? I guess the nouveau drivers still need a lot of patching to get vga_switcheroo working smoothly, right?

frigaut · 2012-09-18 09:12:19

mzanetti wrote:

Nice. Thanks for the update. I'm currently still running 3.6-rc5 with only the one needed gmux patch.

yes, these ones are in now. http://git.kernel.org/?p=linux/kernel/g … a=shortlog look for gmux.

I still have to use Mac OS's gfxCardStatus to force usage of the intel card. Is that one fixed too? I guess the nouveau drivers still need a lot of patching to get vga_switcheroo working smoothly, right?

It seems that this is in too. Although I didn't try to un-force the intel card from osx, after some trials with vga_switcherro I ended up with that in my Xorg.log:

[ 403.669] (--) PCI: (0:0:2:0) 8086:0166:106b:00f7 rev 9, Mem @ 0xc1400000/4194304, 0xb0000000/268435456, I/O @ 0x00003000/64
[ 403.669] (--) PCI:*(0:1:0:0) 10de:0fd5:106b:00f2 rev 161, Mem @ 0xc0000000/16777216, 0x90000000/268435456, 0xa0000000/33554432, I/O @ 0x00002000/128, BIOS @ 0x????????/524288

which indicates to me that the nvidia is the one connected, while I am using the i915 driver (and thus using the card at 0:0:2).
Don't want to reboot right now, will update after testing, explicitly forcing the nvidia with gfxcardstatus.

frigaut · 2012-09-18 13:36:18

errm.
ok, so after a reboot into osx (which forces back the discrete card) and reboot into 3.6-rc6, the situation is the following:
* as I said, I am running an unpatched 3.6-rc6. To get vga_switcheroo working, I have to modeset i915 and nouveau. Nouveau takes control and I get a corrupted frame buffer (console). X doesn't seem to start (I have autologin and X autostarting), from what I can judge (can't read anything, but it looks like it stays on the console). Apparently, the nouveau patches that solve corruption have not made it into this kernel.
* I then tried the nvidia blob (actually using nvidia-bumblebee, which install the blob alongside the intel driver), but when X starts, I get a back screen (no backlight). I still have keyboard access and can reboot through a "sudo reboot". I tried a suspend/resume cycle, which has been said can restore
backlight control, but it doesn't restore backlight in this case.

So, mzanetti, not sure about the answer to your question. It seems that no, bypassing gfxcardstatus is still not possible.

mzanetti · 2012-09-18 13:55:39

Not a big deal as I'm fine right now with using the intel chip only.

Another question: I have installed refind through Mac OS by running ./install.sh. This works fine for the first reboot but everytime I boot Mac OS it gets thrown away again. That means, everytime I boot Mac OS I have to re-install refind and force the integrated graphics to be able to boot back to Linux. Do you see this issue too?

As I boot Mac OS very rarely, both are not big issues to me. But the refind situation is somewhat weird and I'd like to understand whats going on.

frigaut · 2012-09-18 14:32:36

nope, I don't see that. I actually don't have a install.sh in my refind install (you're sure it's refind and not refit?).
I use some bless command, but as I don't have access to the osx command line history right now, I'm not sure of the exact syntax.
Just to be sure: you've installed refind on your osx volume, right? Not the EFI volume? I have it under /efi in my osx volume/partition.

mzanetti · 2012-09-18 15:15:46

Ah, I think I have found it. Reading through the install.sh script (which definately came with the refind package) I see that they check for a file /Library/StartupItems/refitBlesser and print a big fat warning that refind does only start up once if this file is present (yes, I have installed refit some time back for testing). The reason why this warning was not triggered here, is that they check for the file with "test -f" and refit seems to have changed its structure to move this file into a subdirectory with thie same name.

Joekickass · 2012-10-13 10:25:14

Hi guys,

I currently have a working setup of Arch linux on the rMBP using the nvidia proprietary drivers (with the exception of the crappy broadcom wireless and a semi-working sd-card reader). However, I would like to try out the intel driver, which conflicts with the nvidia blob. My question is if the intel and nouveau drivers are working as of this writing, or will I cripple my system?

