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

[Sinetek]

AMD doesn't support the 'leaf' {censored} in cpuid_set_cache_info(), so i bypassed that since it's not useful.

however, topo depends on the number of cores being set correctly. can't just patch the hex.

 

uint64_t msr = rdmsr64(MSR_CORE_THREAD_COUNT);

info_p->core_count   = bitfield32((uint32_t)msr, 19, 16);

info_p->thread_count = bitfield32((uint32_t)msr, 15,  0);

 

 

can patch these to a known value, say, 1

[Bronya]

 

AMD doesn't support the 'leaf' {censored} in cpuid_set_cache_info(), so i bypassed that since it's not useful.

however, topo depends on the number of cores being set correctly. can't just patch the hex.

 

uint64_t msr = rdmsr64(MSR_CORE_THREAD_COUNT);

info_p->core_count   = bitfield32((uint32_t)msr, 19, 16);

info_p->thread_count = bitfield32((uint32_t)msr, 15,  0);

 

 

can patch these to a known value, say, 1

 

Yes! In it is add  "(ecx & 0xff) + 1;" or number cpu ,say, 8 if cpu 8 cores )))))

 

Tomorrow i continue fix kernel ... 

[Sinetek]

or just plain

 

 

info_p->core_count   = 1;
info_p->thread_count = 1;

and hope for the best

[theconnactic]

A probably naive question for you both: how can you guys visualize the C functions inside the hex files/binaries? No hex editor i know do this kind of "translation" properly. If there's a "translation key" for it, perhaps i can help you with the patches themselves, since i'm now quite fluent in C...

 

All the best!

P.S.: 10.8.5 is out, update your computers

[burkett375]

 

 

 

 

 

ok

any graphic glitches?

[theconnactic]

He has a 4xxx series ATI card, Burkett, so he won't see any glitches but get instead the red pixel bug (meaning games and 3-D rendering software are all unusable).

 

All the best!

[burkett375]

Yeah, thought about that after I posted

[tuxtard]

There are some differences in SSE instructions between Intel and AMD chips. Take a look at this:

http://en.wikipedia.org/wiki/SSE4

Maybe you guys could disable everything past SSE3 in kernel for testing purposes? If there are no glitches in that situation, then we will know where to look.

[spakk]

I do not know whether this is of significance:

 

 

*************************************************************AMDhttp://lkml.org/lkml/2005/11/4/173http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/26094.PDFhttp://developer.amd.com/article_print.jsp?id=92 (RH)[7] AMD's 7th  generation  processors return  a CPUID  base    family value of '7'. These include AMD Athlon, AthlonXP,    AthlonMP, and Duron.[8] AMD's  8th generation  processors  return an  effective    CPUID family of '0x0F'. These include AMD Opteron,     Athlon64, and Turion.before 7th gen : ok7th gen:  P-state (performance state) change    UP : warn about time inaccuracy    SMP      sol : disable powernow      Use monotonic pseudo-TSC  STPCLK-Throttling (temperature) : only done on UP, ok    UP : warn about time inaccuracy8th gen :  P-state change    UP : inaccuracy    dual-core : locked-step ; inaccuracy    SMP : may drift      sol : disable powernow    Use monotonic pseudo-TSC  SMP, dual core : C1-clock ramping (halt) (power state : C-state)    sol : idle=poll or disable C1-ramping    Use monotonic pseudo-TSC  STPCLK-Throttling (temperature) :    single processor dual-core ok ; inaccuracy    SMP : NOT ok (rare)      Use monotonic pseudo-TSCUntil TSC  becomes invariant, AMD  recommends that operatingsystem  developers  avoid TSC  as  a  fast  timer source  onaffected systems. (AMD  recommends that the operating systemshould  favor these  time sources  in a  prioritized manner:HPET first,  then ACPI PM  Timer, then PIT.)   The followingpseudo-code shows one way of determining when to use TSC: use_AMD_TSC() { // returns TRUE if ok to use TSC   if (CPUID.base_family http://www.intel.com/design/mobile/specupdt/302209.htm) ), yoursupposition is true:Article V. For Pentium M processors (family [06H], models [09H, 0DH]); forPentium 4 processors, Intel Xeon processors (family [0FH], models [00H, 01H, or02H]); and for P6 family processors: the timestamp counter increments with everyinternal processor clock cycle. The internal processor clock cycle is determinedby the current core-clock to bus-clock ratio. Intel(R) SpeedStep(R) technologytransitions may also impact the processor clock.Article VI. For Pentium 4 processors, Intel Xeon processors (family [0FH],models [03H and higher]): the time-stamp counter increments at a constant rate.That rate may be set by the maximum core-clock to bus-clock ratio of theprocessor or may be set by the frequency at which the processor is booted. Thespecific processor configuration determines the behavior. Constant TSC behaviorensures that the duration of each clock tick is uniform and supports the use ofthe TSC as a wall clock timer even if the processor core changes frequency. Thisis the architectural behavior moving forward.It's a pity they call this sucking TSC behavior an "architectural behaviormoving forward"*************************************************************HPEThttp://www.intel.com/hardwaredesign/hpetspec_1.pdf

 

[madlock]

EM64T is in Pentium4 6xx models, Pentium4 5x1 models and Celeron 3x1 and 3x6 models.

