iBOOT Source - http://iboot-efi.svn.sourceforge.net/viewvc/iboot-efi/
Click to view attachment
Full Version: Iboot Source Released
QUOTE (SunKi @ Apr 19 2011, 11:18 AM) 
for what boot.efi should have access to variables such as BootOrder?
Ninja uploaded not all his code: for example, �SMBIOS_STRUCTURE_POINTER� has no member named �Type131�.
So, he is using patched headers.
Ninja makes stupid mistakes: = instead of ==
AsmCpuid may not work, it's possibly 64-bit specific. But will test.
So many warnings treated as errors... All because he is working in Windows which ignores them.
Crazy check: he checks if BMP image size is less than... 4 Gigabytes!!!
Compiled, now linking errors (memcpy and udivdi3)
Compiled and linked iBoot, now testing (though some things became unsupported such as PNG images and... all CPUs higher than Core 2!)
PNG is because of memcpy (I think it's called in some structure assignment), and Core iX because of udivdi3 - 64bit unsigned division on 32bit CPUs, I can find it only for Atom).
PNG is because of memcpy (I think it's called in some structure assignment), and Core iX because of udivdi3 - 64bit unsigned division on 32bit CPUs, I can find it only for Atom).
Started, but nothing except grey screen - it has no textmode and its GUI is PNG-based)
Ujen woke up and created branch structure.
Took memcpy from Apple.
Compiled!
Ujen woke up and created branch structure.
Took memcpy from Apple.
Compiled!
Click to view attachmentLION BOOTED ON A 32BIT ATOM NETBOOK WITH IBOOT RUNNING ON DUET!!!!!!!!!!!!!!!!!!!
QUOTE (гык-sse2 @ Apr 19 2011, 02:17 PM)
Took memcpy from Apple.
What namely?
I can't find what assignment compiled into _memcpy.
About udiv it is here
(gCPUStructure.TSCFrequency / gCPUStructure.BusRatioMax);
Just comment out. FSBFrequency already known.
QUOTE (гык-sse2 @ Apr 19 2011, 02:59 PM)
LION BOOTED ON A 32BIT ATOM NETBOOK WITH IBOOT RUNNING ON DUET!!!!!!!!!!!!!!!!!!!
it's good news...
QUOTE (гык-sse2 @ Apr 19 2011, 02:59 PM)
LION BOOTED ON A 32BIT ATOM NETBOOK WITH IBOOT RUNNING ON DUET!!!!!!!!!!!!!!!!!!!
Congratulation! You are faster then me.
I already find what's new in iBoot compared to Clover. Thanks to Ninja!
QUOTE (Slice @ Apr 19 2011, 03:06 PM)
What namely?
I can't find what assignment compiled into _memcpy.
About udiv it is here
(gCPUStructure.TSCFrequency / gCPUStructure.BusRatioMax);
Just comment out. FSBFrequency already known.
I can't find what assignment compiled into _memcpy.
About udiv it is here
(gCPUStructure.TSCFrequency / gCPUStructure.BusRatioMax);
Just comment out. FSBFrequency already known.
I also couldn't, but the problem was solved by adding this function to picopng.c (with license).
It's already known by whom?byOsxSmbiosGen? then I'd comment not only this line, because iBoot overrides some part of our patch.
QUOTE (Slice @ Apr 19 2011, 03:10 PM)
Congratulation! You are faster then me.
I already find what's new in iBoot compared to Clover. Thanks to Ninja!
I already find what's new in iBoot compared to Clover. Thanks to Ninja!
Model, SystemSerialNumber, DevicePathsSupported and system-id?
