Project OS X Forums > My Bootloader

FormerlyKnownAs

Jan 20 2011, 05:35 PM

QUOTE (Slice @ Jan 19 2011, 08:41 PM)

There are dust remaining after my attemps to understand what is the best UUID, how to get it, and how to use it.
But in any case System Profiler how me other Hardware UUID. No matter what I did during boot.

I didn't check system profiler. Just reinstalled to test and System UUID vales from both methods show the same in bdmesg but I have different value in system profiler also.

D

Qortado

Jan 21 2011, 02:36 AM

QUOTE (Slice @ Jan 13 2011, 11:24 AM)

Here?

What about VoodooSDHC (1.1.2b1) ?

Slice

Jan 21 2011, 07:33 AM

QUOTE (Qortado @ Jan 21 2011, 06:36 AM)

What about VoodooSDHC (1.1.2b1) ?

Here

Qortado

Jan 21 2011, 12:08 PM

QUOTE (Slice @ Jan 21 2011, 05:33 AM)

Here

Thanks, I was looking under Hardware Information

Azimutz

Jan 21 2011, 04:14 PM

Hi Slice,
bad news

the booter from your Chameleon branch gives me instant reboot right after "boot0: done", any rev.
I don't have any clue of were the problem is coming from at this point!
All i can say is that i also can't boot Meklort's branch; the last rev i can boot from his branch, is 671.
But the problem with his branch is a bit different; i do get an instant reboot, but only after all modules are loaded. Still need to inform Mek about this...
Any idea?
The Assembla version works.. not perfect but i can boot without major problems.

Anyway, i'll be back later for more... got some stuff to talk with you about this post and the "kernel cache" stuff...

See you later.

Slice

Jan 21 2011, 06:43 PM

Did you place the booter into EFI? I didn't try and I think the problem may be here.
And more. You check my compilation or your own? May be different.

Azimutz

Jan 22 2011, 10:57 AM

No Slice, i only use EFI partition if i'm forced to, for testing purposes or so.
Going to try the booter from the first post... i've been compiling it my self.
Thanks for reminding me

Edit: same problem. Will try to check what's going on; if you were using trunk instead of Meklort's branch, it would be easier to check

Will be back.. still need to check some stuff about the platform uuid...

Slice

Jan 22 2011, 11:57 AM

I think the Meklort's module base bootloader give us more flexibility to check: you can exclude almost all modules from loading except Sysmbols.dylib to see what is happen.
Also you can compile some modules with DEBUG_XXX=1 to boot verbose.

Azimutz

Jan 27 2011, 10:04 PM

yeah... i understand what you mean. But, i'm having trouble trying to study "normal" Chameleon's code and also keep up with all the changes on Mek's branch (and all the others, for that matter). I have to take it slow... Don't get me wrong, i do like the modules system and he's certainly doing a great work, all on it's own. It's just that i know "normal" Chameleon a lot better... it would be easier for me to spot a problem if you were using trunk.

Anyway, back to your booter...
You forgot to add "modules_support.s" to boot2 folder on the last rev (714); it wont compile without it.
I still get the instant reboot like i explained before, even without modules; as far as i can see modules don't even start loading.
It's some other problem... i really don't know were to start looking...

Slice

Jan 28 2011, 08:09 AM

As well as Meklort I do not propose to return to trunk code. It is not updated since rev654 no matter how many bugs are found. I am looking forward and working on modules system. I reported bugs with these on voodooproject. New revision is not working and I am in searching why.

Kabyl

Jan 28 2011, 04:32 PM

QUOTE (Slice @ Jan 28 2011, 09:09 AM)

As well as Meklort I do not propose to return to trunk code. It is not updated since rev654 no matter how many bugs are found. I am looking forward and working on modules system. I reported bugs with these on voodooproject. New revision is not working and I am in searching why.

I'm not exactly sure what you're expecting, but it doesn't work that way; projects like this need human resources, time, effort, and will.

There is a difference between how helpful reporting bugs is and proposing a patch to fix a bug. I've come across many posts with a fix (more or less) to bug here or there, but it ends up forgotten and unnoticed most times because no proper applicable patch against trunk is provided.

Slice

Jan 28 2011, 09:46 PM

QUOTE (Kabyl @ Jan 28 2011, 08:32 PM)

I'm not exactly sure what you're expecting, but it doesn't work that way; projects like this need human resources, time, effort, and will.

There is a difference between how helpful reporting bugs is and proposing a patch to fix a bug. I've come across many posts with a fix (more or less) to bug here or there, but it ends up forgotten and unnoticed most times because no proper applicable patch against trunk is provided.

I do not know how to propose a diff that can be considered as "proper applicable patch" because mostly one corrections leads to another in several places. May be in future, when I got more or less final result I can make a diff file and commented every line why I change it.
Nobody ask me why my acpi_patcher as twice as longer then trunc one.
What I did with EDID.h
Why I comment out tens lines in smbios_pather and added others.
What is NVRAM module.
Why two GraphicsEnabler modules.
May be I am wrong in my patches but no any discussions about its.

OK. Let it be my way not needed for others.
But I also encounter problems with core functions of Chameleon that I never touched. I asked questions. I show reports about bugs. And... again, I see no movements.

Kabyl

Jan 29 2011, 05:58 AM

QUOTE (Slice @ Jan 28 2011, 10:46 PM)

I do not know how to propose a diff that can be considered as "proper applicable patch" because mostly one corrections leads to another in several places. May be in future, when I got more or less final result I can make a diff file and commented every line why I change it.
Nobody ask me why my acpi_patcher as twice as longer then trunc one.
What I did with EDID.h
Why I comment out tens lines in smbios_pather and added others.

