Qoopz AnV xnu 10.4.0 Rel4
 

I have recently begun rewriting the patcher (again), it should now work.
I've also modified the segment loading code (custom for non-encrypted on AMD, stock routines on Intel)
The dyld which got built in is a stock Apple dyld with sysenter trap pre-patched.
It should now work on AMD for all apps (including translate under 64bit).
The patched dyld is also separately included under the patches as dyld_patched for those who which to use it with other kernels (install then under /usr/lib/dyld).
This should work for most Qoopz + AnV kernels because it patches sysenter to int 0xFC.

Download including sources:
http://rapidshare.com/files/41366145..._Rel4.zip.html

thanks very much for your efforts
testing hope it works

I am unable to boot using that one giving me kind of page-fault kernel panic...

http://img3.imagebanana.com/img/nd8usno8/kp.jpg

after testing
busratio not detected --> so must use busratio=16
64 bit restart---------> so must use arch=i386
using both flag it work and give panic like above :'(
hope that solved in next version :)

same issue as above, :)

It's not working under 64-32bit if I don't use patcher_opts=2, same kp as the one above. Once it boots in 32bit everything seems fine.

^this

Oh and to clear up, do we have to use -force64 or arch=x86_64 on amd with this kernel?

Well, I still can't play any steam games,and i have problems with some logic plugins aswell, I couldn't with the original dyld either anyway but this is definitely a big upgrade.
Well i guess they are decrypted, thats why they don't work?
Sorry for my limited understanding of this anyway.

Oh, starcraft 2 works now with the "patched" dyld, so I'm almost ready to ditch my old iDeneb 1.5.8 install:)

update:

-force64 doesn't help much for me, most apps crash unless they are 64bit.

So here comes my first testing experience with the Rel4 Kernel: :)

+ no more System-Profiler crash like with Rel3 Kernel
+ my custom display calibration setting gets loaded

without "patcher_opts=2"

- Kernel Panic from above

using "patcher_opts=2"

- Spotlight Prefpane crashes the Finder (Finder doesn't restart)
- iTunes crashes with the exact same error as if I am using -force64 (see screenshot below)
- Volume control in menubar doesn't work due to unpresent decryption routine
+ Quiktime works

using "patcher_opts=2 -force64"

+ booting into 32/64-mode with success
+ Quiktime works
+ Spotlight Prefpane works
- iTunes crashes
- sometimes a white block appears when closing a window (just close - no quit of the corresponding app)
+ Volume control in menubar works without patched/decrypted SystemUIServer - so if the issues with forced patcher (mainly iTunes - maybe other Apps causing similar error??) would get fixed, I might gonna become a happy user of your Kernel even without working decryption routine… :D

Screens of working 64-Bit & left issues:

http://img3.imagebanana.com/img/f2jd...cher_opts2.png http://img3.imagebanana.com/img/2d1b...r_opts2dis.png

So far still too buggy / unstable for using it, but another step into the right direction. :)

Thanks again for your work Andy - I am looking forward to try out further builds of this project. ;)

//// EDIT \\\\
Most Apps (Firefox, VLC, etc) are crashing the same way as iTunes does - same with and without -force64.
I tried booting the old Rel3 Kernel with "patcher_opts=2" and I am having the exact same issues...

Re: - Qoopz AnV xnu 10.4.0 Rel4
 

Hi Andy,

Thanks a lot for this fast release I'm really pleased with the progress and would like to share my thoughts....

Congrats to Andy I don't know how to thank you?

I can confirm the same as above. As soon as I changed kernels I got a KP (page fault) exactly like the one above. Then I added patcher_opt=2 to get into legacy mode. when I use the -force64 flag I get reboot. so i changed arch=86_64 to arch=i386 got into 64bit mode.

in 64bit reaper64 works but firefox crashes here are the details : -

as above safari etc 64bit apps work perfect but 32bit seems to not load

Regards,

Ashish

Same thing here, on my AMD Phenom II based pc : it doesn't boot without "patcher_opts=2" and in x64 mode (forced with -force64) I cannot launch x86 applications.