BR
Tomas

frigaut · 2012-10-14 06:53:38

I am currently running the 3.6.1 kernel from testing, and the intel graphic works fine with no tweaks.
* boot parameters: "root=/dev/something add_efi_memmap initrd=your_initrd" (although I hear the add_efi_memmap is not necessary). I am using refind as a efi bootloader, but there are other options.
* Booting efi direct as seen above.
* Don't forget to un-blacklist nouveau if you have blacklisted it (vgaswitcheroo won't work if any of the i915 or nouveau driver are prevented to load or to modeset)
* You have to have the intel graphic card selected from osx using gfxcardstatus (select "integrated only")
* I have this as 20-intel.conf in /etc/X11/xorg.conf.d/ :

Section "Device"
	Identifier 	"Intel Graphics"
	Driver 		"intel"
	Option		"AccelMethod" "sna"
	BusID 		"PCI:0:2:0"
EndSection

not sure it's necessary either, but it can't hurt.
* you can use vgaswitcheroo to turn off safely the nvidia graphic card (or turn it back on when you need to):

echo OFF > /sys/kernel/debug/vgaswitcheroo/switch

I remember I had to do something for /sys/kernel/debug to appear on the filesystem, but can't remember what right now (google for it if needed).
Good luck. Remember I'm not taking responsibility if you wreck your system and spend hours restoring it. But that should be safe. The worst that can happen is that X doesn't start, but you'll still have console access.

Joekickass · 2012-10-14 18:51:30

Sound good. I'll try it tonight. I've noticed your previous boot params for powersave:

frigaut wrote:

1.2) create or copy from the refind install (see below) the refind_linux.conf, e.g.:
"Boot with default" "root=/dev/sda4 nointremap add_efi_memmap"
"Boot intel powersave" "root=/dev/sda4 nointremap i915.modeset=1 i915.i915_enable_rc6=1 i915.i915_enable_fbc=1 i915.lvds_downclock=1 drm.vblankoffdelay=1 i915.i915_enable_ppgtt=0 add_efi_memmap"
[...]

Are they still relevant?

On another matter. I'm also using refind, with the refind.conf located on my mac partition. I use it to triple boot mac os, windows and arch linux. I've noticed the resolution of refind changed when I restarted the computer with a low-resolution projector connected. Now I cannot reset this resolution. I've tried different settings in refind.conf, e.g.:

resolution 1440 900

Seems to be controlled by MacOS display manager, but I cannot find any documentation regarding this. Any ideas?

Joekickass · 2012-10-15 13:07:00

Successfully installed xf86-video-intel and xf86-video-nouveau yesterday. Only got two real issues .

First off, I couldn't start X with the /etc/X11/xorg.conf.d/20-intel.conf as you suggested. Logs printed:

X.Org X Server 1.13.0
Release Date: 2012-09-05
[     6.951] X Protocol Version 11, Revision 0
[     6.951] Build Operating System: Linux 3.6.0-1-ARCH x86_64 
[     6.951] Current Operating System: Linux toni-arch 3.6.2-1-ARCH #1 SMP PREEMPT Fri Oct 12 23:58:58 CEST 2012 x86_64
[     6.951] Kernel command line: vmlinuz-linux.efi root=/dev/sda5 noapic nointremap nomodeset initrd=\initramfs-linux.img 
[     6.951] Build Date: 05 October 2012  01:57:18PM
[     6.951]  
[     6.951] Current version of pixman: 0.26.2
[     6.951] 	Before reporting problems, check [url]http://wiki.x.org[/url]
	to make sure that you have the latest version.
[     6.951] Markers: (--) probed, (**) from config file, (==) default setting,
	(++) from command line, (!!) notice, (II) informational,
	(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[     6.951] (==) Log file: "/var/log/Xorg.2.log", Time: Mon Oct 15 10:36:24 2012
[     6.951] (==) Using config directory: "/etc/X11/xorg.conf.d"
[     6.952] (==) No Layout section.  Using the first Screen section.
[     6.952] (==) No screen section available. Using defaults.
[     6.952] (**) |-->Screen "Default Screen Section" (0)
[     6.952] (**) |   |-->Monitor "<default monitor>"
[     6.952] (==) No device specified for screen "Default Screen Section".
	Using the first device section listed.
[     6.952] (**) |   |-->Device "Intel Graphics"
[     6.952] (==) No monitor specified for screen "Default Screen Section".
	Using a default monitor configuration.
[     6.952] (==) Automatically adding devices
[     6.952] (==) Automatically enabling devices
[     6.952] (==) Automatically adding GPU devices
[     6.952] (WW) The directory "/usr/share/fonts/OTF/" does not exist.
[     6.952] 	Entry deleted from font path.
[     6.952] (WW) `fonts.dir' not found (or not valid) in "/usr/share/fonts/100dpi/".
[     6.952] 	Entry deleted from font path.
[     6.952] 	(Run 'mkfontdir' on "/usr/share/fonts/100dpi/").
[     6.952] (WW) `fonts.dir' not found (or not valid) in "/usr/share/fonts/75dpi/".
[     6.952] 	Entry deleted from font path.
[     6.952] 	(Run 'mkfontdir' on "/usr/share/fonts/75dpi/").
[     6.952] (==) FontPath set to:
	/usr/share/fonts/misc/,
	/usr/share/fonts/TTF/,
	/usr/share/fonts/Type1/
[     6.952] (==) ModulePath set to "/usr/lib/xorg/modules"
[     6.952] (II) The server relies on udev to provide the list of input devices.
	If no devices become available, reconfigure udev or disable AutoAddDevices.
[     6.952] (II) Loader magic: 0x7fcc20
[     6.952] (II) Module ABI versions:
[     6.952] 	X.Org ANSI C Emulation: 0.4
[     6.952] 	X.Org Video Driver: 13.0
[     6.952] 	X.Org XInput driver : 18.0
[     6.952] 	X.Org Server Extension : 7.0
[     6.954] (--) PCI: (0:0:2:0) 8086:0166:106b:00f7 rev 9, Mem @ 0xc1400000/4194304, 0xb0000000/268435456, I/O @ 0x00003000/64
[     6.954] (--) PCI:*(0:1:0:0) 10de:0fd5:106b:00f2 rev 161, Mem @ 0xc0000000/16777216, 0x90000000/268435456, 0xa0000000/33554432, I/O @ 0x00002000/128, BIOS @ 0x????????/524288
[     6.955] (WW) Open ACPI failed (/var/run/acpid.socket) (No such file or directory)
[     6.955] Initializing built-in extension Generic Event Extension
[     6.955] Initializing built-in extension SHAPE
[     6.955] Initializing built-in extension MIT-SHM
[     6.955] Initializing built-in extension XInputExtension
[     6.955] Initializing built-in extension XTEST
[     6.955] Initializing built-in extension BIG-REQUESTS
[     6.955] Initializing built-in extension SYNC
[     6.955] Initializing built-in extension XKEYBOARD
[     6.955] Initializing built-in extension XC-MISC
[     6.955] Initializing built-in extension SECURITY
[     6.955] Initializing built-in extension XINERAMA
[     6.955] Initializing built-in extension XFIXES
[     6.955] Initializing built-in extension RENDER
[     6.955] Initializing built-in extension RANDR
[     6.955] Initializing built-in extension COMPOSITE
[     6.955] Initializing built-in extension DAMAGE
[     6.955] Initializing built-in extension MIT-SCREEN-SAVER
[     6.955] Initializing built-in extension DOUBLE-BUFFER
[     6.955] Initializing built-in extension RECORD
[     6.955] Initializing built-in extension DPMS
[     6.955] Initializing built-in extension X-Resource
[     6.955] Initializing built-in extension XVideo
[     6.955] Initializing built-in extension XVideo-MotionCompensation
[     6.955] Initializing built-in extension XFree86-VidModeExtension
[     6.955] Initializing built-in extension XFree86-DGA
[     6.955] Initializing built-in extension XFree86-DRI
[     6.955] Initializing built-in extension DRI2
[     6.955] (II) LoadModule: "glx"
[     6.955] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
[     6.955] (II) Module glx: vendor="X.Org Foundation"
[     6.955] 	compiled for 1.13.0, module version = 1.0.0
[     6.955] 	ABI class: X.Org Server Extension, version 7.0
[     6.955] (==) AIGLX enabled
[     6.955] Loading extension GLX
[     6.955] (II) LoadModule: "intel"
[     6.955] (II) Loading /usr/lib/xorg/modules/drivers/intel_drv.so
[     6.956] (II) Module intel: vendor="X.Org Foundation"
[     6.956] 	compiled for 1.13.0, module version = 2.20.9
[     6.956] 	Module class: X.Org Video Driver
[     6.956] 	ABI class: X.Org Video Driver, version 13.0
[     6.956] (II) intel: Driver for Intel Integrated Graphics Chipsets: i810,
	i810-dc100, i810e, i815, i830M, 845G, 854, 852GM/855GM, 865G, 915G,
	E7221 (i915), 915GM, 945G, 945GM, 945GME, Pineview GM, Pineview G,
	965G, G35, 965Q, 946GZ, 965GM, 965GME/GLE, G33, Q35, Q33, GM45,
	4 Series, G45/G43, Q45/Q43, G41, B43, B43, Clarkdale, Arrandale,
	Sandybridge Desktop (GT1), Sandybridge Desktop (GT2),
	Sandybridge Desktop (GT2+), Sandybridge Mobile (GT1),
	Sandybridge Mobile (GT2), Sandybridge Mobile (GT2+),
	Sandybridge Server, Ivybridge Mobile (GT1), Ivybridge Mobile (GT2),
	Ivybridge Desktop (GT1), Ivybridge Desktop (GT2), Ivybridge Server,
	Ivybridge Server (GT2), Haswell Desktop (GT1), Haswell Desktop (GT2),
	Haswell Desktop (GT2+), Haswell Mobile (GT1), Haswell Mobile (GT2),
	Haswell Mobile (GT2+), Haswell Server (GT1), Haswell Server (GT2),
	Haswell Server (GT2+), Haswell SDV Desktop (GT1),
	Haswell SDV Desktop (GT2), Haswell SDV Desktop (GT2+),
	Haswell SDV Mobile (GT1), Haswell SDV Mobile (GT2),
	Haswell SDV Mobile (GT2+), Haswell SDV Server (GT1),
	Haswell SDV Server (GT2), Haswell SDV Server (GT2+),
	Haswell ULT Desktop (GT1), Haswell ULT Desktop (GT2),
	Haswell ULT Desktop (GT2+), Haswell ULT Mobile (GT1),
	Haswell ULT Mobile (GT2), Haswell ULT Mobile (GT2+),
	Haswell ULT Server (GT1), Haswell ULT Server (GT2),
	Haswell ULT Server (GT2+), Haswell CRW Desktop (GT1),
	Haswell CRW Desktop (GT2), Haswell CRW Desktop (GT2+),
	Haswell CRW Mobile (GT1), Haswell CRW Mobile (GT2),
	Haswell CRW Mobile (GT2+), Haswell CRW Server (GT1),
	Haswell CRW Server (GT2), Haswell CRW Server (GT2+),
	ValleyView PO board
[     6.957] (++) using VT number 8