I, personally, didn't notice any of that.

QUOTE

What is NVRAM module.

I'm watching this one, but as I can see, it's still not finished (?)

QUOTE

Why two GraphicsEnabler modules.

Well, it would be clear when you use modules of course, when everything is built into a single boot file, it's not.

The modules idea was there long ago, and has attracted many people, and we had a lot of discussions about it as well, but there were no results, until a single person (Evan) decided to do it all.

I don't see him complaining about the lack of interest or anything like that (this doesn't mean we're not interested, on the contrary), and he has yet to say his work is ready to be merged into trunk, meanwhile, everyone needs to catch up with the new structure of the booter.

QUOTE

May be I am wrong in my patches but no any discussions about its.

OK. Let it be my way not needed for others.
But I also encounter problems with core functions of Chameleon that I never touched. I asked questions. I show reports about bugs. And... again, I see no movements.

People working on this project have other, more important, things to do; real jobs, real life... so it's all about time and will.

Slice

Jan 29 2011, 11:35 AM

QUOTE (Kabyl @ Jan 29 2011, 09:58 AM)

I'm watching this one, but as I can see, it's still not finished (?)

Yes, it is not fully working cause I need to study more about this. Being alone I can't promise a solution soon.
I just have a result
Click to view attachment
while without NVRAM module
Click to view attachment
Further I propose to choose a disk, press restart, and Chameleon restarted into choosen partition. Is it real? Yes! But some problems is still unresolved.

QUOTE

Well, it would be clear when you use modules of course, when everything is built into a single boot file, it's not.

The modules idea was there long ago, and has attracted many people, and we had a lot of discussions about it as well, but there were no results, until a single person (Evan) decided to do it all.

I don't see him complaining about the lack of interest or anything like that (this doesn't mean we're not interested, on the contrary), and he has yet to say his work is ready to be merged into trunk, meanwhile, everyone needs to catch up with the new structure of the booter.

Before rev700 his module's system looks like working. But now some changes turn it to non-working state and now I work around it. Hope meklort too.

QUOTE

People working on this project have other, more important, things to do; real jobs, real life... so it's all about time and will.

I also have restricted time and I prefer to spend it to eliminate bugs then to explain to others why I did it. "Others" are not interesting.

valv

Jan 29 2011, 02:09 PM

QUOTE (Slice @ Jan 29 2011, 12:35 PM)

"Others" are not interesting.

?

Slice

Jan 29 2011, 02:52 PM

QUOTE (valv @ Jan 29 2011, 06:09 PM)

?

Are you ready to discuss a common bug?

valv

Jan 29 2011, 03:43 PM

always

Slice

Jan 29 2011, 04:23 PM

OK.
1. This question is still unresolved.
2. This question also is not too hard.
3. Why Linux is invisible?

CODE

bool EX2Probe (const void *buf)
{
return (OSReadLittleInt16(buf+0x438,0)==0xEF53); //
}

This is right magic. I checked.

valv

Jan 29 2011, 05:00 PM

I already knew about your two first questions. I did no progress so far. Did you ?
Regarding Linux detection, It is working for me. So I don't know where to start analysis from.

Slice

Jan 29 2011, 08:04 PM

QUOTE (valv @ Jan 29 2011, 09:00 PM)

I already knew about your two first questions. I did no progress so far. Did you ?

As I said the path is different. But if for Leo/Tiger there are no problem

CODE

Try cache /System/Library/Caches/com.apple.kernelcaches/kernelcache YES
Loading Darwin 10.4
Loading kernel cache /System/Library/Caches/com.apple.kernelcaches/kernelcache

For Snow I can't get good name

CODE

search for kernelcache /System/Library/Caches/com.apple.kext.caches/Startup/kernelcache_i386Try cache /System/Library/Caches/com.apple.kext.caches/Startup/kernelcache_i386 YES
Loading Darwin 10.6
Try cache failed

because Adler32 procedure is wrong applied. I don't know how to do better so I propose a workaround

CODE

            if(gMacOSVersion[3] == '6') {
                sprintf(gBootKernelCacheFile, "%skernelcache_%s", kDefaultCachePathSnow, (archCpuType == CPU_TYPE_I386) ? "i386" : "x86_64"); //, adler32);
                msglog("search for kernelcache %s", gBootKernelCacheFile);
                //Slice - TODO
                /*
                 - but the name is longer .adler32 and more...
                 kernelcache_i386.E102928C.qSs0
                 so will opendir and scan for some files
                */
                char* name;
                long flagsC;
                long timeC;
                struct dirstuff* cacheDir = opendir(kDefaultCachePathSnow);
                while(readdir(cacheDir, (const char**)&name, &flagsC, &timeC) >= 0)
                {
                    if(strcmp(&name[0], gBootKernelCacheFile) == 0) //?
                    {
                        //char* tmp = malloc(strlen(name) + 1);
                        //strcpy(tmp, name);
                        verbose("find kernelcache=%s\n", name);
                    }
                }

            }
            else //if(gMacOSVersion[3] == '5')
                sprintf(gBootKernelCacheFile, "%skernelcache", kDefaultCachePath);
        }

But this is only a first attempt. Quirky and not finished. There are may be several files and we should to choose most recent. Your propositions?
Also not sure about strcmp().

QUOTE

Regarding Linux detection, It is working for me. So I don't know where to start analysis from.

If it works for you I should more careful compare what is the difference.