CODE
00002a27 movl %eax,0x00000118(%esp)
00002a2e movl %eax,0x0000011c(%esp)
00002a35 movl %edx,%eax
00002a37 shrl $0x03,%eax
00002a3a movl %eax,0x00000120(%esp)
00002a41 shrl $0x02,%edx
00002a44 movl %edx,0x00000124(%esp)
00002a4b movl %ecx,%eax
00002a4d shrl $0x02,%eax
00002a50 movl %eax,0x00000128(%esp)
00002a57 shrl %ecx
00002a59 movl %ecx,0x0000012c(%esp)
00002a60 shrl %ebx
00002a62 movl %ebx,0x00000130(%esp)
00002a69 movl $_png_alloc_find_node,0x000000fc(%esp)
00002a74 movl $_png_alloc_find_node,0x00000100(%esp)
00002a7f movl $_png_alloc_find_node,0x00000104(%esp)
00002a8a movl $_png_alloc_find_node,0x00000108(%esp)
00002a95 movl $_png_alloc_find_node,0x0000010c(%esp)
00002aa0 movl $_png_alloc_find_node,0x00000110(%esp)
00002aab movl $_png_alloc_find_node,0x00000114(%esp)
00002ab6 leal 0x0000008c(%esp),%esi
00002abd movl $0x00000070,0x08(%esp)
00002ac5 movl $_C.54.15003,0x04(%esp)
00002acd movl %esi,(%esp)
00002ad0 calll _memcpy+0x100000000
00002ad5 xorl %ecx,%ecx
00002ad7 leal 0x000000fc(%esp),%ebp
00002ade incl %ecx
00002a2e movl %eax,0x0000011c(%esp)
00002a35 movl %edx,%eax
00002a37 shrl $0x03,%eax
00002a3a movl %eax,0x00000120(%esp)
00002a41 shrl $0x02,%edx
00002a44 movl %edx,0x00000124(%esp)
00002a4b movl %ecx,%eax
00002a4d shrl $0x02,%eax
00002a50 movl %eax,0x00000128(%esp)
00002a57 shrl %ecx
00002a59 movl %ecx,0x0000012c(%esp)
00002a60 shrl %ebx
00002a62 movl %ebx,0x00000130(%esp)
00002a69 movl $_png_alloc_find_node,0x000000fc(%esp)
00002a74 movl $_png_alloc_find_node,0x00000100(%esp)
00002a7f movl $_png_alloc_find_node,0x00000104(%esp)
00002a8a movl $_png_alloc_find_node,0x00000108(%esp)
00002a95 movl $_png_alloc_find_node,0x0000010c(%esp)
00002aa0 movl $_png_alloc_find_node,0x00000110(%esp)
00002aab movl $_png_alloc_find_node,0x00000114(%esp)
00002ab6 leal 0x0000008c(%esp),%esi
00002abd movl $0x00000070,0x08(%esp)
00002ac5 movl $_C.54.15003,0x04(%esp)
00002acd movl %esi,(%esp)
00002ad0 calll _memcpy+0x100000000
00002ad5 xorl %ecx,%ecx
00002ad7 leal 0x000000fc(%esp),%ebp
00002ade incl %ecx
Just a few minutes and I'll upload the code.
It is here
Compiler decided to unroll the cycle and optimize to memcpy!
CODE
for (i = 0; i < 7; i++)
PNG_adam7Pass(out_data, scanlinen->data, scanlineo->data, &scanlines->data[passstart[i]],
info->width, pattern[i], pattern[i + 7], pattern[i + 14], pattern[i + 21],
passw[i], passh[i], bpp);
PNG_adam7Pass(out_data, scanlinen->data, scanlineo->data, &scanlines->data[passstart[i]],
info->width, pattern[i], pattern[i + 7], pattern[i + 14], pattern[i + 21],
passw[i], passh[i], bpp);
Compiler decided to unroll the cycle and optimize to memcpy!
QUOTE (гык-sse2 @ Apr 19 2011, 03:24 PM)
Model, SystemSerialNumber, DevicePathsSupported and system-id?
DataHub usage!
QUOTE (Slice @ Apr 19 2011, 03:33 PM)
DataHub usage!
Weren't we using datahub in VBoxAppleSim?
iNifiga Ne!
Lion boots, Snow Leopard panics.
Maybe I deleted smth important together with GetCpuSomething.
Click to view attachment
I always thought that Lion stalling and KPs are one problem, because if iPhoneTom solved only KP, then Lion wouldn't boot on XPC 0.50.
It's good that I haven't removed Chameleon))
Is there a way to enable Time Machine for a folder?
Lion boots, Snow Leopard panics.
Maybe I deleted smth important together with GetCpuSomething.
Click to view attachment
I always thought that Lion stalling and KPs are one problem, because if iPhoneTom solved only KP, then Lion wouldn't boot on XPC 0.50.
It's good that I haven't removed Chameleon))
Is there a way to enable Time Machine for a folder?
Some correction
And no more _memcpy!
Do not hurry! Nobody wants to run faster then you. We should look carefully what iBoot do and what not.
As I previously said there are two parts of The Bootloader. One is EFI image and other is GUI. Before this moment I proposed to rewrite rEFIt to be my GUI.
And many patches must be in GUI but not in EFI.
iBoot is fine for this purpose. But there is also a work to adopt one for other.
CODE
const UINT32 pattern[28] = { 0, 4, 0, 2, 0, 1, 0, 0, 0, 4, 0, 2, 0, 1, 8, 8, 4, 4, 2, 2, 1, 8, 8,
8, 4, 4, 2, 2 }; // values for the adam7 passes
....
