Jump to content

Mavericks kernel testing on AMD (formerly Mountain Lion kernel testing on AMD)

Mountain Lion AMD legacy kernel x64_86 ssse3 ssse3 emulator

  • Please log in to reply
6379 replies to this topic

#1841
theconnactic

theconnactic

    Stubborn AMD user

  • Local Moderators
  • 2,973 posts
  • Gender:Male
What is this?, they will ask: simple - this is HUGE NEWS!

Congratulations, Sinetek!

#1842
theconnactic

theconnactic

    Stubborn AMD user

  • Local Moderators
  • 2,973 posts
  • Gender:Male

UPDATE 8: an unexpected annoyance, ArchiveUtility seems to stop working after the system is some time up, just like good old Lion XPCHelper-related core service crashes. I have StuffitExpander, so i could choose not to bother, but i think this issue should be investigated, since it could be just another manifestation of the same issue that cause, among other things, the nVidia graphics corruption.

Best regards!

EDIT: here's the entire report for the crash:

Attached File  report.zip   36.44KB   5 downloads

#1843
Sinetek

Sinetek

    InsanelyMac Sage

  • Coders
  • 333 posts
  • Gender:Male

UPDATE 8: an unexpected annoyance, ArchiveUtility seems to stop working after the system is some time up, just like good old Lion XPCHelper-related core service crashes. I have StuffitExpander, so i could choose not to bother, but i think this issue should be investigated, since it could be just another manifestation of the same issue that cause, among other things, the nVidia graphics corruption.

Best regards!

EDIT: here's the entire report for the crash:
Attached File  report.zip   36.44KB   5 downloads

wtf, that's similar to my bug above. something with 'semaphores'/'mutex' in libsystem_kernel.dylib
and/or libdispatch.dylib

@connactic: could you get me the SAME report but with this kernel argument set at boot??
keepsyms=1


Edit: AppleFSCompressionTypeDataless is rearing its ugly head

your log is strange, I can't find a clear reason for the crash in it

EDIT #3.1415: NVDA(DMA): Channel exception! exception type = 0x1f = Fifo: MMU Error
00000069

#1844
Shaneee

Shaneee

    AMD Dev Tester

  • Retired
  • 1,002 posts
  • Gender:Male
  • Location:Scotland
  • Interests:Web Design. OS X. Music.

What is this you ask?

This is Mountain Lion running i386 on AMD with an nVidia card and QE/CI. There is a huge bug somewhere though which is triggered by clicking on any menu.

Sinetek


Could you share your test kernel? I am going to re create my Login loop on ML and try it out. May have different results with different CPU / GPU combo.

#1845
theconnactic

theconnactic

    Stubborn AMD user

  • Local Moderators
  • 2,973 posts
  • Gender:Male
NVDA(DMA): Channel exception! exception type = 0x1f = Fifo: MMU Error 00000069 ---> Well, well, well!

#1846
spakk

spakk

    If you try to please everyone, then you have certainly forgotten

  • Local Moderators
  • 2,026 posts
  • Gender:Male
  • Location:marocain
it may be if the sound works via a working AppleHDA.kext that will be work then the graphics card with CI / QE?. I've adjusted my DSDT and the AppleHDA.kext, but the sound does not work yet hundred percent ! The graphics are much better but still no CI / QE. Can my presumption apply???
I have a Nvidia GTX 550 TI with Realtek ALC889 Onboard Sound

#1847
Adrian Andréas

Adrian Andréas

    InsanelyMac Protégé

  • Members
  • Pip
  • 32 posts

it may be if the sound works via a working AppleHDA.kext that will be work then the graphics card with CI / QE?. I've adjusted my DSDT and the AppleHDA.kext, but the sound does not work yet hundred percent ! The graphics are much better but still no CI / QE. Can my presumption apply???
I have a Nvidia GTX 550 TI with Realtek ALC889 Onboard Sound


I was thinking that I only had problems with audio.
In my case, the audio stutters and the codec it is:
VIA VT1708S

Spakk you ever tried this?
http://olarila.com/f....php?f=3&t=2168

#1848
theconnactic

theconnactic

    Stubborn AMD user

  • Local Moderators
  • 2,973 posts
  • Gender:Male

I was thinking that I only had problems with audio.
In my case, the audio stutters and the codec it is:
VIA VT1708S

Spakk you ever tried this?
http://olarila.com/f....php?f=3&t=2168


At least you have AppleHDA audio, guys! I have to resort to VoodooHDA, and HDMI is yet a distant hope.

