ProjectOSX

Welcome Guest!

Returning User? Login here.

Want membership privileges? Register here.

54 Pages V  « < 10 11 12 13 14 > »   
Start a new topic Add Reply
> Uefi Booting On Ami Aptio Uefi Boards, experiences with other UEFI implementations are welcome
dgsga
post Jun 5 2012, 10:35 AM
Post #221
I can now confirm that HDA injection is the cause of the delay. If I set HDA injection to No in config.plist and instead inject HDA via DSDT.aml then booting is instant.
Asus Maximus V Gene, i3770K@4.6GHz, 8Gb RAM, GeForce660Ti, 10.9.1 via Clover
гык-sse2
post Jun 5 2012, 01:49 PM
Post #222
QUOTE (Slice @ Jun 4 2012, 11:54 AM) *
It does contain AptioFix but 64bit only. Did you compile 32bit?
[attachment=4468:Screen_s...11.52.58.png]

CloverGrower always compiles both 32 and 64, but I looked again in that folder - there is no HFSPlus and no AptioFix, only Smbios and FatBin.
UEFI is a great way to get the best of the best in the world of imagination and creativity. ©Google Scribe
Slice
post Jun 5 2012, 05:57 PM
Post #223
QUOTE (гык-sse2 @ Jun 5 2012, 05:49 PM) *
CloverGrower always compiles both 32 and 64, but I looked again in that folder - there is no HFSPlus and no AptioFix, only Smbios and FatBin.

I compile with

cd Clover
./ebuild.sh -64
cd rEFIt_UEFI
./build64.sh
i3-2120 GA-H61M-S1 UEFI, Radeon HD6670-UEFI, ALC887(VoodooHDA 2.8.6), OS⌘10.9.5, OS⌘ 10.7.5 Clover FakeSMC_plugins_3.3.1 Realtek LAN v3.1.2
dmazar
post Jun 5 2012, 08:41 PM
Post #224
QUOTE (kizwan @ Jun 5 2012, 06:27 AM) *
This time it kernel panic after loading kernelcache & UUID message.

panic(cpu 0 caller 0xffffff8000295a37): "pmap_steal_memory"@/SourceCache/xnu/xnu-1699.22.73/osfmk/vm/vm_resident.c:746
I had pmap_steal_memory when memory map was too big. It seems that your UEFI can not exit boot services if shrinking is done in GetMemoryMap like Aptio driver was doing so far, so I have removed it in the last test - and you got pmap_steal kp - at least you exited boot services, applied patches and get to kernel.
(By the way, is seems that this shrinking is not needed on 10.7.4 and 10.8 here. 10.6.8 also boots without it, but quickly KPs.)

Here is the version that will shrink memmap later in the process. Lets see how far can you go now:
Attached File  OsxAptioFixDrv.efi.zip ( 13.52K ) Number of downloads: 32

HW: Asus P8P67-M, Intel Core i5-2300, 4GB, XFX HD-567X-ZHH3 SW: SL, L, ML: Clover UEFI boot
dmazar
post Jun 5 2012, 08:52 PM
Post #225
QUOTE (jadran @ Jun 4 2012, 10:15 PM) *
Problem is when I add HFSPlus.efi to driver folder;
If I add only HFSPlus.efi to drivers64 folder it will stall, but loading from shell it works... I tired loading in different scenarios,I added numbers before every name like:
1Aptio.efi
2HFSPlus.efi
3FatBin.efi
and all versions around.

problem is the same, will try new cloverx64.efi later, now I must do some other work and I am not in front mine "testing" machine.
I'm joining your club now. For some reason my HFS driver stopped working. If I load Clover from shell then it works. But if I boot directly into it - stops after loading drivers. Actually, stops after volume scanning, since - no volumes can be detected. Scanning for BlockIO protocol returns handles for all volumes, but they are all set to 0. Some weird issue with this driver. As soon as I remove it or replace with VBox - all works.