UINT32 passstart[7]; //= { 0 };
for (i=0; i<7; i++) {
passstart[i] = 0;
}
// UINT32 pattern[28] = { 0, 4, 0, 2, 0, 1, 0, 0, 0, 4, 0, 2, 0, 1, 8, 8, 4, 4, 2, 2, 1, 8, 8,
// 8, 4, 4, 2, 2 }; // values for the adam7 passes
8, 4, 4, 2, 2 }; // values for the adam7 passes
....
UINT32 passstart[7]; //= { 0 };
for (i=0; i<7; i++) {
passstart[i] = 0;
}
// UINT32 pattern[28] = { 0, 4, 0, 2, 0, 1, 0, 0, 0, 4, 0, 2, 0, 1, 8, 8, 4, 4, 2, 2, 1, 8, 8,
// 8, 4, 4, 2, 2 }; // values for the adam7 passes
And no more _memcpy!
Do not hurry! Nobody wants to run faster then you. We should look carefully what iBoot do and what not.
As I previously said there are two parts of The Bootloader. One is EFI image and other is GUI. Before this moment I proposed to rewrite rEFIt to be my GUI.
And many patches must be in GUI but not in EFI.
iBoot is fine for this purpose. But there is also a work to adopt one for other.
Error loading theme.plist 
QUOTE (Slice @ Apr 19 2011, 04:46 PM)
Error loading theme.plist
Click to view attachment
Take efi folder from this archive
QUOTE (Slice @ Apr 19 2011, 04:23 PM)
And many patches must be in GUI but not in EFI.
I think that patches should be separate modules, loaded from any GUI or shell. Like U90Acpi.
QUOTE (гык-sse2 @ Apr 19 2011, 05:30 PM)
I think that patches should be separate modules, loaded from any GUI or shell. Like U90Acpi.
Yes, I mean exactly that.
I booted Snow without problems. But USB cried in kernel,log. All works.
CODE
| +-o platform <class IOService, id 0x10000010e, !registered, !matched, active, busy 0, retain 4>
| {
| "SystemSerialNumber" = <47003800570036004e0032004a000000>
| "Model" = <4c00610074006900740075006400650020004400340032003000200020002000200020002000
20002000200020002000200020002000200020002000200020000000>
| "kext" = <>
| "iboot" = <>
| "name" = <"platform">
| "TSCFrequency" = <80de893f00000000>
| "FSBFrequency" = <406bed0700000000>
| "CPUFrequency" = <80de893f00000000>
| "DevicePathsSupported" = <01000000>
| "system-id" = <0000000000000010800000508d9f42f0>
| }
| {
| "SystemSerialNumber" = <47003800570036004e0032004a000000>
| "Model" = <4c00610074006900740075006400650020004400340032003000200020002000200020002000
20002000200020002000200020002000200020002000200020000000>
| "kext" = <>
| "iboot" = <>
| "name" = <"platform">
| "TSCFrequency" = <80de893f00000000>
| "FSBFrequency" = <406bed0700000000>
| "CPUFrequency" = <80de893f00000000>
| "DevicePathsSupported" = <01000000>
| "system-id" = <0000000000000010800000508d9f42f0>
| }
New system-id and new hardwareID in profiler.
Continue testing.
Same KP.
Now boot Lion. Yes, it started. Gone through xpchelper. Next message was "journal replay done" so error with previous booter was something like "still waiting for ....".
I looked at Lion installer until "10.7 can't be installed on this computer". My computer is "MacBook1,1" with Yonah CPU.
I will try again later with SMBIOS changed to MacBookPro5,1 to see if it does matter.
QUOTE (Slice @ Apr 19 2011, 08:18 PM)
Now boot Lion. Yes, it started. Gone through xpchelper. Next message was "journal replay done" so error with previous booter was something like "still waiting for ....".
I looked at Lion installer until "10.7 can't be installed on this computer". My computer is "MacBook1,1" with Yonah CPU.
I will try again later with SMBIOS changed to MacBookPro5,1 to see if it does matter.
I looked at Lion installer until "10.7 can't be installed on this computer". My computer is "MacBook1,1" with Yonah CPU.
I will try again later with SMBIOS changed to MacBookPro5,1 to see if it does matter.
I also had this message. Mine is MacBook4,1.
I don't remember how did I solve it on my desktop. But it was some smbios info too (this was only when installing from Snow).
Seems like both SystemSerialNumber and Model are in the wrong format, should be like:
"Model" = <"M","a","c","B","o","o","k","P","r","o","8",",","3">
aka: typedef int16_t EFI_CHAR16;
"Model" = <"M","a","c","B","o","o","k","P","r","o","8",",","3">
aka: typedef int16_t EFI_CHAR16;
QUOTE (гык-sse2 @ Apr 19 2011, 10:21 PM)
I also had this message. Mine is MacBook4,1.
I don't remember how did I solve it on my desktop. But it was some smbios info too (this was only when installing from Snow).
I don't remember how did I solve it on my desktop. But it was some smbios info too (this was only when installing from Snow).
As I remember you said something about s/n. It should correspond to board-id. May be I am wrong.
I am comparing your corrections to ninja with my (I did the same work at the same time
). Agree with your excluding patches that we already have.Why you didn't correct warnings about "variable xxx may be used uninitialized"? For me compilation stopped.
For example here
CODE
EFI_STATUS Status = EFI_SUCCESS; //my init
and more dangerous here
CODE
VOID SetupKextload(EFI_HANDLE bootVolume)
{
BOOTVOLUME_DATA* volume; // uninitialized
EFI_DEVICE_PATH_PROTOCOL* DevicePath;
//Slice - dunno what to do
volume = AllocatePool(sizeof(BOOTVOLUME_DATA));
//
// get DevicePath from Selected Volume
DevicePath=DevicePathFromHandle(bootVolume);
// get System Version from DevicePath
FindVersionPlist(DevicePath,volume->Version); //using volume !!!
{
BOOTVOLUME_DATA* volume; // uninitialized
EFI_DEVICE_PATH_PROTOCOL* DevicePath;
//Slice - dunno what to do
volume = AllocatePool(sizeof(BOOTVOLUME_DATA));
//
// get DevicePath from Selected Volume
DevicePath=DevicePathFromHandle(bootVolume);
// get System Version from DevicePath
FindVersionPlist(DevicePath,volume->Version); //using volume !!!
About bmp size of 4Gb lol. I made the follow
CODE
if (BltBufferSize >= 1<<24) //SIZE_4GB) //Slice - not possible for arch32, reduce to 4Mpx x 32bit
About memcpy I said in previous post.
QUOTE (гык-sse2 @ Apr 19 2011, 02:31 AM)
Compiled and linked iBoot, now testing (though some things became unsupported such as PNG images and... all CPUs higher than Core 2!)
No wonder OSX lion would restart right at kernel start on my sandy bridge system. (I wonder why one user claimed that iboot did in fact work with his sandy bridge system however.) There is code in iboot's GetCPUProperties() that tries to identify sandy bridge core i7s, but it always mis-identified my system as 1 core, 1 thread, so it looks like that code hasn't been working right.QUOTE (bcc9 @ Apr 19 2011, 01:57 PM)
There is code in iboot's GetCPUProperties() that tries to identify sandy bridge core i7s, but it always mis-identified my system as 1 core, 1 thread, so it looks like that code hasn't been working right.
Digging deeper, from inspection, the bug is that the intelCPU variable in GetCPUProperties() is initialized from gCPUStructure.Model, but that field has not yet been initialized, so the value is garbage.
So the subsequent switch (intelCPU) logic doesn't match the CPU type correctly.
QUOTE (гык-sse2 @ Apr 19 2011, 12:00 PM)
iNifiga Ne!
Lion boots, Snow Leopard panics.
Maybe I deleted smth important together with GetCpuSomething.
Click to view attachment
I always thought that Lion stalling and KPs are one problem, because if iPhoneTom solved only KP, then Lion wouldn't boot on XPC 0.50.
It's good that I haven't removed Chameleon))
Is there a way to enable Time Machine for a folder?
Lion boots, Snow Leopard panics.
Maybe I deleted smth important together with GetCpuSomething.
Click to view attachment
I always thought that Lion stalling and KPs are one problem, because if iPhoneTom solved only KP, then Lion wouldn't boot on XPC 0.50.
It's good that I haven't removed Chameleon))
Is there a way to enable Time Machine for a folder?
How to compile this?
QUOTE (kocoman @ Apr 20 2011, 09:18 AM)
How to compile this?
To compile I included this line into clover's DuetPkgIa32.dsc
CODE
#
# Sample Application
#
#MdeModulePkg/Application/HelloWorld/HelloWorld.inf
# MdeModulePkg/Application/VariableInfo/VariableInfo.inf
iboot-efi/iboot.inf
# Sample Application
#
#MdeModulePkg/Application/HelloWorld/HelloWorld.inf
# MdeModulePkg/Application/VariableInfo/VariableInfo.inf
iboot-efi/iboot.inf
And search iboot.efi in Build folder.
QUOTE
Digging deeper, from inspection, the bug is that the intelCPU variable in GetCPUProperties() is initialized from gCPUStructure.Model, but that field has not yet been initialized, so the value is garbage.
So the subsequent switch (intelCPU) logic doesn't match the CPU type correctly.
So the subsequent switch (intelCPU) logic doesn't match the CPU type correctly.
I have more correct CPU identification in Clover. So I and usr-sse2 just excluded cpu.c from iboot.
QUOTE (Slice @ Apr 19 2011, 11:00 PM)
To compile I included this line into clover's DuetPkgIa32.dsc
And search iboot.efi in Build folder.
I have more correct CPU identification in Clover. So I and usr-sse2 just excluded cpu.c from iboot.
CODE
#
# Sample Application
#
#MdeModulePkg/Application/HelloWorld/HelloWorld.inf
# MdeModulePkg/Application/VariableInfo/VariableInfo.inf
iboot-efi/iboot.inf
# Sample Application
#
#MdeModulePkg/Application/HelloWorld/HelloWorld.inf
# MdeModulePkg/Application/VariableInfo/VariableInfo.inf
iboot-efi/iboot.inf
And search iboot.efi in Build folder.
I have more correct CPU identification in Clover. So I and usr-sse2 just excluded cpu.c from iboot.
So I added the above iboot.inf line to edk2/DuetPkg/cloverefiboot/DuetPkgIa32.dsc, and put usr-sse2's version of iboot into edk2/iboot-efi. Initially build32.sh fails as usr-sse2's SmBios.h doesn't match the #include "Smbios.h" in my case sensitive filesystem.
After I fixed that, I get a working Efildr32.
So I:
CODE
cat StartHFS.com efi32.com2 edk2/Build/CloverIA32/DEBUG_XCODE32/FV/Efildr32 > /Volumes/clover/boot
and I'm able to boot clover and start booting into OSX lion's install disk.
It hangs after "Loading kernel cache file ..." and "root device uuid is...", however.
I guess I'm lost as to how to make use of iboot with clover - I never saw any iboot-like menu of the disks to boot from...
Edit: if I copy the built iboot.efi to /boot/bootIA32.efi, then after clover starts bootIA32.efi, it drops back into the uefi bios menu after printing "Boot Failed. EFI USB Device". If I copy the shell .efi into bootIA32.efi, that shell efi code runs correctly.
iBoot is not started because of absent of support files. Take its from гык-sse2 post#20.
Ninja is wrong in cpu.c
AsciiStrnCpy(destination, source, length);
CODE
AsmCpuid(0x80000002, reg, NULL, NULL, NULL);
AsciiStrnCpy((CHAR8 *)reg, &str[0], 16);
AsciiStrnCpy((CHAR8 *)reg, &str[0], 16);
AsciiStrnCpy(destination, source, length);
QUOTE (bcc9 @ Apr 20 2011, 04:04 AM)
and put usr-sse2's version of iboot into edk2/iboot-efi.
I'm a bit dense this morning, but which version is гык-sse2's?
QUOTE (rammjet @ Apr 20 2011, 03:50 PM)
I'm a bit dense this morning, but which version is гык-sse2's?
This is my recent version. The folder is renamed to keep separate from original one.
Click to view attachment
QUOTE (Slice @ Apr 20 2011, 08:37 AM)
This is my recent version. The folder is renamed to keep separate from original one.
Click to view attachment
Click to view attachment
Thanks. I caught the "cl" name change. I also found гык-sse2's files in post 17.
QUOTE (гык-sse2 @ Apr 19 2011, 04:00 PM)
iNifiga Ne!
Lion boots, Snow Leopard panics.
Lion boots, Snow Leopard panics.
CODE
EFI_INTERFACE_DEVICE_PROPERTIES mDeviceProperties=
{
NULL,
NULL,
NULL,
NULL,
GetDeviceProperties
};
{
NULL,
NULL,
NULL,
NULL,
GetDeviceProperties
};
Remember our discussion about 5th and 6th.
Well, my luck is running about the same as before.
On the good side, I can run /edk2/Build/CloverIA32/DEBUG_XCODE32/IA32/iboot-cl/iboot/DEBUG/iboot.efi and get the GUI. This is using Slice's iboot files.
But when I try to boot something, I get the same old results: with DSDT.aml, in Clover, I get a black screen hang with cursor. Without a DSDT.aml, I get the ACPI KP.
On the good side, I can run /edk2/Build/CloverIA32/DEBUG_XCODE32/IA32/iboot-cl/iboot/DEBUG/iboot.efi and get the GUI. This is using Slice's iboot files.
But when I try to boot something, I get the same old results: with DSDT.aml, in Clover, I get a black screen hang with cursor. Without a DSDT.aml, I get the ACPI KP.
Why doesn't OS_Ninja post in this thread..
There's now iboot 1.7 in the irc, but it does not say what is new in the changelog..
I try to boot v1.7, but still get
The code would be
iboot's platformdata.c
Also, how to compile iboot into 64bit? I ask because I get
"Image type IA32 is not supported by this X64 shell"
Is it possible to compile a 64-bit Efildr20? (like the one that is working for my MSI board from the Hacklion.zip)
I tried to boot the iboot.efi from slice, but it does not load. I think my Efildr20 is 64bit, because it uses /efi/boot/bootx64.efi
When I used the boot file from the 0.4 script by Jadran, then put boot0, boot1h, and boot like "Chameleon". I only get the WELCOME TO EFI WORLD
How to solve these problems? Usually the WELCOME TO EFI WORLD should not be seen, on "successful" EfiLdr20's I get MBR Start, then GUI... If I load shell.efi, then type ver,
EFI Specification Revision: 2.10
EFI Vendor : EDK II
EFI Revision : 1.0
Can anyone suggest something?
There's now iboot 1.7 in the irc, but it does not say what is new in the changelog..
I try to boot v1.7, but still get
QUOTE
can't locate smbios protocol, not found
then
there were errors converting smbios system table : not found
then
there were errors converting smbios system table : not found
The code would be
iboot's platformdata.c
QUOTE
EFI_STATUS InjectPlatformData()
{
EFI_STATUS Status;
Status = gBootServices->LocateProtocol (&gEfiSmbiosProtocolGuid,NULL,(VOID **)&Smbios);
if (EFI_ERROR (Status)) {
CreateDesktop();
DrawLogo();
InitPointer();
ShowPointer();
SHOW_ON_ERROR(Status,"Cannot locate SMBIOS protocol!",STOP_ICON);
}
{
EFI_STATUS Status;
Status = gBootServices->LocateProtocol (&gEfiSmbiosProtocolGuid,NULL,(VOID **)&Smbios);
if (EFI_ERROR (Status)) {
CreateDesktop();
DrawLogo();
InitPointer();
ShowPointer();
SHOW_ON_ERROR(Status,"Cannot locate SMBIOS protocol!",STOP_ICON);
}
Also, how to compile iboot into 64bit? I ask because I get
"Image type IA32 is not supported by this X64 shell"
Is it possible to compile a 64-bit Efildr20? (like the one that is working for my MSI board from the Hacklion.zip)
QUOTE
iBOOT also comes in two flavor's: iBOOT 32bit and iBOOT 64bit. The reason being is to accomodate any version Intel based Apple� Computer and EFI compatible PC.
I tried to boot the iboot.efi from slice, but it does not load. I think my Efildr20 is 64bit, because it uses /efi/boot/bootx64.efi
When I used the boot file from the 0.4 script by Jadran, then put boot0, boot1h, and boot like "Chameleon". I only get the WELCOME TO EFI WORLD
How to solve these problems? Usually the WELCOME TO EFI WORLD should not be seen, on "successful" EfiLdr20's I get MBR Start, then GUI... If I load shell.efi, then type ver,
EFI Specification Revision: 2.10
EFI Vendor : EDK II
EFI Revision : 1.0
Can anyone suggest something?
QUOTE (гык-sse2 @ Apr 20 2011, 11:27 AM)
rammjet, change FDF and DSC files to use non-Clover GenericBdsLib, PlatformBdsLib, AcpiTable and exclude AcpiPlatform. This will undo Slice's ACPI patch.
Thanks. That got a different result, even though it wasn't success.
SL with no DSDT.aml: upon kernel load, immediate shutdown of the computer - completely off
SL with DSDT.aml: Kext loading disabled and immediate CPU halted messages
Lion with and without DSDT.aml: no HPETs message
Just for general info, it should be noted that although I normally boot SL with DSDT.aml, I have proven that I can do so without it. I normally boot Lion using XPC without DSDT.aml because with DSDT.aml I would not get VoodooPS2 to work. And I used to boot Lion with a version of iBoot without DSDT.aml but I had to remove kernelcaches every time or I would not get audio.
QUOTE (kocoman @ Apr 20 2011, 07:39 PM)
How to solve these problems? Usually the WELCOME TO EFI WORLD should not be seen, on "successful" EfiLdr20's I get MBR Start, then GUI...
iPhoneTom and other people just disable this WELCOME string.
I disabled also MBR Start! in "pre-Clover" on applelife.
Hm, OS_Ninja doesn't add me to project, although "it's open and everybody is welcome to create branches"...
QUOTE (Slice @ Apr 20 2011, 01:45 AM)
iBoot is not started because of absent of support files. Take its from гык-sse2 post#20.
Thanks for the pointer. I had thought it would start but just lack gui elements without the extra /efi files.
I tried the 1.2 efi config files, and it boots into iboot gui now. However most of the entries in the system menu are blank (like all the important ones such as cpu frequency, cpu bus speed, product name).
I instead put the 1.7 efi config files into /efi, and now those fields are filled in. However the cpu bus speed is still mis-identified as 66. Changing any values under the system menu seem to have no effect (the changes don't show up when you go back to the M menu).
If I proceed with booting into the lion DP2 install disk, I get a bit further than before, but it still hangs after displaying "EfiMode: 0".
So I see the iboot code just does rudimentary read&writes of efi/theme/config.bin to store the setting menu data. The data is just dumped directly from the GUI_MENU_DATA structure, and that structure keeps changing between versions. This would explain why it's not OK to just use the iboot 1.2 /efi directory with 1.7-based code.
I'm guessing that writing to config.bin fails because writing to hfs files doesn't work with the posted HFSPlusMP21.efi. Thus config changes don't stick.
I'm guessing that writing to config.bin fails because writing to hfs files doesn't work with the posted HFSPlusMP21.efi. Thus config changes don't stick.
For the 64bit EFI, I looked at the Conf/tools_def.txt, and it says
What is ABI ? Firmware? when will this problem be resolved? or do I need to compile in Visual studio 2005?
I tried to build anyway, but it says
I did change the
but build.py still complains
how to fix it?
QUOTE
##################
# X64 definitions - still a work in progress. This tool chain does not produce
# the correct ABI, it is just used to compile the code....
##################
# X64 definitions - still a work in progress. This tool chain does not produce
# the correct ABI, it is just used to compile the code....
##################
What is ABI ? Firmware? when will this problem be resolved? or do I need to compile in Visual studio 2005?
I tried to build anyway, but it says
QUOTE
build.py...
: error 2000: Invalid parameter
Active platform supports [IA32] only, but [X64] is given.
: error 2000: Invalid parameter
Active platform supports [IA32] only, but [X64] is given.
I did change the
QUOTE
Find the TARGET_ARCH definition in Conf/target.txt and change the corresponding line to match this
CODE
TARGET_ARCH = X64
CODE
TARGET_ARCH = X64
but build.py still complains
how to fix it?
QUOTE (rammjet @ Apr 20 2011, 12:03 PM)
Thanks. That got a different result, even though it wasn't success.
SL with no DSDT.aml: upon kernel load, immediate shutdown of the computer - completely off
SL with DSDT.aml: Kext loading disabled and immediate CPU halted messages
Lion with and without DSDT.aml: no HPETs message
Just for general info, it should be noted that although I normally boot SL with DSDT.aml, I have proven that I can do so without it. I normally boot Lion using XPC without DSDT.aml because with DSDT.aml I would not get VoodooPS2 to work. And I used to boot Lion with a version of iBoot without DSDT.aml but I had to remove kernelcaches every time or I would not get audio.
SL with no DSDT.aml: upon kernel load, immediate shutdown of the computer - completely off
SL with DSDT.aml: Kext loading disabled and immediate CPU halted messages
Lion with and without DSDT.aml: no HPETs message
Just for general info, it should be noted that although I normally boot SL with DSDT.aml, I have proven that I can do so without it. I normally boot Lion using XPC without DSDT.aml because with DSDT.aml I would not get VoodooPS2 to work. And I used to boot Lion with a version of iBoot without DSDT.aml but I had to remove kernelcaches every time or I would not get audio.
I managed to boot to Lion. Because of the HPET message, I removed the CPUManagement kext. During boot, it sat at IOAPIC for about a minute. I was ready to abort when it suddenly launched and took me to the Desktop. I should point out that I previously made the clover changes suggested by гык-sse2 in post 42.
QUOTE (bcc9 @ Apr 20 2011, 07:17 PM)
So I see the iboot code just does rudimentary read&writes of efi/theme/config.bin to store the setting menu data. The data is just dumped directly from the GUI_MENU_DATA structure, and that structure keeps changing between versions. This would explain why it's not OK to just use the iboot 1.2 /efi directory with 1.7-based code.
I'm guessing that writing to config.bin fails because writing to hfs files doesn't work with the posted HFSPlusMP21.efi. Thus config changes don't stick.
I'm guessing that writing to config.bin fails because writing to hfs files doesn't work with the posted HFSPlusMP21.efi. Thus config changes don't stick.
Smbios Patching is a work in progress. Yes, new menu entries will corrupt the bin file, so i suggest deleting it if your updating to a new release.
Also, smbios patching is only working on most genuine intel mobo's for now until i read more of the spec and try creating a new smbios struct with pointers and tables.
QUOTE (OS_Ninja @ Apr 21 2011, 10:38 AM)
Also, smbios patching is only working on most genuine intel mobo's for now until i read more of the spec and try creating a new smbios struct with pointers and tables.
? I thought iboot was successfully booting lion on many intel-chipset motherboards including those from gigabyte. This is as reported by users over in the main lion thread @insanelymac. Some users report success with iboot and sandy bridge gigabyte motherboards so I'm at a loss as to what is failing in my case.QUOTE (bcc9 @ Apr 20 2011, 04:35 PM)
Thanks for the pointer. I had thought it would start but just lack gui elements without the extra /efi files.
I tried the 1.2 efi config files, and it boots into iboot gui now. However most of the entries in the system menu are blank (like all the important ones such as cpu frequency, cpu bus speed, product name).
I instead put the 1.7 efi config files into /efi, and now those fields are filled in. However the cpu bus speed is still mis-identified as 66. Changing any values under the system menu seem to have no effect (the changes don't show up when you go back to the M menu).
If I proceed with booting into the lion DP2 install disk, I get a bit further than before, but it still hangs after displaying "EfiMode: 0".
I tried the 1.2 efi config files, and it boots into iboot gui now. However most of the entries in the system menu are blank (like all the important ones such as cpu frequency, cpu bus speed, product name).
I instead put the 1.7 efi config files into /efi, and now those fields are filled in. However the cpu bus speed is still mis-identified as 66. Changing any values under the system menu seem to have no effect (the changes don't show up when you go back to the M menu).
If I proceed with booting into the lion DP2 install disk, I get a bit further than before, but it still hangs after displaying "EfiMode: 0".
If your trying to boot lion and the bootargs efimode dumps a number other than EfiMode:64, you arent booting the newest version iBOOT.
I implemented the 64bit bootargs struct in newer versions, starting with 1.5
Also, use the config.bin file from topic in my irc channel, it is formatted to work with all the newest menu entries.
QUOTE (OS_Ninja @ Apr 21 2011, 05:16 PM)
If your trying to boot lion and the bootargs efimode dumps a number other than EfiMode:64, you arent booting the newest version iBOOT.
Actually I was using what I had compiled myself from the source code, so it's the current 1.7 version with slice/sse2 changes&fixes to work with the osx edk2 build environment. I think some of the code that they commented out in bootprocess.c is causing things to not be initialized right, I'll go try with some of that turned back on.QUOTE (OS_Ninja @ Apr 21 2011, 05:16 PM)
so, use the config.bin file from topic in my irc channel, it is formatted to work with all the newest menu entries.
I'm all set with the 1.7 config.bin binary. My point was to highlight how using the 1.2 config.bin binary breaks things, as that is what was recommended a few times in this thread.
QUOTE (bcc9 @ Apr 21 2011, 09:14 PM)
Actually I was using what I had compiled myself from the source code, so it's the current 1.7 version with slice/sse2 changes&fixes to work with the osx edk2 build environment. I think some of the code that they commented out in bootprocess.c is causing things to not be initialized right, I'll go try with some of that turned back on.
I'm all set with the 1.7 config.bin binary. My point was to highlight how using the 1.2 config.bin binary breaks things, as that is what was recommended a few times in this thread.
I'm all set with the 1.7 config.bin binary. My point was to highlight how using the 1.2 config.bin binary breaks things, as that is what was recommended a few times in this thread.
How are you booting iBOOT with a vanilla EDK2 Efildr20?
iBOOT doesnt read FAT binary headers...
I have this module in my edk2 source. Maybe I'll open my edk2 source soon.
Good Idea OS_Ninga
QUOTE (OS_Ninja @ Apr 22 2011, 07:36 AM)
How are you booting iBOOT with a vanilla EDK2 Efildr20?
iBOOT doesnt read FAT binary headers...
I have this module in my edk2 source. Maybe I'll open my edk2 source soon.
iBOOT doesnt read FAT binary headers...
I have this module in my edk2 source. Maybe I'll open my edk2 source soon.
Not vanilla edk2, but Clover, it reads fat modules fine. Also it's 32-bit.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.