Now, to the news:

I can confirm the latest amd_kernel (which is attached in the first post of the topic) works with no performance loss and no errors with 10.8.3. Contrary to all common sense, i did the App Store update (but i recommend you use the combo update) and the only modifications i had to make was adding my card's ID in the plists of both ATI5000Controller.kext and AMDRadeonAccelerator.kext. I did this to the 10.8.3 stock kexts and all went fine, but i saved my 10.8.2 patched kexts just in case. That's what i recommend to you, fellow ATI users who have a stable system.

Best regards!

#1849
burkett375

burkett375

    InsanelyMac Protégé

  • Members
  • PipPip
  • 80 posts

At least you have AppleHDA audio, guys! I have to resort to VoodooHDA, and HDMI is yet a distant hope.

Now, to the news:

I can confirm the latest amd_kernel (which is attached in the first post of the topic) works with no performance loss and no errors with 10.8.3. Contrary to all common sense, i did the App Store update (but i recommend you use the combo update) and the only modifications i had to make was adding my card's ID in the plists of both ATI5000Controller.kext and AMDRadeonAccelerator.kext. I did this to the 10.8.3 stock kexts and all went fine, but i saved my 10.8.2 patched kexts just in case. That's what i recommend to you, fellow ATI users who have a stable system.

Best regards!


are you still seeing graphics glitches? That's the only thing that's keeping me back from using it as a daily driver. and how do you patch the kexts, cause I just tried and I don't think I did it right.

Thanks!

#1850
theconnactic

theconnactic

    Stubborn AMD user

  • Local Moderators
  • 2,973 posts
  • Gender:Male
Hi!

Heavy graphic glitches with Safari, so i am using Chrome, which is absolutely fine. Other than that, no glitches at all, and only minor annoyances, like having to reinstall VLC every time i want to use it (but install VLC means copy it to the Applications folder, so there's no big deal). Games, movies and animations work just fine with absolutely no graphics corruption. Least but not last, i got my higher Geekbench2 score ever with this machine.

I am already using Mountain Lion as my primary system here. But i confess that, since is a HTPC, a computer for fun and content consumption, not for work, i cannot lightly recommend this for everyone.

The method i used for my graphics card will not necessarily work for yours. That said, i simply added the ID of my card to the info.plists of ATI5000Controller.kext, under IOPCIMatch, and also in AMDRadeonAccelerator, under cedar (the family of my card). I can boot with GE=Y or GE=N, but GE=N makes the system report my card as an AMD5000, not an AMD5450. With this method, only the default framebuffer, Eulemur, appears to work: all others will boot to a black screen.

All the best!

#1851
theconnactic

theconnactic

    Stubborn AMD user

  • Local Moderators
  • 2,973 posts
  • Gender:Male
10.8.3: extensive tests

Attached File  Screen Shot 2013-03-15 at 1.35.16 AM.png   314.84KB   34 downloads

Attached File  Screen Shot 2013-03-15 at 2.24.35 AM.png   765.38KB   26 downloads

Attached File  Screen Shot 2013-03-15 at 2.28.46 AM.png   35.01KB   16 downloads

Attached File  Screen Shot 2013-03-15 at 2.26.07 AM.png   182.65KB   15 downloads

Attached File  Screen Shot 2013-03-15 at 2.30.05 AM.png   246.3KB   11 downloads

Attached File  Screen Shot 2013-03-15 at 2.30.21 AM.png   222.29KB   11 downloads

Attached File  Screen Shot 2013-03-15 at 2.30.55 AM.png   15.04KB   16 downloads

Attached File  Screen Shot 2013-03-15 at 2.31.41 AM.png   239.51KB   18 downloads

Attached File  Screen Shot 2013-03-15 at 2.32.59 AM.png   177.65KB   17 downloads

Attached File  Screen Shot 2013-03-15 at 2.33.21 AM.png   252.8KB   19 downloads

Attached File  Screen Shot 2013-03-15 at 2.34.56 AM.png   155.2KB   18 downloads

Attached File  Screen Shot 2013-03-15 at 2.35.06 AM.png   161.66KB   17 downloads

Attached File  Screen Shot 2013-03-15 at 2.35.42 AM.png   239.54KB   16 downloads

Attached File  Screen Shot 2013-03-15 at 2.36.22 AM.png   253.37KB   16 downloads

P.S.: All the programs of the Adobe CS6 suite were open at the same time. They took a while to load, but once open, all they worked without any glitch or lack of comparative performance against 10.7.5 or Windows 8.

10.8.3, the upsides: the archive utility bug (see the top of the page, where Sinetek answers to the post i describe it) went away. The amd_kernel (first post of the topic) works fine. No loss of performance (see topic Geekbench Scores).

10.8.3 the downsides: Safari remais unusable with the HD5450. VLC still cannot be reopened once closed, needing a reinstallation. The graphics are a bit more buggy: "phantoms" (persistent pixels) are left when a window is closed, but they can be erased by selecting the region affected, or shaking another window over it. Launchpad showed glitches once, but i wasn't able to reproduce it, so it should be an unfrequent glitch.