I'm not sure what happened here that caused it to stop working. At some point I have removed one partition from the drive with Disk Utility in MountainLion and clicked Verify Disk for some partitions, but I'm not sure if driver problem occurred right after that.
HW: Asus P8P67-M, Intel Core i5-2300, 4GB, XFX HD-567X-ZHH3 SW: SL, L, ML: Clover UEFI boot
dmazar
post Jun 5 2012, 10:01 PM
Post #226
QUOTE (dmazar @ Jun 5 2012, 10:52 PM) *
I'm joining your club now. For some reason my HFS driver stopped working.
Got it back. But, check this:

1. Booting from USB stick: if I have cloverx64.efi renamed to bootx64.efi and start from stick - then it does not work.
If bootx64.efi is a shell -> boot into shell -> start clover from shell - it works

2. HDD, EFI partition: cloverx64.efi added as a boot option and boot option is default one (first one to boot) - power on -> clover starts -> then it works

3. HDD, EFI partition: cloverx64.efi added as a boot option, but not set as default one (not first) - power on -> using F8 to select clover from the list of boot options -> and it does not work

4. completely the same as 3. but some USB stick is attached - then it works

Bloody hell! blink.gif

(All above "does not work" means that clover can not read volumes because all BlockIO handles returned in ScanVolumes() are 0, which is wrong.)
HW: Asus P8P67-M, Intel Core i5-2300, 4GB, XFX HD-567X-ZHH3 SW: SL, L, ML: Clover UEFI boot
Slice
post Jun 6 2012, 06:43 AM
Post #227
I don't know what part of codes contains this mistakes. I think it is a bug of UEFI. EDK2 is updated everydays to eliminate bugs but your BIOS.
So my proposition is to think about replacement of LocateHandle function.
USB device is removable device and if there is no media then BlockIo == NULL.
And see MdeModulePkg/Core/Dxe/Hand/Locate.c (last update march2011)
CODE
  for (;;) {
    //
    // Get the next handle.  If no more handles, stop
    //
    Handle = GetNext (&Position, &Interface);
    if (NULL == Handle) {
      break;
    }

What if NULL will appear at the begin of the array? Stop searching?
We can just replace in Clover
Status = gBS->LocateHandleBuffer(ByProtocol, ...
by
Status = OurLocate....
If we understand better algo.
i3-2120 GA-H61M-S1 UEFI, Radeon HD6670-UEFI, ALC887(VoodooHDA 2.8.6), OS⌘10.9.5, OS⌘ 10.7.5 Clover FakeSMC_plugins_3.3.1 Realtek LAN v3.1.2
kizwan
post Jun 6 2012, 08:53 AM
Post #228
QUOTE (dmazar @ Jun 6 2012, 04:41 AM) *
I had pmap_steal_memory when memory map was too big. It seems that your UEFI can not exit boot services if shrinking is done in GetMemoryMap like Aptio driver was doing so far, so I have removed it in the last test - and you got pmap_steal kp - at least you exited boot services, applied patches and get to kernel.
(By the way, is seems that this shrinking is not needed on 10.7.4 and 10.8 here. 10.6.8 also boots without it, but quickly KPs.)

Here is the version that will shrink memmap later in the process. Lets see how far can you go now:
Attached File  OsxAptioFixDrv.efi.zip ( 13.52K ) Number of downloads: 32

This time it go further:-
CODE
  Loading kernel cache file 'System\Library\Caches\com.apple.kext.caches\Startup\kernelcache'...
....................
root device uuid is '7E8241F0-1D45-324F-9CF0-2D5FD08D103F'
Loading drivers...
................................................................................
................................................................................
................................................................................
................................................................................
................................................................................
................................................................................
................................................................................
................................................................................
...........................................................

After Loading drivers, it reboot. DebugLog is not created (or deleted). I'll try to obtain DebugLog.

