InfiniteMac OSx86  


Reply
 
Thread tools Display modes
  #31  
Old 05-29-2012, 03:23 PM
sulphide sulphide is offline
 
Join Date: Dec 2009
Posts: 37
fx-8150, 64-bit, reboot right after kexts loaded. previous kernel works ok even in 64 bits.
Reply With Quote
  #32  
Old 05-29-2012, 05:41 PM
R:A:W:X86's Avatar
R:A:W:X86 R:A:W:X86 is offline
 
Join Date: Mar 2012
Location: Germany
Posts: 95
Thanks for testing so far.
Still working on some more improvements to the code, which I hope will make this more attractive for other people to work on this.

Quote:
Originally Posted by gils View Post
excuses moi de mon Français , pour te dire que ton kernel fonctionne sur mon Xéon E3 1230 en 32 bit et avec le bug finder (je suppose que c'est dû au mode 32 bit ) , demain , je testerai sur mon x4 960t .
[...]
je reste toujours en 32 bit .
This behavior is absolutely unexpected, unless you are using -legacy bootflag.

The Kernel still enters 64-Bit enabled boot-up by default.
-force64 is not associated with any function and not required on SSSE3-AMD - even less on Vanilla-Intel.

Just to note:
Using this Kernel permanently on a Vanilla capable system is not recommended.
The Kernel source code Apple releases doesn't 100% match the Kernel, they are releasing with their OS.
Beeing compatible is just a feature for stuff like multiplatform Boot-CD solutions, etc.

AMD FX 4100
MSI 760GM-P23
Radeon HD6570
OS X Lion Retail 10.7.5
Lion Kernel Project
Reply With Quote
  #33  
Old 05-29-2012, 08:06 PM
gils's Avatar
gils gils is offline
Panther
 
Join Date: Dec 2011
Posts: 194
Quote:
Originally Posted by R:A:W:X86 View Post
Thanks for testing so far.
Still working on some more improvements to the code, which I hope will make this more attractive for other people to work on this.



This behavior is absolutely unexpected, unless you are using -legacy bootflag.

The Kernel still enters 64-Bit enabled boot-up by default.
-force64 is not associated with any function and not required on SSSE3-AMD - even less on Vanilla-Intel.

Just to note:
Using this Kernel permanently on a Vanilla capable system is not recommended.
The Kernel source code Apple releases doesn't 100% match the Kernel, they are releasing with their OS.
Beeing compatible is just a feature for stuff like multiplatform Boot-CD solutions, etc.

hello,

Actually, I had mistakenly left-legacy in the plist (hoops!) For my tests, I using the same OSX 10.7.4 on my AMD Phenom and my Intel xeon .

That said, your kernel is royal! perfs are on the rise and smoother video! Hats off Sir!

There are only 64 bit mode (k 10) will fix the problem of various bug and we will have AMD configuration at the top!

The Phenom / Atlon still remains beautiful and deserve our respect!

I hope to that you will come to us out the kernel in 64 bit AMD K10.

Big thank you again for your work.

google translation.










je précise , le proc est bien unlocker en x6 à 3,01 ghz mais pas à 3,41 ghz comme affiché dans les tests , je suppose que c'est le turbo .

Macmini early 2006 /blanGsak 10.7.2
Acer 8930 G / core2duo T6500 / NV 9600 GS m / Lion 10.7.3 / ML 10.8.2 / w7/8
Asus M5A97pro / FX 8350 / Asus HD 7950 CUII v2
GA-78LMT SP2 / FX 6100 / P x4 960t / Sapphire HD 4850 1024
MSI K9neo v1 / Atlon 4800+/ Gigabyte HD 5450 512
test kernel 12.0 AMD

Tuto : http://www.hack-my-mac.fr/index.php/...fx-buldo#28896

CG opérationnelle : HD 5450 512 / HD 4850 1 / GT 610

Last edited by gils; 05-30-2012 at 09:49 PM.
Reply With Quote
  #34  
Old 05-29-2012, 10:25 PM
rrtn's Avatar
rrtn rrtn is offline
Cheetah
 
Join Date: Jul 2008
Posts: 4
Quote:
Originally Posted by R:A:W:X86 View Post
update :

I have changed the CPU identification methods with those written by mercurysquad & AnV.

The aim is to make the Kernel work on most AMD and Intel CPU that at least support SSE3 (including Vanilla Kernel compatible).

At the moment this build has to be considered unstable / for testing purposes.


looking for testers

It works on both of my workstations (FX-4100 and Xeon E3)

Anyway I would be glad if a lot of people would do some testing on this.