All in all, i recommend the update, as long your ATI graphics card is already working with 10.8.2.

P.S.: i might as well try to rollback the ATI kexts to the 10.8.2 ones to improve graphics stability. I'll test it, as soon as i back up my system.

#1852
rd8111

rd8111

    InsanelyMac Protégé

  • Members
  • Pip
  • 28 posts
  • Gender:Male
Hi Guys,

Let me pass on how I made the rarely known FirePro v4900 integrate and run with Snow Leopard on my FX-8350, Gigabyte GA-990FXA-UD5. I think it will be useful regardless of which OS X version. I used a couple different tools between lspci and System Info to make sure the Device ID (0x6749) and Vendor ID (0x1002) for the v4900 were consistent. Then upon noticing a set of ATI kernel extensions in S/L/E, I reviewed all ATI info.plist content for device and vendor id matches. I found one exact match within the ATIRadeonX3000 kext. I used a web site of listed PCI data and the Chameleon 2187 ati.c source code file to find the graphics family and other device ids in the same family. It turns out that the v4900 is part of the Turks family. Other video card models have different graphic families. Most importantly though, once you find the graphics family for the card you have, you may find that existing kexts already have yours and/or many of the graphics family members already present.

When reviewing all of the info.plist content via Show Package Contents (pop up menu in Finder) and navigating via Finder to Contents | info.plist, I happened to find the controller file listing Turks family members (not the v4900 though). Thinking that only two kexts would be enough, I booted up and expected graphics to magically work without forcing Graphics Mode. The screen was all blue, and I was pretty sure the desktop was hiding somewhere. With some more research, I found a couple posts implying that three items are needed for successful ATI graphics:

ATISupport.kext
A Controller kext
A Video Card kext (not sure what else to call it)

Anyhow, since I had determined which ATIController kext already uses other Turks family members, I added the v4900 to the IOPCIMatch key within the ATI6000Controller.kext. I doubt the order has any significance, since the listing was not ascending or descending. I did place it around other device ID values closest to the v4900 though. Keep in mind that an entry is

0xDeviceIDVendorID

The v4900 looks like: 0x67491002

After saving this change to the info.plist, updating the kernel caches and permissions with Kext Wizard (much faster than typing in Terminal), I rebooted without modifying any Chameleon 2.0, RC5 boot flags. After the gray apple screen was done, the screen went dark for about five seconds, then lo and behold, full graphics support appeared. After getting this to work, Graphics Mode and GraphicsEnabler values are ignored (possibly overridden).

I have been on the AMD OS X scene for a short time, so there may be something I am not aware of in Mountain Lion and Lion. As long as the more recent OS X releases do something similar with maintaining support for ATI video cards, there is a realistic chance that this will work. I suppose the most recently released ATI cards may not be so easily supported. So, your mileage may vary...

It looks odd on paper, but the kexts that are running the v4900 correctly are:
ATISupport.kext
ATI6000Controller.kext (only added the v4900 to the IOPCIMatch value set)
ATIRadeonX3000.kext

I can change the resolution just like on my MacBook Pro, modify refresh rate, run the screen saver, run the Grab application, etc. The connection is DVI dual link between the v4900 and my ancient Acer AL1917W monitor. I made no modifications to the DSDT. In fact only the default DSDT and smbios settings from the Chameleon 2 RC5 boot loader are in use at this time. Now that I have found the magic three lines to have a bootable USB via a cloned disk image boot successfully, I will be testing the more recent boot loader versions and determining upgrading to Mountain Lion.

#1853
Bitcore

Bitcore

    AMD Fanboy

  • Members
  • PipPipPip
  • 150 posts
  • Gender:Male
  • Location:The Netherlands
  • Interests:Producing Music
Hi guys,

I'm back again.
I see you did great progress on video distortion problems!
Have to test the aty_something and rd8111's way of getting rid of it.
When i have a new HDD, i'll try to join the development again :)
Might get a cheap ass HD5450 aswel :)
LETS GET FULL QE/CI and no distorion FOR ALL ATI (and Nvidea) CARDS !!!

#1854
burkett375

burkett375

    InsanelyMac Protégé

  • Members
  • PipPip
  • 80 posts

Hi!

Heavy graphic glitches with Safari, so i am using Chrome, which is absolutely fine. Other than that, no glitches at all, and only minor annoyances, like having to reinstall VLC every time i want to use it (but install VLC means copy it to the Applications folder, so there's no big deal). Games, movies and animations work just fine with absolutely no graphics corruption. Least but not last, i got my higher Geekbench2 score ever with this machine.

I am already using Mountain Lion as my primary system here. But i confess that, since is a HTPC, a computer for fun and content consumption, not for work, i cannot lightly recommend this for everyone.

The method i used for my graphics card will not necessarily work for yours. That said, i simply added the ID of my card to the info.plists of ATI5000Controller.kext, under IOPCIMatch, and also in AMDRadeonAccelerator, under cedar (the family of my card). I can boot with GE=Y or GE=N, but GE=N makes the system report my card as an AMD5000, not an AMD5450. With this method, only the default framebuffer, Eulemur, appears to work: all others will boot to a black screen.

All the best!


So is it worth it to fiddle around with the kext if everything is working fine minus graphic artifacts? I get artifcats using Chrome or Safari, with both being unusable for some sites. I see that in the iopcimatch, there's several device IDs, do I delete them all except mine? Thanks for your help!

#1855
theconnactic

theconnactic

    Stubborn AMD user

  • Local Moderators
  • 2,973 posts
  • Gender:Male
I don't think so, burkett375, and i think the biggest merit of rd811's method is getting the FirePro v4900 to work with OSX, not solving the graphics corruption bug related to kernel itself (which the method didn't do: the card simply appears not to experience it). I think that, if you already have QE/CI, your work with the kexts is done, and you will have to look elsewhere to solve this very annoying graphics corruption issue.

Rd811, i'll ask the staff to move your post to the Genius Bar as a tutorial: i'm pretty sure lots of FirePro users, regardless of being on Intel or AMD, can benefit of it.

All the best!

#1856
rd8111

rd8111

    InsanelyMac Protégé

  • Members
  • Pip
  • 28 posts
  • Gender:Male

I don't think so, burkett375, and i think the biggest merit of rd811's method is getting the FirePro v4900 to work with OSX, not solving the graphics corruption bug related to kernel itself (which the method didn't do: the card simply appears not to experience it). I think that, if you already have QE/CI, your work with the kexts is done, and you will have to look elsewhere to solve this very annoying graphics corruption issue.

Rd811, i'll ask the staff to move your post to the Genius Bar as a tutorial: i'm pretty sure lots of FirePro users, regardless of being on Intel or AMD, can benefit of it.

All the best!


Thank you, theconnactic. I think the approach will work for more than FirePro users. I found so many ATI graphic families that my head was spinning...

@burkett375, I would never remove already existing IDs from a kext. For a future hardware change (e.g. the latest video card with 10 GB Video RAM coming soon to a computer near you) the removal of valid IDs may be forgotten 2 years from now. Adding a DeviceIDVendorID to a kext that has other items in the same device family seems better. I have followed directions for replacing the contents within the VoodooHDA kext though, but that is specific to the "how to solve problems" FAQ the VoodooHDA team has out there. Besides, as long as you have backups via your preferred utility program, it is much much easier to try something and restore to a previous point if needed.

#1857
theconnactic

theconnactic

    Stubborn AMD user

  • Local Moderators
  • 2,973 posts
  • Gender:Male
Indeed, rd811: it could be a tutorial for any ATI user: could you rewrite your post in the form of a tutorial, so i can ask the staff to move it to the Genius Bar?

#1858
Shaneee

Shaneee

    AMD Dev Tester

  • Retired
  • 1,002 posts
  • Gender:Male
  • Location:Scotland
  • Interests:Web Design. OS X. Music.
Apple seem to be helping us out I think. Just updated to 10.8.3 and with no kext modifications yet I've booted without safe mode to desktop with buggy graphics and no QE/CI on my GT 430. Will keep you updated...

#1859
Adrian Andréas

Adrian Andréas

    InsanelyMac Protégé

  • Members
  • Pip
  • 32 posts
I bought a USB sound card.
However has the same stuttering in the sound!
I believe it is kernel problem.
The same happens with Lion.Com Sow leopard works perfectly

#1860
rd8111

rd8111

    InsanelyMac Protégé

  • Members
  • Pip
  • 28 posts
  • Gender:Male

Indeed, rd811: it could be a tutorial for any ATI user: could you rewrite your post in the form of a tutorial, so i can ask the staff to move it to the Genius Bar?


Sure. Do I post it here? I will look at a couple other tutorials and follow the format.





Also tagged with one or more of these keywords: Mountain Lion, AMD, legacy kernel, x64_86, ssse3, ssse3 emulator


2 user(s) are reading this topic

0 members, 2 guests, 0 anonymous users

© 2014 InsanelyMac  |   News  |   Forum  |   Downloads  |   OSx86 Wiki  |   Mac Netbook  |   PHP hosting by CatN  |   Designed by Ed Gain  |   Logo by irfan  |   Privacy Policy