This post has been edited by kizwan: Jun 6 2012, 08:54 AM
1) EFI (Chameleon 2.0-RC4) + Retail Mac OS X 10.6
** Acer Aspire 9420 (9424WSMi) Intel C2D T5600 @ 1.83GHz ** 2GB DDR2 ** NVIDIA GeForce Go 7300 128MB ** Dell Wireless 1395 (rebrand to AirPort Extreme) ** Intel HDA ALC883 - all devices working
2) EFI (munky's method & Chameleon 2.0-RC1) + Retail Mac OS X 10.5.6 + Vista Ultimate (3rd partition) (one GUID disk)
** Acer Aspire 3680 (3684NWXMI) Intel Celeron M 440 @ 1.87GHz ** 1GB DDR2 ** Intel GMA950 ** Linksys WPC54G v3.1 (Broadcom 4318) ** Intel HDA ALC883 - all devices working
3) Windows 7 Home Premium 64-bit
** Dell Studio 1557 - Intel Core i7 720QM ** 4GB DDR3 1333Mhz ** ATI HD4570 512MB ** Intel WiFi 5300 ** Seagate Momentus 5400.6 SATA 3Gb/s 320-GB Hard Drive 5400RPM ** WLED 1366x768 15.6" **
dmazar
post Jun 6 2012, 10:02 AM
Post #229
I do not know why it loads drivers after kernelcache. Previously you had panic after kernelcache, meaning kernel was started. Hm.
HW: Asus P8P67-M, Intel Core i5-2300, 4GB, XFX HD-567X-ZHH3 SW: SL, L, ML: Clover UEFI boot
kizwan
post Jun 6 2012, 03:56 PM
Post #230
You're right. It shouldn't loading drivers. It loading drivers because earlier I boot (using Chameleon USB boot disk) with -x. I fixed it by rebuilding kernelcache. However, I got instant reboot after I see this message:-
CODE
  Loading kernel cache file 'System\Library\Caches\com.apple.kext.caches\Startup\kernelcache'...
....................
root device uuid is '7E8241F0-1D45-324F-9CF0-2D5FD08D103F'


When it auto-reboot, DebugLog variable is deleted. So, no DebugLog.
1) EFI (Chameleon 2.0-RC4) + Retail Mac OS X 10.6
** Acer Aspire 9420 (9424WSMi) Intel C2D T5600 @ 1.83GHz ** 2GB DDR2 ** NVIDIA GeForce Go 7300 128MB ** Dell Wireless 1395 (rebrand to AirPort Extreme) ** Intel HDA ALC883 - all devices working
2) EFI (munky's method & Chameleon 2.0-RC1) + Retail Mac OS X 10.5.6 + Vista Ultimate (3rd partition) (one GUID disk)
** Acer Aspire 3680 (3684NWXMI) Intel Celeron M 440 @ 1.87GHz ** 1GB DDR2 ** Intel GMA950 ** Linksys WPC54G v3.1 (Broadcom 4318) ** Intel HDA ALC883 - all devices working
3) Windows 7 Home Premium 64-bit
** Dell Studio 1557 - Intel Core i7 720QM ** 4GB DDR3 1333Mhz ** ATI HD4570 512MB ** Intel WiFi 5300 ** Seagate Momentus 5400.6 SATA 3Gb/s 320-GB Hard Drive 5400RPM ** WLED 1366x768 15.6" **
STLVNUB
post Jun 6 2012, 05:08 PM
Post #231
QUOTE (Slice @ Jun 6 2012, 03:57 AM) *
I compile with

cd Clover
./ebuild.sh -64
cd rEFIt_UEFI
./build64.sh

CloverGrowerV1.0 will default to using ebuild.sh for Clover and build32/64.sh for rEFIt.
Azrock Z77 Pro3, G1610 AMD HD7750 Clover/Ozmosis
Toshiba A660-0MR00R I7-7740QM GT330M 8 Gig Clover