[     6.961] (EE) No devices detected.
[     6.961] 
Fatal server error:
[     6.961] no screens found
[     6.961] (EE) 
Please consult the The X.Org Foundation support 
	 at [url]http://wiki.x.org[/url]
 for help. 
[     6.961] (EE) Please also check the log file at "/var/log/Xorg.2.log" for additional information.
[     6.961] (EE)

Solved by removing /etc/X11/xorg.conf.d/20-intel.conf

The next, more serious issue, is that gnome-shell is consuming around 50% of all CPU threads (~400% CPU in top) all the time. I did:

* Installed xf86-video-intel and xf86-video-nouveau, removed nvidia blob.
* Switched gfxcardstatus to "integrated only"
* Boot params in refind (root=/dev/blahblah add_efi_memmap")

I will investigate this further. Thanks for the help by the way...

Last edited by Joekickass (2012-10-15 16:30:17)

litemotiv · 2012-10-15 13:33:17

Joekickass, did you mean to use code tags instead of quote tags?

Arch Linux

#26 2012-08-01 14:25:16

Re: MacBook Pro 2012 Retina

#27 2012-08-01 14:48:32

Re: MacBook Pro 2012 Retina

#28 2012-08-01 15:00:59

Re: MacBook Pro 2012 Retina

#29 2012-08-02 13:35:27

Re: MacBook Pro 2012 Retina

#30 2012-08-03 08:20:46

Re: MacBook Pro 2012 Retina

#31 2012-08-03 15:20:58

Re: MacBook Pro 2012 Retina

#32 2012-08-13 01:28:18

Re: MacBook Pro 2012 Retina

#33 2012-08-14 04:59:06

Re: MacBook Pro 2012 Retina

#34 2012-08-16 17:30:35

Re: MacBook Pro 2012 Retina

#35 2012-08-16 23:11:01

Re: MacBook Pro 2012 Retina

#36 2012-08-17 09:41:56

Re: MacBook Pro 2012 Retina

#37 2012-08-17 18:19:21

Re: MacBook Pro 2012 Retina

#38 2012-08-18 12:25:11

Re: MacBook Pro 2012 Retina

#39 2012-09-18 00:23:44

Re: MacBook Pro 2012 Retina

#40 2012-09-18 08:59:20

Re: MacBook Pro 2012 Retina

#41 2012-09-18 09:12:19

Re: MacBook Pro 2012 Retina

#42 2012-09-18 13:36:18

Re: MacBook Pro 2012 Retina

#43 2012-09-18 13:55:39

Re: MacBook Pro 2012 Retina

#44 2012-09-18 14:32:36

Re: MacBook Pro 2012 Retina

#45 2012-09-18 15:15:46

Re: MacBook Pro 2012 Retina

#46 2012-10-13 10:25:14

Re: MacBook Pro 2012 Retina

#47 2012-10-14 06:53:38

Re: MacBook Pro 2012 Retina

#48 2012-10-14 18:51:30

Re: MacBook Pro 2012 Retina

#49 2012-10-15 13:07:00

Re: MacBook Pro 2012 Retina

#50 2012-10-15 13:33:17

Re: MacBook Pro 2012 Retina

Board footer