EDITED:
This is also a place to discuss

CODE

            // TODO: fix this, the order does matter, and it's not correct now.
            // Also try to load Extensions from boot helper partitions.
            if (gBootVolume->flags & kBVFlagBooter)
            {
                strcpy(dirSpecExtra, "/com.apple.boot.P/System/Library/");
                if (FileLoadDrivers(dirSpecExtra, 0) != 0)
                {
                    strcpy(dirSpecExtra, "/com.apple.boot.R/System/Library/");
                    if (FileLoadDrivers(dirSpecExtra, 0) != 0)
                    {
                        strcpy(dirSpecExtra, "/com.apple.boot.S/System/Library/");
                        FileLoadDrivers(dirSpecExtra, 0);
                    }
                }
            }

PRS = Paper, Rock, Scissor
Order must be:
1. If P and R then P.
2. If P and S then S.
3. If R and S then R.

valv

Jan 29 2011, 11:21 PM

QUOTE (Slice @ Jan 29 2011, 09:04 PM)

Also not sure about strcmp().

I do prefer strstr() for things like this. after what we can use strcpy() to go forward.

Slice

Jan 31 2011, 07:33 AM

QUOTE (valv @ Jan 30 2011, 03:21 AM)

I do prefer strstr() for things like this. after what we can use strcpy() to go forward.

It is the happiest day of my life: somehow I got a help!
Now the problem with kernelcache for snow is resolved

CODE

search for kernelcache kernelcache_i386
find kernelcache=/System/Library/Caches/com.apple.kext.caches/Startup/kernelcache_i386.E102928C
Try cache /System/Library/Caches/com.apple.kext.caches/Startup/kernelcache_i386.E102928C YES
Loading Darwin 10.6
Loading kernel cache /System/Library/Caches/com.apple.kext.caches/Startup/kernelcache_i386.E102928C

I think the codes may be commited into trunc.

Now I am going to resolve next two problems: starting from CD and Linux detection.
No Valv, your bootloader also didn't detect Linux. It's not a big problem to check.
Click to view attachment
As you see FreeBSD detected but Linux - no while it presents here
Click to view attachment
So go to think about.

And about starting from CD. The same problem with your bootloader.
Click to view attachment
If you don't see a problem it doesn't mean the problem doesn't exists.

My newest bootloader compiled version is in topic.

wojnar

Jan 31 2011, 11:15 AM

@slice

I tried your latest bootloader.

bdmesg:

CODE

Last login: Mon Jan 31 11:57:09 on console
/Users/wojnar/Desktop/RC5m-721/i386/bdmesg ; exit;
MacBook-Pro-wojnar:~ wojnar$ /Users/wojnar/Desktop/RC5m-721/i386/bdmesg ; exit;
Using PCI-Root-UID value: 0
00:00.0 [060000] [8086:2a00] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x0,0x0)
00:01.0 [060400] [8086:2a01] (subsys [0000:0000]):: PciRoot(0x0)/Pci(0x1,0x0)
01:00.0 [030000] [10de:0609] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1,0x0)/Pci(0x0,0x0)
00:1a.0 [0c0300] [8086:2834] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1a,0x0)
00:1a.1 [0c0300] [8086:2835] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1a,0x1)
00:1a.7 [0c0320] [8086:283a] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1a,0x7)
00:1b.0 [040300] [8086:284b] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1b,0x0)
00:1c.0 [060400] [8086:283f] (subsys [0000:0000]):: PciRoot(0x0)/Pci(0x1c,0x0)
02:00.0 [028000] [14e4:4328] (subsys [1028:0226]):: PciRoot(0x0)/Pci(0x1c,0x0)/Pci(0x0,0x0)
00:1c.1 [060400] [8086:2841] (subsys [0000:0000]):: PciRoot(0x0)/Pci(0x1c,0x1)
03:00.0 [018000] [1095:3531] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1c,0x1)/Pci(0x0,0x0)
00:1c.2 [060400] [8086:2843] (subsys [0000:0000]):: PciRoot(0x0)/Pci(0x1c,0x2)
04:00.0 [020000] [10ec:8168] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1c,0x2)/Pci(0x0,0x0)
00:1c.3 [060400] [8086:2845] (subsys [0000:0000]):: PciRoot(0x0)/Pci(0x1c,0x3)
00:1c.4 [060400] [8086:2847] (subsys [0000:0000]):: PciRoot(0x0)/Pci(0x1c,0x4)
00:1d.0 [0c0300] [8086:2830] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1d,0x0)
00:1d.1 [0c0300] [8086:2831] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1d,0x1)
00:1d.2 [0c0300] [8086:2832] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1d,0x2)
00:1d.7 [0c0320] [8086:2836] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1d,0x7)
00:1e.0 [060401] [8086:2448] (subsys [0000:0000]):: PciRoot(0x0)/Pci(0x1e,0x0)
0c:06.0 [0c0010] [11c1:5811] (subsys [11c1:5811]):: PciRoot(0x0)/Pci(0x1e,0x0)/Pci(0x6,0x0)
00:1f.0 [060100] [8086:2811] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1f,0x0)
00:1f.1 [01018a] [8086:2850] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1f,0x1)
00:1f.2 [010601] [8086:2829] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1f,0x2)
00:1f.3 [0c0500] [8086:283e] (subsys [107b:0690]):: PciRoot(0x0)/Pci(0x1f,0x3)
CPU: BrandString Intel® Core™2 Duo CPU     T8300  @ 2.40GHz
measure_tsc_frequency = 2394MHz
msr(0x00cd): MSR_FSB_FREQ 200.0MHz
msr(0x00ce): platform_info 80170d28-43450813
msr(0x0198): MSR_IA32_PERF_STATUS 06170d28-06000d28
max: 13 current: 13
msr(0x0017): MSR_IA32_PLATFORM_ID 0xb9f48c22
DMI Chassis Info:
 Type=a
 Boot-up State=3
 Power Supply=3 Thermal State=3
CPU is Mobile
CPU: Brand:                 Intel® Core™2 Duo CPU     T8300  @ 2.40GHz
CPU: Vendor/Model/ExtModel: 0x756e6547/0x17/0x1
CPU: Family/ExtFamily:      0x6/0x0
CPU: MaxCoef/CurrCoef/Turbo:      0xd/0xd/0xd
CPU: MaxDiv/CurrDiv:        0x1/0x1
CPU: TSCFreq:               2394MHz
CPU: FSBFreq:               200MHz
CPU: CPUFreq:               2600MHz
CPU: NoCores/NoThreads:     2/0
CPU: Features:              0xbfebfbff
Attempting to read GPT
Read GPT
Reading GPT partition 1, type 48465300-0000-11AA-AA11-00306543ECAC
scanVolume 0
Attempting to read GPT
Read GPT
Reading GPT partition 1, type C12A7328-F81F-11D2-BA4B-00A0C93EC93B
Reading GPT partition 2, type 48465300-0000-11AA-AA11-00306543ECAC
Reading GPT partition 3, type 48465300-0000-11AA-AA11-00306543ECAC
scanVolume 1
diskScanFDiskBootVolumes 130
scanVolume 2
Buffer location: 0x1000FAF8 status: 1
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
getResolution: 8192x38616x32
Buffer location: 0x1000FAAC status: 1
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 
NVRAM started with ModulesLoaded
Start NVRAM reading
Name[0]:
Value[0]:
Name[1]:
Value[1]:
Name[2]:
Value[2]:
Required boot device is <array><dict><key>IOMatch</key><dict><key>IOProviderClass</key><string>IOMedia</string><key>IOPropertyMatch</key><dict><key>UUID</key><string>9B79B44D-2C32-4183-941F-FFFCE94A325E</string></dict></dict><key>BLLastBSDName</key><string>disk0s3</string></dict></array>
Found SMBIOS System Information Table 1
Customizing SystemID with : 108adec0-ab9b-11dc-805a-efb9f38380ca
NVRAM get a name efi-boot-device
NVRAM add name=efi-boot-device value=<array><dict><key>IOMatch</key><dict><key>IOProviderClass</key><string>IOMedia</string><key>IOPropertyMatch</key><dict><key>UUID</key><string>9B79B44D-2C32-4183-941F-FFFCE94A325E</string></dict></dict><key>BLLastBSDName</key><string>disk0s3</string></dict></array> length=265
NVRAM get a name efi-boot-device-data
 length=4 name=efi-boot-device-data value=
NVRAM get a name options
NVRAM add name=options value= length=2
NVRAM get a name 
GUI module screen width=8192 height=38616
GUI MAX VESA screen width=1280 height=1024
GUI: set mode 353: 1280x800x32
Setting boot-uuid to: 2AB62D86-BD05-3E36-BA0D-A1E31F89A1BC
Address loadConfigFile=2a072  bootInfo=43dc4
header read size 200
Incorrect image signature
search for kernelcache kernelcache_x86_64
find kernelcache=/System/Library/Caches/com.apple.kext.caches/Startup/kernelcache_x86_64.1304BDFF
Try cache /System/Library/Caches/com.apple.kext.caches/Startup/kernelcache_x86_64.1304BDFF NO
Loading Darwin 10.6
Loading kernel /mach_kernel
Wifi Controller [14e4:4328] :: PciRoot(0x0)/Pci(0x1c,0x0)/Pci(0x0,0x0)
Intel GM965/GL960 DRAM Controller [8086:2a00] at 00:00.0
Frequency detected: 333 MHz (666) Dual Channel 5-5-5-15
CPU is Intel® Core™2 Duo CPU     T8300  @ 2.40GHz, family 0x6, model 0x17
Patched DMI Table
Platform CPU Info:
 FSB=200
 MaxSpeed=0
 CurrentSpeed=2394
DMI CPU Info:
 FSB=0
 MaxSpeed=2048
 CurrentSpeed=2400
DMI CPU Info 2:
 Family=bf
 Socket=4
 Cores=85 Enabled=50 Threads=69
Overclocked CPU from 65364MHz to 1280MHz
Found SMBIOS System Information Table 1
Customizing SystemID with : 108adec0-ab9b-11dc-805a-efb9f38380ca
Actual model name is 'P-6831FX'
SSDT with CPU C-States generated successfully
FADT: Restart Fix applied!
Patched SMBIOS CPU Info:
 FSB=0
 MaxSpeed=2048
 CurrentSpeed=2400
 Family=bf Socket=4
 Cores=85 Enabled=50 Threads=69
LoadDrivers: Loading from [/Extra/Extensions.mkext]
LoadDrivers: Loading from [/System/Library/Caches/com.apple.kext.caches/Startup/Extensions.mkext]
Starting Darwin x86_64
Setting Legacy USB Off on controller [8086:2836] at 00:1d.7
Legacy USB Off Done
EHCI controller [8086:2836] at 00:1d.7 DMA @f2704c00
EHCI Acquire OS Ownership done
UHCI controller [8086:2832] at 00:1d.2 base c4(1881) reset
UHCI controller [8086:2831] at 00:1d.1 base c3(1861) reset
UHCI controller [8086:2830] at 00:1d.0 base c2(1841) reset
Setting Legacy USB Off on controller [8086:283a] at 00:1a.7
Legacy USB Off Done
EHCI controller [8086:283a] at 00:1a.7 DMA @f2704800
EHCI Acquire OS Ownership done
UHCI controller [8086:2835] at 00:1a.1 base c1(1821) reset
UHCI controller [8086:2834] at 00:1a.0 base c0(1801) reset

logout

[Proces zakończony]

Please tell me, your CPU is T8300 (as mine) and yours working with FSB=200, have you any issiues with that speed?
Time going slower, Sound are cracky or movie preview are cracky too??

When i tried boot with that value I have thats problems.

Have you any solution for thats problems??

Here is zip file with my .aml files, ioreg.

CODE

http://www.mediafire.com/?dbs9xr83wnczkgw

Cheers

Slice

Jan 31 2011, 11:48 AM

QUOTE (wojnar @ Jan 31 2011, 03:15 PM)

@slice

I tried your latest bootloader.

bdmesg:

Please tell me, your CPU is T8300 (as mine) and yours working with FSB=200, have you any issiues with that speed?
Time going slower, Sound are cracky or movie preview are cracky too??

When i tried boot with that value I have thats problems.

Have you any solution for thats problems??

Here is zip file with my .aml files, ioreg.

CODE

http://www.mediafire.com/?dbs9xr83wnczkgw

Cheers

Yes, I have exactly the same CPU and my FSB=200. But I see some diffs in your bdmesg

CODE

CPU: MaxCoef/CurrCoef/Turbo:      0xd/0xd/0xd
CPU: MaxDiv/CurrDiv:        0x1/0x1
CPU: TSCFreq:               2394MHz
CPU: FSBFreq:               200MHz
CPU: CPUFreq:               2600MHz
CPU: NoCores/NoThreads:     2/0

While I have

CODE

CPU: MaxCoef/CurrCoef/Turbo:      0xd/0xc/0xd
CPU: MaxDiv/CurrDiv:        0x1/0x1
CPU: TSCFreq:               2394MHz
CPU: FSBFreq:               200MHz
CPU: CPUFreq:               2400MHz

Why do you have current coeff at turbo speed while TSCFreq the same? It's a puzzle.

And next message is strange too

CODE

Overclocked CPU from 65364MHz to 1280MHz

First value is from incorrect DMI but why 1280? I need a time to think.
Sure, I have no problems with time, sound and movie.
May be your problem arises with c-state generator? Try without it.

wojnar

Jan 31 2011, 12:04 PM

QUOTE (Slice @ Jan 31 2011, 12:48 PM)

Yes, I have exactly the same CPU and my FSB=200. But I see some diffs in your bdmesg

CODE

CPU: MaxCoef/CurrCoef/Turbo:      0xd/0xd/0xd
CPU: MaxDiv/CurrDiv:        0x1/0x1
CPU: TSCFreq:               2394MHz
CPU: FSBFreq:               200MHz
CPU: CPUFreq:               2600MHz
CPU: NoCores/NoThreads:     2/0

While I have

CODE

CPU: MaxCoef/CurrCoef/Turbo:      0xd/0xc/0xd
CPU: MaxDiv/CurrDiv:        0x1/0x1
CPU: TSCFreq:               2394MHz
CPU: FSBFreq:               200MHz
CPU: CPUFreq:               2400MHz

Why do you have current coeff at turbo speed while TSCFreq the same? It's a puzzle.

And next message is strange too

CODE

Overclocked CPU from 65364MHz to 1280MHz

First value is from incorrect DMI but why 1280? I need a time to think.
Sure, I have no problems with time, sound and movie.
May be your problem arises with c-state generator? Try without it.

Look in my CABP, I don't use GenerateCStates.

I'm currently under trunk r699, with busratio=13 and FSB=184, without thats issues.

Please take your time to think

valv

Jan 31 2011, 12:23 PM

QUOTE (Slice @ Jan 31 2011, 08:33 AM)

It is the happiest day of my life: somehow I got a help!
Now the problem with kernelcache for snow is resolved

that's good news. well done Slice. btw, the thing with linux detection could probably be related to the used disk architecture and/or the way linux was installed. Can also be the VM.
Let me tell what works (in my case):
disk architecture is guid, on both internal hdd and usb stick. grub is on the same partition as the os itself (as u know). Now, no matter from where I do boot, linux partition is shown and I can boot to it without any problem. So the issue is probably limited to cdboot. On the first screenshot, I see bootloader is on en(0,0). Could u tell me which one is that ? I guess it's the CD, right ?
I would look further at this later on.
Also, Slice, unrelated and u probably knew about it already, but pay attention to the fix over here. It looks interesting to me.

Keep up the good work

Slice

Jan 31 2011, 01:43 PM

QUOTE (valv @ Jan 31 2011, 04:23 PM)

that's good news. well done Slice. btw, the thing with linux detection could probably be related to the used disk architecture and/or the way linux was installed. Can also be the VM.
Let me tell what works (in my case):
disk architecture is guid, on both internal hdd and usb stick. grub is on the same partition as the os itself (as u know). Now, no matter from where I do boot, linux partition is shown and I can boot to it without any problem. So the issue is probably limited to cdboot. On the first screenshot, I see bootloader is on en(0,0). Could u tell me which one is that ? I guess it's the CD, right ?
I would look further at this later on.

Yes, it is CD prepared with your latest bootloader (as you see from screen).
I see one difference. I use MBR instead of GPT. So I will to propose that it works with GPT but not with MBR. Right?
Usually Linux do not use GPT, it uses MBR.

QUOTE

Also, Slice, unrelated and u probably knew about it already, but pay attention to the fix over here. It looks interesting to me.

Keep up the good work

Yes, it looks interesting. I want to know how to check. In what situation I can see that the bootloader behavior is changed? May be the problem Nr2? I can check with CD boot. From USB stick I see no problems so far.

valv

Jan 31 2011, 03:12 PM

QUOTE (Slice @ Jan 31 2011, 02:43 PM)

Yes, it is CD prepared with your latest bootloader (as you see from screen).
I see one difference. I use MBR instead of GPT. So I will to propose that it works with GPT but not with MBR. Right?

Most likely yes.

QUOTE (Slice @ Jan 31 2011, 02:43 PM)

Usually Linux do not use GPT, it uses MBR.

Well, it doesn't have to mess with MBR neither. It is accomplishable by relying on Disk Utility (osx) for making the hdd GPT first and preparing linux partitions afterwards. Only then we can set linux up, while knowing it is not going to mess with mbr (if we put its boot-loader on the same partition as the os).

QUOTE (Slice @ Jan 31 2011, 02:43 PM)

Yes, it looks interesting. I want to know how to check. In what situation I can see that the bootloader behavior is changed? May be the problem Nr2? I can check with CD boot. From USB stick I see no problems so far.

Neither had I. Honestly, I don't remember where the Nr2 issue is. But iirc, wojnar0 also asked me to fix an issue like that. Only tests could confirm it though.

vincent99

Jan 31 2011, 06:19 PM

Hi slice , i have a problem with last rev from svn, from verbose i see this in log, with consequent kernel panic on voodoohda:

CODE

error: failed to open PCI device

and immediately panic on voodoohda.

This happens on my 2 hack with voodoohda.

Second problem: autoresolution on boot doesn't work, instead with chazi, meklort, autores works well.

Slice

Feb 1 2011, 07:50 AM

QUOTE (vincent99 @ Jan 31 2011, 10:19 PM)

Hi slice , i have a problem with last rev from svn, from verbose i see this in log, with consequent kernel panic on voodoohda:

CODE

error: failed to open PCI device

and immediately panic on voodoohda.

This happens on my 2 hack with voodoohda.

Did you have AppleHDA installed? Did you use KextPatcher.dylib?

QUOTE

Second problem: autoresolution on boot doesn't work, instead with chazi, meklort, autores works well.

As well as for me. Thank you for info that with meklort autores works. Going to compare.

PS. Your bdmesg?

vincent99

Feb 1 2011, 10:32 AM

QUOTE

Did you have AppleHDA installed? Did you use KextPatcher.dylib?

I have Applehdadisabler in .mkext in Extra and it has always works well, nothing problem with other loaders.
No, i don't use the kext patcher module.

QUOTE

PS. Your bdmesg?

Here, bdmesg with valv boot, ATM

Click to view attachment

Slice

Feb 1 2011, 11:20 AM

You found a problem that is new for me.
Try to delete caches
sudo rm -r /System/Library/Caches/com.apple.kext.caches/*

vincent99

Feb 1 2011, 11:45 AM

Hi slice, now it works

strange problem, very strange,only on this...

btw the new bdmesg with your boot:

Click to view attachment

Now the problem is autores

Good work slice..

EDIT: the problem is back

Something is wrong. Is of little to use delete every time the kext cache, i don't know...

Slice

Feb 1 2011, 02:54 PM

1. I know the problem with resolution and I can't find a difference with meklort except I manually restrict initial resolution because my screen is not switched at first screen. I am thinking about.

CODE

GUI module screen width=1280 height=800 -- the module Resolution works fine
GUI: set mode 280: 1024x768x32 -- my manual restriction

2. About cache. You seems to use Extra/Extensions. But system doesn't know anything about it so it created kernelcache without Extra.
All other bootloaders can't use kernelcache and so no conflict with Extra.
May be I need to check if e/e/* present and in this case do not use kernelcache as others did.

Kappy

Feb 11 2011, 04:51 AM

@ Slice:

I created a minor problem I don't know how to fix. I was experimenting with the Chameleon bootloader doing an install that was supposed to be on a USB flash drive. Inadvertently I installed Chameleon on my Mac Pro's boot volume. I have removed the 'boot' file, so the Mac Pro is booting OK. However, it KPs when restarting or shutting down. I assume because the 'boot0' and 'boot1h' files have been written to the Mac Pro's boot drive.

I don't know what to 'fdisk' and/or 'dd' to fix it. I assume there's a way to restore the real OS X files to fix this, but I don't know how. If you can help me I'd much appreciate it.

Kappy

Slice

Feb 11 2011, 08:39 AM

No, I can't help with real Mac. I don't know what is boot0 and boot1h must be here. Reinstall MacOSX?

Kappy

Feb 11 2011, 07:20 PM

QUOTE (Slice @ Feb 11 2011, 03:39 AM)

No, I can't help with real Mac. I don't know what is boot0 and boot1h must be here. Reinstall MacOSX?

@ Slice:

OK. I was trying to avoid going through a reinstallation. Thanks, anyway.

barnum

Feb 28 2011, 03:25 PM

Hi slice,

I'm looking for your bootloader.
For the moment I use the AnVal and I want to test Yours.
Does I just need to copy your boot to my boot partition et copy modules to "/Extra/module" ?
or I must also modify my smbios.plist and my "com.apple.boot.plist"

regards,
Barnum

Slice

Mar 2 2011, 11:36 AM

QUOTE (barnum @ Feb 28 2011, 07:25 PM)

Hi slice,

I'm looking for your bootloader.
For the moment I use the AnVal and I want to test Yours.
Does I just need to copy your boot to my boot partition et copy modules to "/Extra/module" ?
or I must also modify my smbios.plist and my "com.apple.boot.plist"

regards,
Barnum

Yes, just copy boot and modules.
smbios.plist and boot.plist is the same. If you use some special features from AnVal then my bootloader may ignore them as useless or automatic performed.

yeehaa

Mar 2 2011, 09:28 PM

@slice: will you please change the ati graphics enabler to match kabyl's branch? it has a feature to specify custom framebuffers which helps ati users a lot. thanks in advance!!

also all ati users should check out bcc9's new connector info mod over at insanely which lets you patch the connector info so you are not restricted to the hard coded ones for your frame buffer..!

Slice

Mar 3 2011, 07:38 AM

QUOTE (yeehaa @ Mar 3 2011, 01:28 AM)

@slice: will you please change the ati graphics enabler to match kabyl's branch? it has a feature to specify custom framebuffers which helps ati users a lot. thanks in advance!!

also all ati users should check out bcc9's new connector info mod over at insanely which lets you patch the connector info so you are not restricted to the hard coded ones for your frame buffer..!

I made the changes in svn but I can't check if it works. Compile and check by yourself.

I am not sure that I will continue this development because of Lion. It seems EFI-based bootloader needed but no open-source project exists so can't say anything about the future yet.

yeehaa

Mar 3 2011, 08:37 PM

QUOTE (Slice @ Mar 3 2011, 07:38 AM)

I made the changes in svn but I can't check if it works. Compile and check by yourself.

I am not sure that I will continue this development because of Lion. It seems EFI-based bootloader needed but no open-source project exists so can't say anything about the future yet.

i have a basic test system setup which is very unstable. let me try to install xcode on it..

abt the bootloader n lion: dont u think all these "features" will get merged to a newer true EFI based chameleon project? most probably these module system will work out pretty well during that portin!

Slice

Mar 4 2011, 07:33 AM

It's a problem there are no any open-source EFI-based bootloader. DUET can be compiled only under Windows/VS that I have no.
So I don't know from what to start.
There is a change to improve Chameleon -> ChameLion but we should to understand what is wrong.
Testers have the same panic
... no kexts summary
What is mean? Why it is not happen with EFI?

Kabyl

Mar 4 2011, 08:05 AM

QUOTE (Slice @ Mar 4 2011, 08:33 AM)

It's a problem there are no any open-source EFI-based bootloader. DUET can be compiled only under Windows/VS that I have no.
So I don't know from what to start.

You can compile EDKII under OS X just fine.

QUOTE

There is a change to improve Chameleon -> ChameLion but we should to understand what is wrong.

It seems someone was able to boot Lion with Chameleon; http://netkas.org/?p=624#comment-153363

QUOTE

Testers have the same panic
... no kexts summary
What is mean? Why it is not happen with EFI?

Simply because you're using Apple's boot.efi.

Slice

Mar 4 2011, 09:55 AM

QUOTE (Kabyl @ Mar 4 2011, 12:05 PM)

You can compile EDKII under OS X just fine.

EDKII yes, but DUET - no. XCode project doesn't include DUET target.

QUOTE

It seems someone was able to boot Lion with Chameleon; http://netkas.org/?p=624#comment-153363

I already do this and got the same KP as in screenshot. I expected that this bootloader may be good for some user but no one my tester was successful.

QUOTE

Simply because you're using Apple's boot.efi.

What if Chameleon will load boot.efi instead of mach_kernel?

Kabyl

Mar 4 2011, 05:04 PM

QUOTE (Slice @ Mar 4 2011, 10:55 AM)

EDKII yes, but DUET - no. XCode project doesn't include DUET target.

I've been building it from Terminal, and it works fine, and it's probably easier now with the efitools found in cctools.
http://sourceforge.net/apps/mediawiki/tian...s#Mac_OS_X_10.5
http://sourceforge.net/apps/mediawiki/tian...php?title=Xcode

QUOTE

I already do this and got the same KP as in screenshot. I expected that this bootloader may be good for some user but no one my tester was successful.

The DT structure seems to be the same.

It's obviously something to do with the mkext (kexts) loading, maybe the booter extracts (or creates, in the case of kexts) a structure before setting the DriversPackage-% (or the Driver-%) properties.

QUOTE

What if Chameleon will load boot.efi instead of mach_kernel?

Probably too much work, I would prefer working on a DUET based booter.

Slice

Mar 4 2011, 06:02 PM

QUOTE (Kabyl @ Mar 4 2011, 09:04 PM)

I've been building it from Terminal, and it works fine, and it's probably easier now with the efitools found in cctools.
http://sourceforge.net/apps/mediawiki/tian...s#Mac_OS_X_10.5
http://sourceforge.net/apps/mediawiki/tian...php?title=Xcode

The DT structure seems to be the same.

It's obviously something to do with the mkext (kexts) loading, maybe the booter extracts (or creates, in the case of kexts) a structure before setting the DriversPackage-% (or the Driver-%) properties.

Probably too much work, I would prefer working on a DUET based booter.

I will try to follow this guide to compile DUET. If successful then I can take part.
About Chameleon I think not so much work as with DUET. But the last one is more perspective.
We just need to make real EfiRuntimeServices, then Load("boot.efi") then jump to it.

The common problem for both case is a set of drivers. Bootloader needs HDD, USB_MassStorage, DVD, PS2, VGA.
1. Chameleon works with BIOS drivers. This is universal approach and always works. But we can't handle these drivers into EfiRuntime after kernel started.
2. EFI-based loader uses own drivers, universal, but they don't work in a half of configs.
3. GRUB (?) uses drivers from linux_kernel more or less working almost with any hardware, but it should load ~20Mb before start.

May be it is possible to make a mix? EFI bootloader but with BIOS drivers?

Kabyl

Mar 4 2011, 06:24 PM

QUOTE (Slice @ Mar 4 2011, 07:02 PM)

I will try to follow this guide to compile DUET. If successful then I can take part.
About Chameleon I think not so much work as with DUET. But the last one is more perspective.
We just need to make real EfiRuntimeServices, then Load("boot.efi") then jump to it.

We probably don't need the RuntimeServices, but we, absolutely, need BootServices, which I still think is too much. And we would also need to consider supporting the 64bit part of boot.efi

more work...

IMHO, working and fixing DUET is a better.

QUOTE

The common problem for both case is a set of drivers. Bootloader needs HDD, USB_MassStorage, DVD, PS2, VGA.
1. Chameleon works with BIOS drivers. This is universal approach and always works. But we can't handle these drivers into EfiRuntime after kernel started.
2. EFI-based loader uses own drivers, universal, but they don't work in a half of configs.

One could work on fixing that.

QUOTE

3. GRUB (?) uses drivers from linux_kernel more or less working almost with any hardware, but it should load ~20Mb before start.

grub2 has its own drivers, and it could be optimized. But it would also require Lion support, same thing with current Chameleon.

Slice

Mar 4 2011, 10:06 PM

Almost successful installing all tools but gcc-4.3.0 report an error

CODE

ld: duplicate symbol _init_inline_once in libbackend.a(tree-inline.o) and tree-inline.o
collect2: ld returned 1 exit status
make[1]: *** [cc1plus-dummy] Error 1

and Edk2 main build

CODE

"ld" -arch i386 -u __ModuleEntryPoint -e __ModuleEntryPoint -preload -segalign 0x20 -pie -all_load -dead_strip -seg1addr 0x220 -read_only_relocs suppress -map /Developer/Projects/EFI/edk2/Build/Unix/RELEASE_XCODE32/IA32/MdeModulePkg/Core/Dxe/DxeMain/DEBUG/DxeCore.map -o /Developer/Projects/EFI/edk2/Build/Unix/RELEASE_XCODE32/IA32/MdeModulePkg/Core/Dxe/DxeMain/DEBUG/DxeCore.dll  -filelist /Developer/Projects/EFI/edk2/Build/Unix/RELEASE_XCODE32/IA32/MdeModulePkg/Core/Dxe/DxeMain/OUTPUT/static_library_files.lst
Undefined symbols:
  "_memcpy", referenced from:
      _LzmaDec_AllocateProbs in LzmaDecompressLib.lib(LzmaDec.obj)
ld: symbol(s) not found
make: *** [/Developer/Projects/EFI/edk2/Build/Unix/RELEASE_XCODE32/IA32/MdeModulePkg/Core/Dxe/DxeMain/DEBUG/DxeCore.dll] Error 1

build.py...
: error 7000: Failed to execute command
    make tbuild [/Developer/Projects/EFI/edk2/Build/Unix/RELEASE_XCODE32/IA32/MdeModulePkg/Core/Dxe/DxeMain]

build.py...
: error F002: Failed to build module
    /Developer/Projects/EFI/edk2/MdeModulePkg/Core/Dxe/DxeMain.inf [IA32, XCODE32, RELEASE]

and Duet

CODE

/Developer/Projects/EFI/edk2/Build/DuetPkgIA32/DEBUG_XCODE32/IA32/UefiCpuPkg/CpuDxe/CpuDxe/OUTPUT/Ia32/IvtAsm.iii:60:0 assumed for missing expression
/Developer/Projects/EFI/edk2/Build/DuetPkgIA32/DEBUG_XCODE32/IA32/UefiCpuPkg/CpuDxe/CpuDxe/OUTPUT/Ia32/IvtAsm.iii:60:Rest of line ignored. 1st junk character valued 92 (\).
make: *** [/Developer/Projects/EFI/edk2/Build/DuetPkgIA32/DEBUG_XCODE32/IA32/UefiCpuPkg/CpuDxe/CpuDxe/OUTPUT/Ia32/IvtAsm.obj] Error 1

build.py...
: error 7000: Failed to execute command
make tbuild [/Developer/Projects/EFI/edk2/Build/DuetPkgIA32/DEBUG_XCODE32/IA32/UefiCpuPkg/CpuDxe/CpuDxe]

build.py...
: error F002: Failed to build module
/Developer/Projects/EFI/edk2/UefiCpuPkg/CpuDxe/CpuDxe.inf [IA32, XCODE32, DEBUG]

- Failed -

mighty

Mar 17 2011, 12:20 AM

Hi Slice
Your bootloader works really good. It detects my CPU just as it is unlike the trunk version which gave wrong FSB. However there is a small issue which makes the sound a little crackling. I dont recall this crackling sound in trunk version. But your version and AnVAL both get me this crackling sound. I m using a patched AppleHDA, i tried VoodooHDA which turned out the same matter.
Can you please take a look this possible bug?
Thanks