Chumnick
post Jun 7 2012, 12:23 PM
Post #232
Hi dmazar,
having two chips with the BIOS 0301 (APTIO2.0) and 3304 (APTIO2.31), I tested the Clover for UEFI and demonstrated the following:
with APTIO2.0 - SelfDevicePath=Acpi(PNP0A03,0)/Pci(1F|2)/Sata(1,0)/HD(Part1,SigF0947B66-41F4-4A93-9AC4-1FAA99CDAC64) @211F418, boot and work normally, and UUID - taken from the config as it should be
with APTIO2.31 - SelfDevicePath=PciRoot(0x0)/Pci(0x1f,0x2)/Sata(0x1,0x0)/HD(1,GPT,f0947b66-41f4-4a93-9ac4-1faa99cdac64,0x28,0x64000) @C167B18, boot normally, but UUID - not taken from the config and it is observed only at boot UEFI.
Tested with all drivers OsxAptioFixDrv and last versions Clover.
Sorry for my English. Thank you.
ASUS P8Z68 Deluxe/Gen3, Intel i7-2600k, 16Gb 1133Mh DDR3, Sapphire Radeon HD 6870 1024Mb, OSZ Vertex3 SSD 240Gb, MacOS X 10.8, Clover, iMac 12,2
Gigabyte GA-Z77X-UP5 TH, Intel i7 3770k, 16Gb 1600Mh, ASUS GeForce GTX 670 2048 MB, Intel SSD 120Gb, MacOS X 10.8, Clover, iMac 12,2
dmazar
post Jun 7 2012, 01:43 PM
Post #233
Hi Chumnick,

All device paths (including SelfDevicePath) are converted to string with DevicePathToText protocol/driver. Aptio 2.0 does not have it's own and then Clover is using it's own internal version (Acpi(PNP0A03,0)/Pci(1F|2)...).

Aptio 2.3 has it's own version and Clover uses that and generates different text (PciRoot(0x0)/Pci(0x1f,0x2)...). That's why you have different SelfDevicePaths. But this is only different textual representation of the same binary device path. That is normal and not a problem.

About not injecting UUID from config: I had that situation plus some FireWare errors plus slow "DSMOS arrived" message, but only when UEFI runtime services were not passed properly to OSX. It was the same situation here as when starting 64bit Lion from 32 bit Clover - in that case (U)EFI runtime services were not used by OSX because they were not compatible (32 bit / 64 bit).

When you boot on Aptio 2.3, do you have working runtime services - does "nvram -p" in terminal works? Does it print out some variables?

I do not know what is happening in your case. Make sure that you are starting 64 bit OSX and try with AptioFix debug version from post 224
http://www.projectosx.com/forum/index.php?...amp;#entry18588

This one should write some debug info to NVRAM. After trying booting with it, you can extract debug info from efi shell or from OSX (if OSX can be started from UEFI):

from efi shell:
- boot into shell, move to EFI or some FAT32 partition (fs0: or fs1: or fs2: ...) and do
> dmpstore DebugLog -s debug.txt

from OSX terminal (works only with latest AptioFix debug version from post 224):
> nvram -xp > nvram.plist

You can also post DarwinDumper results from OSX booted on Aptio2.3 with Clover/AptioFix driver, and I can try to see what's going on.

Just noticed, you have the same partition UID on Aptio2.0 and Aptio2.3 machine. Are you using the same drive and switching it between machines?

This post has been edited by dmazar: Jun 7 2012, 01:46 PM
HW: Asus P8P67-M, Intel Core i5-2300, 4GB, XFX HD-567X-ZHH3 SW: SL, L, ML: Clover UEFI boot
Chumnick
post Jun 7 2012, 02:08 PM
Post #234
QUOTE (dmazar @ Jun 7 2012, 02:43 PM) *
Just noticed, you have the same partition UID on Aptio2.0 and Aptio2.3 machine. Are you using the same drive and switching it between machines?

dmazar,
It's one machine for which I have two chips with two BIOSes and for testing change them. I have the same partition UID on Aptio2.0 and Aptio2.3 and same drive.
Other reports, I'll post later today. Thanks.
ASUS P8Z68 Deluxe/Gen3, Intel i7-2600k, 16Gb 1133Mh DDR3, Sapphire Radeon HD 6870 1024Mb, OSZ Vertex3 SSD 240Gb, MacOS X 10.8, Clover, iMac 12,2
Gigabyte GA-Z77X-UP5 TH, Intel i7 3770k, 16Gb 1600Mh, ASUS GeForce GTX 670 2048 MB, Intel SSD 120Gb, MacOS X 10.8, Clover, iMac 12,2
Chumnick
post Jun 7 2012, 04:56 PM
Post #235
QUOTE (dmazar @ Jun 7 2012, 01:43 PM) *
Hi Chumnick,

When you boot on Aptio 2.3, do you have working runtime services - does "nvram -p" in terminal works? Does it print out some variables?

