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

[Sinetek]

not really surprised that it's rebooting.. could someone test on FX?

[DJ PhoeniX]

I can write installer on USB and try to boot with your kernel. Which kernel file and parameters I need?

[spakk]

 

10.9 DP7 sinetek kernel 

http://speedy.sh/fGpKk/mv

song to go with it

http://www.youtube.com/watch?v=4Z6ZJQEGVGY

Hi Sinetek, wonderful to know that you've released the first AMD-test-kernel for maverick,  I will tonight when I'm at home, test it with my prepared external HDD  and will report my results. This is very good news, no matter what first test results are obtained, which the main thing that lives on the development.

10.9 DP7 sinetek kernel.zip

[AkimoA]

instant reboot on Phenom as well 

 

any flags to go with it `? i figured not since its a binpatch i guess

 

way to go Tx for your affords keep it up 

 

have a good day

[Sinetek]

Many thanks to all my testers, love you guys

 

Might be the MSR's that's tripping up AMD boards. Let's cross fingers. Anyone would like to help me out with my web illiteracy find some hosting spot?

[Shaneee]

Anyone would like to help me out with my web illiteracy find some hosting spot?

 

If you mean to host the kernel you could use dropbox or mediafire.

[Sinetek]

there are so many possible issues. let's try some fixes.

 

 

http://cl.ly/20320Z0G1910

 

 

 

[spakk]

• Sinetek do not worry about of the many kernel-testers, we are hotter than New Year's  2012/13 than it came to Mountain Lion AMD kernel

• the first test was without success, I will now test your second kernel

[theconnactic]

Just tested. Instant reboot, unfortunately. Keep'em coming!

 

All the best!

[spakk]

there are so many possible issues. let's try some fixes.

http://cl.ly/20320Z0G1910

 

second test: only with boot flag:-nossse3 gray screen (in the background load the kexts) shortly thereafter for a fraction of a second to see the gray apple, then restart

I have tried almost everything, unfortunately no success, no kernel panic message. the system re-starts only.

Edited September 12, 2013 by spakk

[bcobco]

what about hosting this   darwin kernel non-intel port   builds and sources in sites like googlecode, github, ... ?

we would have proper bug tracking

[joe75]

Apple already host sources

[Bronya]

omg , interesting . i here . i only that test it is kernel and fix  

1) i checked and answer :

 

no problem . i can fix it is ... 

[Sinetek]

you know the drill

 

http://cl.ly/2D033i0Z1E2C

[AkimoA]

same same 

[Sinetek]

@bronya   and now?

[Bronya]

 

you know the drill

http://cl.ly/2D033i0Z1E2C

It is divide error !

 

I know it is problem . it is number cpu in cpuid . I can fix ..... i need time ...

But i fix "no linesize" and normal . please wait . i need check "single user"  

[Sinetek]

then we're getting a bit closer. that means it's skipping over the Intel CPU cache code. It's weird though, the AMD folks should be able to see the divide by 0 panic

[Bronya]

then we're getting a bit closer. that means it's skipping over the Intel CPU cache code. It's weird though, the AMD folks should be able to see the divide by 0 panic

it amd no know pseudo code "cpuid" from "intel" .

[Sinetek]

@bronya   да))  no sources.. so difficult. it bombs in cpu_thread_alloc.

 

maybe AMD crashes sooner??? MSR's.......

[Bronya]

@bronya   да))  no sources.. so difficult. it bombs in cpu_thread_alloc.

 

maybe AMD crashes sooner??? MSR's.......

NO . it is cpuid.

 

 

it is cpu number in cupid_set_info . 

you see :

 

for  intel :

cpuid_fn(0, cpuid_result);

 

info_p->cpuid_cores_per_package 

= bitfield32(reg[eax], 31, 26) + 1;

 

for amd :

cpuid_fn(0x80000008, cpuid_result);

info_p->cpuid_cores_per_package 

= (ecx & 0xff) + 1;

 

I checked - don't working single user .

[Sinetek]

@bronya   weird, it's the first i hear of that.   CPUID function 0 is valid on AMD 

[Bronya]

@bronya   weird, it's the first i hear of that.   CPUID function 0 is valid on AMD 

amd don't support "cupid function" 0 for "cpu number" ))))  support only 0x80000008 )

 

sinetek, you can speak to me how used qemu with "terminal gdb" ? i try if have time ...

Edited September 12, 2013 by Bronya

[Sinetek]

ugh. it's a hard problem. CPU topology depends on checking number of cores. div by 0 error everywhere. need to fill in a structure, but how?

[Bronya]

ugh. it's a hard problem. CPU topology depends on checking number of cores. div by 0 error everywhere. need to fill in a structure, but how?

yes . replace hex code in cupid_set_info and add cpuid in tsc_init ))) . i it is trying ...