They are still quite common and widely used processors.

 

Many of us still have Pentium D series 64-bit processor, which has Intel EM64T too but only SSE3.

SSSE3 in Core2Duo has only 4 multimedia orders more. It would be really nice thing to give the

working legacy_kernel for this useful Pentium D processor.

 

Fortunately 10.6.8 works well with some 10.6.7 kexts and voodooHDA 2.7.4 , but working Lion

or Mountain Lion would be definitely nice.

This Kernel works perfekt on Pentium D with ML :-)

[Bronya]

Sinetek, i corrected your mach_kernel_10.9 and working cpu_number and on amd . 

But don't working "singleuser" . 

 

mach_10_9_sinetek_by_bronzovka_fix.zip

[iOmicron]

Has development on the ML kernel stopped now, to move onto Mavericks? So this would this be the latest (and final) kernel for 10.8? http://www.insanelymac.com/forum/topic/281450-mountain-lion-kernel-testing-on-amd-dont-ask-help-here-use-the-help-topic/page-142?do=findComment&comment=1933704

There was nowhere that linked to the latest RC so I had to search through every page...

[theconnactic]

block USB

What? It goes until the USB fail? Does it mount the root device?

[spakk]

all try were unsuccessful

 

 

Edit:

booting with Hack OS X boot CD the driver from S / L / E are loaded  and at the end it stucks on .... Patched DMI table.
should I make a picture and upload?

 

Edit:

 

I boot with flag:-nossse3 npci= 0x2000 fsb = 200 -x

 

 

Edited September 15, 2013 by spakk

[spakk]

 

Edit:

 

there is nothing matrimonial, it reboots when I start without Hack CD

[spakk]

any news?

all try were unsuccessful

 

 

Edit:

booting with Hack OS X boot CD the driver from S / L / E are loaded  and at the end it stucks on .... Patched DMI table.
should I make a picture and upload?

 

Edit:

 

I boot with flag:-nossse3 npci= 0x2000 fsb = 200 -x

what is the cause of:

• efi_inject_get_devprop_string NULL trying stringdata 
• Patched DMI table

       (see pic #3027 )

 

as a cause a problem with chameleon bootloader?

Edited September 15, 2013 by spakk

[Bronya]

spakk , try delete efi_inject_get_devprop_string and smbios.plist .

[spakk]

spakk , try delete efi_inject_get_devprop_string and smbios.plist .

bronzovka where I can find efi_inject_get_devprop_string, it is the efi-string in the chameleon.boot.plist? I have already been removed,

[Bronya]

bronzovka where I can find efi_inject_get_devprop_string, it is the efi-string in the chameleon.boot.plist? I have already been removed,

hmm, if mountain kernel working , but don't working maverick kernel - then problem kernel on your amd .

[spakk]

hmm, if mountain kernel working , but don't working maverick kernel - then problem kernel on your amd .

ok,i try your last mountain lion rc9_fix and report

 

Edit:

unfortunately, the system stops at the same place. no changes.

without this {censored}ing_hack CD but with the your last Mountain Lion rc9_fix_Kernel

I get a kernel panic, I'll post the same picture

ok,i try your last mountain lion rc9_fix and report

 

Edit:

unfortunately, the system stops at the same place. no changes.

without this {censored}ing_hack CD but with the your last Mountain Lion rc9_fix_Kernel

I get a kernel panic, I'll post the same picture

should I take a rollback of the acpi and ioacpifamily from mountain lion 10.8 ? what do u think?

after change the kexts from Mountain Lion 10.8  I get a similar kernel panic

[Bronya]

ok,i try your last mountain lion rc9_fix and report

 

Edit:

unfortunately, the system stops at the same place. no changes.

without this {censored}ing_hack CD but with the your last Mountain Lion rc9_fix_Kernel

I get a kernel panic, I'll post the same picture

 

should I take a rollback of the acpi and ioacpifamily from mountain lion 10.8 ? what do u think?

rc9_fix normal working . but panic - it extensions . kernel_10.9 don't working for you amd . i should fix , but best on FX . 

[spakk]

rc9_fix normal working . but panic - it extensions . kernel_10.9 don't working for you amd . i should fix , but best on FX . 

sorry but  I have a Phenom II X6

 

 

Edit 1:

Info:

is that booting with the smbios.plist then the System reboot !!! without smbios.plist the kexts are loaded and get a KP

 

Edit 2:

it does not matter which smbios i use , strange behavior ,,,???

 

Edit 3:

what experience have you so done ??  are there any test results ??  ( FYI, I'm using for this test, the first Maverick Preview 1 Version )

Edited September 15, 2013 by spakk

[Sinetek]

ML kernel is done for the most part.. Nvidia problem is something to do with Nvidia driver.

 

Mavericks is still borked on AMD for the moment. Bronya and I are looking into it. Me at snail's pace as I'm a bit busy.

[spakk]

ML kernel is done for the most part.. Nvidia problem is something to do with Nvidia driver.

 

Mavericks is still borked on AMD for the moment. Bronya and I are looking into it. Me at snail's pace as I'm a bit busy.

 I understand you correctly? should we stop the test for a few days or weeks? please answer my question, sinetek

[Bronya]

gils83, blocked only at "launchd" ! Usb working .