I do not know what is happening in your case. Make sure that you are starting 64 bit OSX and try with AptioFix debug version from post 224
http://www.projectosx.com/forum/index.php?...amp;#entry18588

This one should write some debug info to NVRAM. After trying booting with it, you can extract debug info from efi shell or from OSX (if OSX can be started from UEFI):

from OSX terminal (works only with latest AptioFix debug version from post 224):
> nvram -xp > nvram.plist

You can also post DarwinDumper results from OSX booted on Aptio2.3 with Clover/AptioFix driver, and I can try to see what's going on.

Dmazar,
sending my reports according to your requirements:
Attached File  nvram.plist.zip ( 3.14K ) Number of downloads: 6

Attached File  DarwinDumper_v1.2w_Nickhost_11.4.0_x86_64_American_Megatrends_Lion.zip ( 6.46K ) Number of downloads: 8

Attached File  nvram_p.txt.zip ( 1.96K ) Number of downloads: 7


This post has been edited by Chumnick: Jun 7 2012, 04:57 PM
ASUS P8Z68 Deluxe/Gen3, Intel i7-2600k, 16Gb 1133Mh DDR3, Sapphire Radeon HD 6870 1024Mb, OSZ Vertex3 SSD 240Gb, MacOS X 10.8, Clover, iMac 12,2
Gigabyte GA-Z77X-UP5 TH, Intel i7 3770k, 16Gb 1600Mh, ASUS GeForce GTX 670 2048 MB, Intel SSD 120Gb, MacOS X 10.8, Clover, iMac 12,2
dmazar
post Jun 7 2012, 07:04 PM
Post #236
QUOTE (kizwan @ Jun 6 2012, 05:56 PM) *
When it auto-reboot, DebugLog variable is deleted. So, no DebugLog.
Hey kizwan, try this version:
Attached File  OsxAptioFixDrv.efi.zip ( 13.43K ) Number of downloads: 24

I have changed DebugLog a bit - it will be a bit shorter and should stay after your reboot ... I hope.
HW: Asus P8P67-M, Intel Core i5-2300, 4GB, XFX HD-567X-ZHH3 SW: SL, L, ML: Clover UEFI boot
dmazar
post Jun 7 2012, 08:43 PM
Post #237
QUOTE (Chumnick @ Jun 7 2012, 06:56 PM) *
sending my reports according to your requirements:
DebugLog looks fine. Is this UUID the only problem you have?

There is a platform-uuid NVRAM variable that Clover should set to your CustomUUID, but in your case it is some other value. Can you boot into efi shell and check if maybe platform-uuid exists in NVRAM? In shell:
> dmpstore platform-uuid
Should not be there, but if exists, delete it with:
> dmpstore -d platform-uuid

You can also delete DebugLog from there:
> dmpstore -d DebugLog
Or from OSX terminal:
> sudo nvram -d DebugLog

If above does not solve the problem (deleting platform-uuid), send me please DarwinDump/DMIDump and IORegDump folders - just want to see what is written where.
HW: Asus P8P67-M, Intel Core i5-2300, 4GB, XFX HD-567X-ZHH3 SW: SL, L, ML: Clover UEFI boot
Chumnick
post Jun 7 2012, 09:28 PM
Post #238
QUOTE (dmazar @ Jun 7 2012, 08:43 PM) *
DebugLog looks fine. Is this UUID the only problem you have?