I'm releasing a patch that will also patch the libSystem sysenter traps (pre-patched).
That should clear some of these issues...
I'll continue the work on the kernel ;)

good work andy hope new one will be the best kernel for AMD :D

With -force64 (and syscenter fix) it now loads 32bit and 64bit apps, :) nice work!

Thanks Andy. I can now run both 32bit and 64bit apps in my AMD PhenomII X6 1090T using AnV_Qoopz_xnu_10.4.0_Rel4.zip, AMD_sysenter_10.6.4_patcher.zip, latest Chameleon2RC5 (rev411) and booting with

arch=i386 busratio=16 cpus=6 maxmem=4096 patcher_opts=2 -force64

For some reason the combination

arch=x86_64 -force64

does not work (instant reboot) and I have not tested yet if 8GB RAM is possible.

OK, I've applied your patch, and now I can run x86 applications while in x64 mode ( so I can also use more than 4 GB of RAM :) ).

http://img52.imageshack.us/img52/227/screenfg.th.png

The weird thing is that the kernel doesn't identify itself as x64 if you boot with "-force64" (eg. in System Profiler), but that seems to be only cosmetic stuff.

Thanks for your work, Andy :)

Thanks again for the quick fix, there are some improvements over the last that I've identified, for instance i can now use Logic with the 32bit audio unit bridge in 64bit mode, that is very nice! however some plugins still crashes the audiounit validation and i still can't play games in steam, this is not an issue on 10.5.8.

http://img831.imageshack.us/img831/9...00821at840.png

My boot flags now are:

arch=i386 -force64 cpus=3 patcher_opts=2

I have to use cpus=3 because if i don't i get glitchy audio with my presonus firebox fw interface, this was not an issue until i did a clean install a few weeks ago when i had a HD failure, but this is irrelevant to this fix never the less.

This fix patches the sytem dyld, not the one thats injected in the kernel right, should i use std_dyld=Yes ?

Oh, and about the steam issue, am i the only one experiencing this?

First, big thanks to Andy for his great work on this, it is really amazing!

I could boot a fresh (unpatched) 10.6.4 install (10.6.0 + combo update) with this kernel on an AMD Athlon64 3200+ and it mostly works (even iTunes started) but there are some small problems:
- booting in 64bit (without arch=i386 or -x32 options) results in instant reboot (this is not a big problem for me as I have less than 4G RAM)
- booting without patcher_opts=2 results in kernel panic as others noted
- when it boots I saw some crashes and freezes randomly but generally it's very promissing.

I was trying to make power management working (OK, maybe that's too much for a start but I'd like my system run cooler). While installing the VoodooPower package I got a kernel panic during the finishing kextcache -update-volume / step (screenshot attached).

After resetting I got another kernel panic, looks like VoodooPower does not work for my system? (another screenshot attached).

I also have a strange problem with missing dates and times almost everywhere in the GUI (dates show right in the Terminal) and some graphics artifacts in some buttons. The video card is an Nvidia gForce 7600 GS, QE/CI works with Chameleon's GraphicsEnabler option after fixed _UID, One->Zero in DSDT (Gigabyte MB with nForce4 chipset). Third screenshot attached, circles show missing dates, arrows show graphics artifacts. It may have to do with the bootloader because Chameleon also seems to ignore the com.apple.Boot.plist in /Extra folder (I have to type boot options, but it reads DSDT.aml and smbios.plist from the same place). I have Chameleon v2.0-RC4_PCEFI-10.5_AsereBLN. Any ideas anyone about any of these?

@bltz : I think they've dropped support for AMD processors in VoodooPower some time ago. And even when they did support AMD it wasn't always working good. If you want to boot in 64-bit mode use "-force64" flag (without "arch=...").

@cynargo: Thanks for the reply. I think the VoodooPower I was trying should have some support for AMD based on this:
The -force64 option does the same (instant reboot while booting) like any other way trying to boot 64bit on my system (but as I said it's not important to me). Sadly it panics randomly even in 32bit so I think I have to wait for some improved version.

Thank you Andy for providing the new kernel and patches.
I Installed the kernel + AMD sysenter trap patch and added the following kernel flags to the boot string:
arch=i386 patcher_opts=2 -force64

Everything now seems to be working fine. Although I noticed I am still getting cs_invalid_page errors. Here are some of them:

CODE SIGNING: cs_invalid_page(0x1000): p=47[loginwindow] clearing CS_VALID
CODE SIGNING: cs_invalid_page(0x1000): p=46[mds] clearing CS_VALID
CODE SIGNING: cs_invalid_page(0x1000): p=88[fontd] clearing CS_VALID
CODE SIGNING: cs_invalid_page(0x1000): p=110[Dock] clearing CS_VALID
CODE SIGNING: cs_invalid_page(0x1000): p=111[SystemUIServer] clearing CS_VALID
CODE SIGNING: cs_invalid_page(0x1000): p=112[Finder] clearing CS_VALID
CODE SIGNING: cs_invalid_page(0x1000): p=115[fontd] clearing CS_VALID

Is there a way to get rid of them?

Thank you!
FX

Front Row doesn't work
I have patched

/System/Library/Frameworks/CoreServices.framework/Versions/A/Frameworks/CarbonCore.framework/Versions/A
/System/Library/Quicktime

Also Marvin's AMD utility doesn't work too and iTunes stops itself the reproduction.

Sorry for my bad english
Anyway, thx!

Thank you..but patching is not working
 

Only patcher_opts=2 works and all others like patcher_opts=18 or patcher_opts=34 or the no patcher_opts gives me a KP saying:

EDIT: forgot to add that my processor is AMD Athlon X2 4400+

After a few days everything is running as it should, (except for Marvin, crashlog below) although I've noticed that the system seems a bit slow sometimes. Another thing I've noticed is that some soft, like pacifist, won't open with qoopz kernel but it runs fine with yours.
Thanks again for this Andy.

agree, system is slower than using nawcom kernel, every benchmark i have tested confirm that, and graphics ar glitchy also, that was not the case using just 32bit system... dont know where is the problem.. but my experience is that

I noticed with -force64 its rather slow myself but without that flag its as efficient as it should be, anyhow this is the first revision of a kernel/dyld that can use 64bit binaries simultaneously as 32 bit on amd so i guess it can take a bit time until its as reliable as it is on a intel cpu:)

Also, my terminal gets bombarded with:
dyld: shared cached file was build against a different libSystem.dylib, ignoring cache

not a big deal though

I guess I should do some benchmarks myself to confirm the slow downs, working with anything coreaudio with -force64 at the moment is just too glitchy.

Well, I've done some benchmarks and I got the oppposite result as you guys, my system gets better scores in 64bit, and quite better as you can see. As for the slowdown problem, maybe it's something else, like the graphics, don't know what I can do to find the issue.

@Rayflower: try rebuilding the prebindings, I got that problem the first time and that solved it. I always use mainmenu for that purpose but there are many tools for that.



http://img832.imageshack.us/img832/4...reencap.th.jpg

ok here my benchmark, as i sed, there is differences, and quit a bit,
booted without -force64 flag, and my god. how everything was smooth (was on 64-32 by 2 or 3 days) it was so obvious that, that with -force64 system was on some lag on performance and especially on graphics, you can feel that on naked eye.. (google earth was just lagging like crazy) almost every app was taking longer time to boot, (but i have to admit, pictures files and pdf's was opened damn quick much quicker than on 32bit system) strange.. anyways maybe better patch coming soon :) lets hope so.

32bit
http://i33.tinypic.com/1zu6tk.jpg

64bit
http://i33.tinypic.com/2nv6wyp.jpg

Thanks eMatos, that did the trick!

Also for my benchmarking:
http://img840.imageshack.us/img840/2015/92920607.png

Turns out the tests done there gives -force64 the better hand but my results there are marginal.

An on other news i noticed some new(to me) errors in some software i have:

http://img101.imageshack.us/img101/1888/aics5.png

and

http://img824.imageshack.us/img824/9525/pscs3.png

Mine seems to be normal:

32-bit Geekbench on 32-bit Snow:
http://i33.tinypic.com/jf94dw.png

32-bit Geekbench on 64-bit Snow:
http://i36.tinypic.com/2061oa9.png

64-bit Geekbench on 64-bit Snow:
http://i37.tinypic.com/m8e99t.png

Ignore these, the kernel has a patch integrated which forces CS_VALID after detecting any com.apple.xxx code signature...

They don't have any backtraces...
It's too bad as I can't find what is causing the issue this way (damn, too bad).
It's also too bad I don't have an AMD around so I could test it myself.
I'll continue the work though ;)

I'm seeing a nice boost in 64bit mode. In 32bit mode I got 3230 in Geekbench. In 64bit mode I got 3755. In Windows 7 64bit, however, I got 3154 which is a bit surprising since that's the OS that's actually suppose to run on this computer.

I still get "TSC: Frequency = 0.0000MHz" for my Pentium D 2.8Ghz (with busratio set in boo plist), but otherwise everything seems to work.

how strange that almost everyone except me got better running system with 32+64... im jump back to 32bit only, and damn how much faster it is.... :'(

I tried again with a fresh install of 10.6.0 with this 10.4.0 Rel4 kernel, Chameleon RC5 prerelease 18 Build_112 bootloader, fixed DSDT and smbios, nForceATA, nForceLAN, NullCPUPowerManagement and FakeSMC kexts on a machine with AMD Athlon64 3200+, nForce4 chipset. It boots and plays the Welcome movie but after the Welcome screen appears it gives me the attached kernel panic. Does anyone know what this means? (Actually it's the same kernel panic I've reported already in post#17 but now it always happens after boot on a clean install.) Some messages from kernel.log that may be interesting:

Update: I've narrowed it down to the GraphicsEnabler bootloader option. It does not panic if I don't use it (but then QE/CI does not work either). I could get through the initial configuration without GraphicsEnabler but I have the same graphics problems I have asked about previously and booting with GraphicsEnabler=y gets this KP. The same system works well with 10.5.6 with efistring but I cannot use that in 10.6 because Chameleon 2 ignores my com.apple.Boot.plist. (Maybe it's confused by two HDDs with 3 HFS+ partitions on them? Does anyone know how to get debug messages from Chameleon?)

use bootflag: patcher_opts=2

I have a small problem with some .mov and .mp4 files. The QuickLookHelper is quitting.

Of course I'm using patcher_opts=2 (and also arch=i386) otherwise it would not boot at all. With these options it boots, but get a KP after the welcome movie. I've also tried with legacy_kernel-10.4.0 but it also panics at the same place.

QuickLookHelper is known to be problematic even on real Macs. Try searching for it and see for yourself. So this might not be related to this kernel but could be.

Is there anyway i could make it a bit more verbose? i looked trough all my logs and I didn't find any more info about the crashes there I'm afraid.
Also I should add, the cs3 photoshop crash occurred when i tried to "save as.." while the crash in Illustrator cs5 occurred when I opened the app, I'll try to see if i can find more info about it.

Edit:

While opening PSCS3 in the terminal it got:

And with ILCS5:

The crash reporter adobe uses also leaves no backtrace so I guess that doesn't help much.

I have a problem where WindowServer, Safari, QuickTime, and other applications crash randomly. Before, I used to have crashes due to the sysenter trap in libSystem.b.dylib (divide by zero error) but after patching those libraries and using this kernel with patcher_opts=2, I now have this (the only difference is that there is no backtrace):

These crashes, like the sysenter ones, always happen a few hours after booting up. Again, this happens in all applications.

It seems like some widgets won't run now:
Only widgets made by apple will work, the ones I've installed won't.


New console messages that seems related to patch/kernel don't know if it helps but this is the only error I could get, besides some crashes without backtrace:

http://img199.imageshack.us/img199/9185/capjg.th.jpg
http://img23.imageshack.us/img23/4770/cap2do.jpg

Need any help, Andy? By the way, when I try to boot into 64-bit mode, it just restarts.

Works quite well so far on Phenom II X6 1055T, many thanks!

However I can't use your patched 1.2.3 version of VoodooPower.kext - it keeps saying it's unloading. I've tried removing Disabler.kext with AppleIntelCPUPowerManagement.kext but Disabler.kext always keeps coming back after a cache clean + reboot, even though it is not present in the /System/Library/Extensions folder. Is the kernel registering it on the fly?

I just replace the mach_kernel,it unable to boot

Try booting with the following boot arguments:

arch=i386 patcher_opts=2

That should work just fine...

Sep 12 16:01:07 davisin66s-Mac-Pro kernel[0]: CODE SIGNING: cs_invalid_page(0x3c9000): p=407[Front Row] clearing CS_VALID

And

12-09-10 15:50:12 com.apple.launchd.peruser.501[123] (com.apple.RemoteUI) The following job tried to hijack the service "com.apple.RemoteUI" from this job: 0x10bfe0.anonymous.Front Row

Then

12-09-10 15:50:12 FrontRow[212] Message Handler Installation Failed: bootstrap_check_in(): 0x44c: Bootstrap not privileged

12-09-10 15:50:12 FrontRow[212] ******** Application Startup ********

12-09-10 15:50:12 kernel CODE SIGNING: cs_invalid_page(0x3c9000): p=212[Front Row] clearing CS_VALID

12-09-10 15:50:13 login[208] DEAD_PROCESS: 208 ttys000

12-09-10 15:50:15 FrontRow[213] ******** Application Startup ********
12-09-10 15:50:15 com.apple.launchd.peruser.501[123] (com.apple.RemoteUI[213]) Exited: Killed

12-09-10 15:50:15 com.apple.launchd.peruser.501[123] (com.apple.RemoteUI) Throttling respawn: Will start in 10 seconds

12-09-10 15:50:15 kernel CODE SIGNING: cs_invalid_page(0x3c9000): p=213[Front Row] clearing CS_VALID

12-09-10 15:50:25 FrontRow[215] ******** Application Startup ********

12-09-10 15:50:25 kernel CODE SIGNING: cs_invalid_page(0x3c9000): p=215[Front Row] clearing CS_VALID
And continues...

Maybe because of that Front Row doesnt works and I have fixed QuickLook
http://archivos2.tie.cl/motumbo/quic...davisin666.zip

Replace QuickLook (/System/Library/Frameworks/QuickLook.framework/Versions/A/Resources)

Here it's the original file
http://archivos2.tie.cl/motumbo/quicklookd_original.zip

Thanks to Andy by making the kernel xD

@davisin666: I've noticed a while ago many crashreports from quicklook when it tried to preview some movie files. Now I'm running your modified quicklook and it's working just fine, thanks a lot!
After the quicklook replacement I've tested FrontRow and it seems to work as expected, though I haven't tested it much. Just so you know, I've never patched/cpuid frontrow app cause it always crashes if I do so.

EDIT: Would you mind patching dashboardclient.app too? I can't run any non-apple widget and it used to work fine in 32bit mode. Do you have this problem?

Dear, when I try to compile the kernel using voodoobuild I get this error message.
What is missing in my system ?
Thank you.


[3] Processing cctools (cctools-750) (1 min) 23:09:30
Extracting (cctools-750.tar.gz)...
Building...
../lto.c:272: error: ‘LTO_SYMBOL_DEFINITION_WEAK’ undeclared (first use in this function)
../lto.c:276: error: ‘LTO_SYMBOL_DEFINITION_TENTATIVE’ undeclared (first use in this function)
../lto.c:282: error: ‘LTO_SYMBOL_DEFINITION_UNDEFINED’ undeclared (first use in this function)
../lto.c:287: error: ‘LTO_SYMBOL_PERMISSIONS_MASK’ undeclared (first use in this function)
../lto.c:288: error: ‘LTO_SYMBOL_PERMISSIONS_CODE’ undeclared (first use in this function)
../lto.c:292: error: ‘LTO_SYMBOL_PERMISSIONS_DATA’ undeclared (first use in this function)
../lto.c:296: error: ‘LTO_SYMBOL_PERMISSIONS_RODATA’ undeclared (first use in this function)
make[2]: *** [lto.o] Error 1
make[1]: *** [static] Error 2
make: *** [all] Error 1
ERROR

You first need to build the latest LLVM library from:
http://llvm.org
and build that for i386 and x86_64 like this:
After that, the building of cctools will go a lot better ;)