Particulary I would be interested in having it tested on all kind of non-SSSE3-capable AMD & Intel-Nehalem CPUs


Thank You R:A:W:X86 !
i installed 10.7.3 using bronzovka_mach_kernel_v4 since then just left it alone til next benchmark. so i see you posted a kernel for 10.7.4 so i upgraded to 10.7.4 tried your kernel and got reboot after system files load. so i used the 'roll back trick' i used either bronzovka_mach_kernel_v4 or v6 and replaced the kexts AppleFileSystemDriver.kext & IOStorageFamily.kext with the ones from 10.7.3. however, before i could post my results i see your new kernel and give it a shot and voilà lion loads! so heres my 2¢ on your kernel...great work so far
thanks for all your hard work R:A:W:X86

my specs:
ASUS A8N-SLI Deluxe
AMD Athlon 64X2 4400+ Toledo
4GB Patriot DDR 400 (PC 3200)
eVGA GeForce 8800GTS 640Mb 320-bit GDDR3
500GB IDE WD Caviar Blue (OSx86 Snow Leopard)
120GB SATA3 Corsair Force 3 SSD (Windows 7)
1TB SATA2 WD Caviar Black
1TB SATA2 WD Caviar Black
40GB SATA WD Caviar (OSx86 Lion)
NEC 16X Double Layer DVD±RW IDE
D-Link DWL-AG530 IEEE 802.11a/b/g PCI Tri-Mode
ASPIRE X-Superalien ATXA6SW-BK/500W
Reply With Quote
  #35  
Old 05-31-2012, 12:14 PM
R:A:W:X86's Avatar
R:A:W:X86 R:A:W:X86 is offline
 
Join Date: Mar 2012
Location: Germany
Posts: 95
update 2012-05-31:


The main part of the code changes that once have been gathered by AnV are now implemented in a working Kernel.

Most likely non-SSSE3 capable AMD systems will still freeze after hard disk initalisation, unless you are using the -legacy bootlag. (altough I could not test this)
We still need to find a fix for this.

atm NOT implemeted functions are:
  • auto patcher function for mach-o executables
  • some detection code in the real time clock handler (should not be so important as we are all using decent bootloaders)
In addition to the Kernel diff the Download contains two additional diff-patches that will add those code changes.
Both will compile fine, but cause malfunction during boot up.
I am going to diff those files between xnu-1504 & xnu-1699, maybe it is fixable by some "Snow Leopard reimplementations"...


Some answers - before you ask the corresponding questions:
  • Most likely an integration of the auto-patcher will not fix 64-Bit boot-up for the non-SSSE3 CPU users. (Actually I am not 100% sure what the origin of the problem is & I do not hold a hardware platforms that allows testing on this.)
  • Why I am working on this anyway: I hope this will solve a lot of application crashes and other bugs & add some stability to the -legacy mode.
  • Some people (including Bulldozer users) may need to advice the correct clock multiplier to the Kernel as auto-detection fails. Examples: busratio=13 (multiplier 13) or busratio=105 (multiplier 10.5)


this build still has to be considered an experimental release, made for testing purposes & to allow further development!



As already mentioned above, source patch is included.
Would be great if some people with more competencies would contribute to this.

AMD FX 4100
MSI 760GM-P23
Radeon HD6570
OS X Lion Retail 10.7.5
Lion Kernel Project

Last edited by R:A:W:X86; 06-03-2012 at 10:16 PM. Reason: corrected / added some information
Reply With Quote
  #36  
Old 05-31-2012, 03:38 PM
misterfrista misterfrista is offline
Puma
 
Join Date: May 2012
Posts: 21
R:A:W:X86 first I want to thank you for your great work!!

I tested your first kernel and my fresh 10.7.4 installation works great with flags -v npci=0x2000 -legacy!

But your newest one from today just restarts my pc over and over again, I tried the following flags in many combinations: -force64 -legacy arch=x86_64/i386 -v -f npci=0x2000 nothings helps!

If i can help you anyway tell me


My System:
- Bulldozer 8120
- Asus M5A97
- Geforce 9500GT
- 10.7.4 from Appstore with Myhack and your kernel a clean install without any problem
Reply With Quote
  #37  
Old 05-31-2012, 05:52 PM
sulphide sulphide is offline
 
Join Date: Dec 2009
Posts: 37
Quote:
Originally Posted by misterfrista View Post
I tested your first kernel and my fresh 10.7.4 installation works great with flags -v npci=0x2000 -legacy!

But your newest one from today just restarts my pc over and over again, I tried the following flags in many combinations: -force64 -legacy arch=x86_64/i386 -v -f npci=0x2000 nothings helps!

My System:
- Bulldozer 8120
- Asus M5A97
- Geforce 9500GT
- 10.7.4 from Appstore with Myhack and your kernel a clean install without any problem
Same for me, fx-8150, m5a97 (same), geforce 560ti. Restart after all kexts. This is without -legacy flag, of course.

Last edited by sulphide; 05-31-2012 at 06:09 PM.
Reply With Quote
  #38  
Old 05-31-2012, 09:22 PM
gils's Avatar
gils gils is offline
Panther
 
Join Date: Dec 2011
Posts: 194
Quote:
Originally Posted by R:A:W:X86 View Post
update 2012-05-31:


The main part of the code changes that once have been gathered by AnV are now implemented in a working Kernel.

Most likely non-SSSE3 capable AMD systems will still freeze after hard disk initalisation, unless you are using the -legacy bootlag. (altough I could not test this)
We still need to find a fix for this.

atm NOT implemeted functions are:
  • auto patcher function for mach-o executables
  • some detection code in the real time clock handler (should not be so important as we are all using decent bootloaders)
In addition to the Kernel diff the Download contains two additional diff-patches that will add those code changes.
Both will compile fine, but cause malfunction during boot up.
I am going to diff those files between xnu-1504 & xnu-1699, maybe it is fixable by some "Snow Leopard reimplementations"...


Some answers - before you ask the corresponding questions:
  • If the on-the-fly opcode patching routines can be fixed, they will not affect sysenter crashes on 64-Bit enabled AMD (Bulldozer) systems. Currently it will handle mach-o executables only, no shared librarys and frameworks
  • Most likely it will also not fix 64-Bit boot-up for the others. (Actually I am not sure what the problem is & I do not hold a hardware platforms that allows testing on this.)
  • Why I am working on this anyway: I hope this will solve a lot of application crashes and other bugs & add some stability to the -legacy mode.


this build still has to be considered an experimental release, made for testing purposes & to allow further development!



As already mentioned above, source patch is included.
Would be great if some people with more competencies would contribute to this.

hello,

I tested your new kernel (raw), big step forward, starting the PC is a little space! I no longer crash application, I FaceTime which is very long to open, save it for QuickTime bug screen,
I have a CPU that bubbles!

Code: raw -legacy GraphicsEnabler=No PCIRootUID=1-v-f

the rest is in the plist.

thank you for your job!









Macmini early 2006 /blanGsak 10.7.2
Acer 8930 G / core2duo T6500 / NV 9600 GS m / Lion 10.7.3 / ML 10.8.2 / w7/8
Asus M5A97pro / FX 8350 / Asus HD 7950 CUII v2
GA-78LMT SP2 / FX 6100 / P x4 960t / Sapphire HD 4850 1024
MSI K9neo v1 / Atlon 4800+/ Gigabyte HD 5450 512
test kernel 12.0 AMD

Tuto : http://www.hack-my-mac.fr/index.php/...fx-buldo#28896

CG opérationnelle : HD 5450 512 / HD 4850 1 / GT 610
Reply With Quote
  #39  
Old 06-02-2012, 03:18 AM
ham4ever ham4ever is offline
Jaguar
 
Join Date: Dec 2011
Posts: 37
i tested your both kernels , and both great in this specs:

AMD Athlon II x3 435





second kernel open apps soo fast and lion so smooth

just so far i noticed i cant see app in activity monitor like in pic




Update : after using 1 hour, apps appears in Activity monitor

Last edited by ham4ever; 06-02-2012 at 04:12 AM.
Reply With Quote
  #40  
Old 06-03-2012, 01:12 PM
R:A:W:X86's Avatar
R:A:W:X86 R:A:W:X86 is offline
 
Join Date: Mar 2012
Location: Germany
Posts: 95
Quote:
Originally Posted by sulphide View Post
Same for me, fx-8150, m5a97 (same), geforce 560ti. Restart after all kexts. This is without -legacy flag, of course.
Some people (including Bulldozer users) may need to advice the correct clock multiplier to the Kernel as auto-detection fails. Examples: busratio=13 (multiplier 13) or busratio=105 (multiplier 10.5)

update 2012-06-03


As the Kernel features some patch made by AnV that automaticaly turns code signature to valid, I was able to fix 32-Bit Apps to work on 64-Bit enabled AMD machines (Bulldozer).

This requires sysenter traps to be patched in /usr/lib/dyld and /usr/lib/system/libsystem_kernel.dylib
Maybe some additional shared libraries need to be patched if XCode is installed (untested).

There is an Lion compatible Version of amd_insn_patcher by AnV that works for Lion. Old versions will not work as the sysenter trap has changed in Lion.
Get it from here:
http://osx86.co/f36/legacy-kernel-10...post55947.html

Usage: amd_insn_patcher unpatched_file patched_file

Be sure to fix file permissions before you copy over & use "sudo update_dyld_shared_cache" afterwards.



So far all my Apps 32- & 64-Bit are working without issues.



💡 Deploy cloud instances seamlessly on DigitalOcean. Free credits ($100) for InfMac readers.


AMD FX 4100
MSI 760GM-P23
Radeon HD6570
OS X Lion Retail 10.7.5
Lion Kernel Project

Last edited by R:A:W:X86; 06-03-2012 at 01:16 PM.
Reply With Quote
Reply