There is a platform-uuid NVRAM variable that Clover should set to your CustomUUID, but in your case it is some other value. Can you boot into efi shell and check if maybe platform-uuid exists in NVRAM? In shell:
> dmpstore platform-uuid
Should not be there, but if exists, delete it with:
> dmpstore -d platform-uuid

You can also delete DebugLog from there:
> dmpstore -d DebugLog
Or from OSX terminal:
> sudo nvram -d DebugLog

If above does not solve the problem (deleting platform-uuid), send me please DarwinDump/DMIDump and IORegDump folders - just want to see what is written where.

Yes, only UUID problem!
There is no platform-uuid in NVRAM.
Attached File  DMIDump.zip ( 4.7K ) Number of downloads: 5

[attachment=4490:IORegDump.zip]
[attachment=4491:IMG_0015_1.JPG]
ASUS P8Z68 Deluxe/Gen3, Intel i7-2600k, 16Gb 1133Mh DDR3, Sapphire Radeon HD 6870 1024Mb, OSZ Vertex3 SSD 240Gb, MacOS X 10.8, Clover, iMac 12,2
Gigabyte GA-Z77X-UP5 TH, Intel i7 3770k, 16Gb 1600Mh, ASUS GeForce GTX 670 2048 MB, Intel SSD 120Gb, MacOS X 10.8, Clover, iMac 12,2
dmazar
post Jun 7 2012, 10:02 PM
Post #239
QUOTE (Chumnick @ Jun 7 2012, 11:28 PM) *
There is no platform-uuid in NVRAM.
Ah, shell2. This will not be enough there. You see, apart from name and value, every NVRAM variable has some GUID. This GUID is like some kind of a folder or category. GUID + variable name is unique. If you do not specify guid in dmpstore in shell2, then it assumes some default GUID - meaning, the same variable with different GUID will not be printed/deleted. DebugLog and platform-uuid have different GUID then that default one.

To simplify, delete them from OSX terminal:
> sudo nvram -d DebugLog
> sudo nvram -d platform-uuid

and then reboot.

Or, boot into older shell - comes with Clover as /EFI/tools/shellX64.efi - you can start it from Clover menu. Commands from previous post will work there.

To finish this NVRAM platform-uuid stuff - please do in efi shell (old one, from Clover):
> dmpstore > vars.txt
or shell2:
> dmpstore -all > vars.txt
and then check vars.txt if it still contains DebugLog or platform-uuid. I will not be convinced that this variable does not exists until I see this var.txt.
HW: Asus P8P67-M, Intel Core i5-2300, 4GB, XFX HD-567X-ZHH3 SW: SL, L, ML: Clover UEFI boot
Chumnick
post Jun 7 2012, 10:41 PM
Post #240
QUOTE (dmazar @ Jun 7 2012, 10:02 PM) *
Or, boot into older shell - comes with Clover as /EFI/tools/shellX64.efi - you can start it from Clover menu. Commands from previous post will work there.

To finish this NVRAM platform-uuid stuff - please do in efi shell (old one, from Clover):
> dmpstore > vars.txt

and then check vars.txt if it still contains DebugLog or platform-uuid. I will not be convinced that this variable does not exists until I see this var.txt.

DMAZAR,
Thank you very much! I am with my UUID!
[attachment=4492:IMG_1313.JPG]
But I can not find vars.txt. And tell me please, why do I have these symptoms?
I am very grateful to you!

ASUS P8Z68 Deluxe/Gen3, Intel i7-2600k, 16Gb 1133Mh DDR3, Sapphire Radeon HD 6870 1024Mb, OSZ Vertex3 SSD 240Gb, MacOS X 10.8, Clover, iMac 12,2
Gigabyte GA-Z77X-UP5 TH, Intel i7 3770k, 16Gb 1600Mh, ASUS GeForce GTX 670 2048 MB, Intel SSD 120Gb, MacOS X 10.8, Clover, iMac 12,2

54 Pages V  « < 10 11 12 13 14 > » 
Add Reply Start a new topic
2 User(s) are reading this topic (2 Guests and 0 Anonymous Users)
0 Members: