Clover General discussion

Micky1979 · January 24, 2017

ok, time for me to update Pandora. Thanks @polyzaregone for testing with me

working 10.12.3, 10.12.4. https://applelife.ru/threads/clover.42089/page-694#post-639537

thanks for testing. Can you add the link to insanelymac to that "tiny" patch on AppleLife? thanks

polyzargone · January 24, 2017

Thanks to you for your responsiveness :thumbsup_anim: !

PMheart · January 25, 2017

ok, the time to compile Clover with this:

UINT8 KBESieSearchEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0xEB, 0x05, 0xE8, 0x9E, 0x05, 0x00, 0x00 };

UINT8 KBESieReplaceEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0x90, 0x90, 0xE8, 0x9E, 0x05, 0x00, 0x00 }

//UINT8 KBESieSearchEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0xEB, 0x05, 0xE8, 0x7E, 0x05, 0x00, 0x00 };

//UINT8 KBESieReplaceEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0x90, 0x90, 0xE8, 0x7E, 0x05, 0x00, 0x00 };

2 minutes..

Warning: this will fail to boot extensions in 10.12, 10.12.1, 10.12.2 and 10.12.3 from Clover dir. You should try to boot only 10.12.4 from usb. If works later will be udjusted

Hi. Why not delete 'e8 9e 05 00 00'? The only thing we should do is 'eb 05' ==> '90 90'. And I can assure that only one set of this data can be found.

BTW. Tested. KernelPm/KernelCPU/KernelLapic/AppleRTC/FakeCPUID will still work.

Sherlocks · January 25, 2017

@Micky1979

nice work. thank you

@Slice i cleaned up kext_inject.c.

we dont need

+// Sierra debug kernel
+UINT8 KBESieDebugSearchEXT[] = { 0xE8, 0x47, 0x00, 0x00, 0x00, 0xE9, 0x09, 0x00, 0x00, 0x00, 0x48, 0x8B, 0x7D, 0xE8, 0xE8, 0xD9 };
+UINT8 KBESieDebugReplaceEXT[] = { 0xE8, 0x47, 0x00, 0x00, 0x00, 0x90, 0x90, 0x00, 0x00, 0x00, 0x48, 0x8B, 0x7D, 0xE8, 0xE8, 0xD9 };
+UINT8 KBESieDebugSearchSIP[] = { 0x31, 0xC9, 0x39, 0xC1, 0x0F, 0x85, 0x3C, 0x00, 0x00, 0x00, 0x48, 0x8B, 0x85, 0xF8, 0xFE, 0xFF };
+UINT8 KBESieDebugReplaceSIP[] = { 0x31, 0xC9, 0x39, 0xC1, 0xEB, 0x80, 0x90, 0x90, 0x90, 0x90, 0x48, 0x8B, 0x85, 0xF8, 0xFE, 0xFF };
+
+// Sierra develop kernel
+UINT8 KBESieDevelSearchEXT[] = { 0xE8, 0x47, 0x00, 0x00, 0x00, 0xE9, 0x09, 0x00, 0x00, 0x00, 0x48, 0x8B, 0x7D, 0xE8, 0xE8, 0xD9 };
+UINT8 KBESieDevelReplaceEXT[] = { 0xE8, 0x47, 0x00, 0x00, 0x00, 0x90, 0x90, 0x00, 0x00, 0x00, 0x48, 0x8B, 0x7D, 0xE8, 0xE8, 0xD9 };
+UINT8 KBESieDevelSearchSIP[] = { 0x31, 0xC9, 0x39, 0xC1, 0x0F, 0x85, 0x3C, 0x00, 0x00, 0x00, 0x48, 0x8B, 0x85, 0xF8, 0xFE, 0xFF };
+UINT8 KBESieDevelReplaceSIP[] = { 0x31, 0xC9, 0x39, 0xC1, 0xEB, 0x80, 0x90, 0x90, 0x90, 0x90, 0x48, 0x8B, 0x85, 0xF8, 0xFE, 0xFF };

it's same hex code. we have to choose one(just prefer to choose name name. not important name)

clover always enter + else if (NumSieDebug == 1) {

+ else if (NumSieDebug == 1) {
+ Num = SearchAndReplace(Kernel, KERNEL_MAX_SIZE, KBESieDebugSearchEXT, sizeof(KBESieDebugSearchEXT), KBESieDebugReplaceEXT, 1) +
+ SearchAndReplace(Kernel, KERNEL_MAX_SIZE, KBESieDebugSearchSIP, sizeof(KBESieDebugSearchSIP), KBESieDebugReplaceSIP, 1);
+ DBG_RT(Entry, "==> kernel Sierra Debug: %d replaces done.\n", Num);
+ }
+ else if (NumSieDevel == 1) {
+ Num = SearchAndReplace(Kernel, KERNEL_MAX_SIZE, KBESieDevelSearchEXT, sizeof(KBESieDevelSearchEXT), KBESieDevelReplaceEXT, 1) +
+ SearchAndReplace(Kernel, KERNEL_MAX_SIZE, KBESieDevelSearchSIP, sizeof(KBESieDevelSearchSIP), KBESieDevelReplaceSIP, 1);
+ DBG_RT(Entry, "==> kernel Sierra Development: %d replaces done.\n", Num);
+ }
+

never enter + else if (NumSieDevel == 1) {.

because same hex code.

if one of hex code has difference, i never mention it now

i remove this one

and added 10.12.4+ hex code by micky

thank you

kext_inject.c-clean2.zip

PMheart · January 25, 2017

@Micky1979

nice work. thank you

@Slice i cleaned up kext_inject.c.

we dont need

it's same hex code. we have to choose one(just prefer to choose name name. not important name)

clover always enter + else if (NumSieDebug == 1) {

never enter + else if (NumSieDevel == 1) {.

because same hex code.

i remove this one

and added 10.12.4+ hex code by micky

thank you

I think there's a typo...

KBESie4*EXT is for 10.12.0 - 10.12.3, and KBESie*EXT is for 10.12.4+... You can check the hex code.

(Previously it's 'e8 7e' and now it's 'e8 9e'.)

BTW. I think the code for KBESieDebugReplaceEXT is wrong. Only NOP out 'e9 09' is not enough. Actually the 'e9 09 00 00 00' is the integrated jmpq and thus we should fill it with '90 90 90 90 90' (5x NOPs).

Edit: Here is the fixed one..

Edited January 25, 2017 by PMheart

Sherlocks · January 25, 2017

I think there's a typo...

KBESie4*EXT is for 10.12.0 - 10.12.3, and KBESie*EXT is for 10.12.4+... You can check the hex code.

(Previously it's 'e8 7e' and now it's 'e8 9e'.)

Edit: Here is the fixed one..

new_kext_inject.c.zip

i take a micky code. your are right. i checked old hex code. i changed kext_inject.c now

PMheart · January 25, 2017

i take a micky code. your are right. i checked old hex code. i changed kext_inject.c now

I updated it. Check above.

Sherlocks · January 25, 2017

I updated it. Check above.

checked. you just did change hex order. "(Previously it's 'e8 7e' and now it's 'e8 9e'.)"

UINT8 KBESieSearchEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0xEB, 0x05, 0xE8, 0x9E, 0x05, 0x00, 0x00 };

UINT8 KBESieReplaceEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0x90, 0x90, 0xE8, 0x9E, 0x05, 0x00, 0x00 };

UINT8 KBESie4SearchEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0xEB, 0x05, 0xE8, 0x7E, 0x05, 0x00, 0x00 };

UINT8 KBESie4ReplaceEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0x90, 0x90, 0xE8, 0x7E, 0x05, 0x00, 0x00 };

to

UINT8 KBESieSearchEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0xEB, 0x05, 0xE8, 0x7E, 0x05, 0x00, 0x00 };

UINT8 KBESieReplaceEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0x90, 0x90, 0xE8, 0x7E, 0x05, 0x00, 0x00 };

UINT8 KBESie4SearchEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0xEB, 0x05, 0xE8, 0x9E, 0x05, 0x00, 0x00 };

UINT8 KBESie4ReplaceEXT[] = { 0xE8, 0x25, 0x00, 0x00, 0x00, 0x90, 0x90, 0xE8, 0x9E, 0x05, 0x00, 0x00 };

it's not important order. but it help more clear.

thank you

kext_inject.c-clean2.zip

PMheart · January 25, 2017

checked. you just did change hex order. "(Previously it's 'e8 7e' and now it's 'e8 9e'.)"

to

it's not important order. but it help more clear.

thank you

Also check the Debug section. Only NOP out part of a jmpq is not enough.

Sherlocks · January 25, 2017

Also check the Debug section. Only NOP out part of a jmpq is not enough.

if i have develop kernel, i can surely check. but i can't surely mention it now.

PMheart · January 25, 2017

if i have develop kernel, i can surely check. but i can't surely mention it now.

Right. I don't have debugging kernel as well. But the issue is on the replacement/patch. Just fix it.

cecekpawon · January 25, 2017

Please distribute new unstable kernel patch version pattern via "KernelToPatch" (this features exists for this kind of purposes) and not into source, they will be junk someday IMHO

EDIT: Im trying to patch Smbios Table134 days ago to store smc revision. They exist in header & seems didnt do anything. Am I doing wrong / this completely unnecessary?

Edited January 25, 2017 by cecekpawon

abdouoi · January 25, 2017

After this update my backlight control stopped working. Am I the only one?

Slice · January 25, 2017

Please distribute new unstable kernel patch version pattern via "KernelToPatch" (this features exists for this kind of purposes) and not into source, they will be junk someday IMHO

Yes!

	<key>KernelAndKextPatches</key>
	<dict>
		<key>KernelToPatch</key>
		<array>
			<dict>
				<key>Comment</key>
				<string>Kext inject allowed</string>
				<key>MatchOS</key>
				<string>10.12.4</string>
				<key>Find</key>
				<data>6CUAAADrBeieBQAA</data>
				<key>Replace</key>
				<data>6CUAAACQkOieBQAA</data>
			</dict>
		</array>

PMheart · January 25, 2017

Why not consider reducing the search/replace patterns?

'e8 25 00 00 00 eb 05 e8' to 'e8 25 00 00 00 90 90 e8'

This keeps the patch unique. And IMHO it will not cause confusions... (Patch more than 1 time)

Micky1979 · January 25, 2017

too short IMHO.

but ...we have the "decode kernel" function in Clover?

Slice · January 25, 2017

too short IMHO.

but ...we have the "decode kernel" function in Clover?

No

SavageAUS · January 25, 2017

Rev 3879

Skylake support, by goodwin_c.
	<key>CPU</key>
	<dict>
		<key>HWPEnable</key>
		<true/>
	</dict>
This variable switches on Intel Speed Shift technology.

Also named Hardware-Controlled Performance States

Present on MacBook9,1

Is this needed to be set on skylake if using an ssdt?

Does checking the box (clover configurator) enable this feature? or disable it?

Never mind, trialled it and setting the checkbox set my CPU @ 4ghz and stayed at that.

Micky1979 · January 25, 2017

i take a micky code. your are right. i checked old hex code. i changed kext_inject.c now

Please distribute new unstable kernel patch version pattern via "KernelToPatch" (this features exists for this kind of purposes) and not into source, they will be junk someday IMHO

EDIT: Im trying to patch Smbios Table134 days ago to store smc revision. They exist in header & seems didnt do anything. Am I doing wrong / this completely unnecessary?

No

pity.

Guys why not like this?

else if (NumSie == 1) {

Num = SearchAndReplace(Kernel, KERNEL_MAX_SIZE, KBESieSearchSIP, sizeof(KBESieSearchSIP), KBESieReplaceSIP, 1);

if ((Num +1) != (Num + SearchAndReplace(Kernel, KERNEL_MAX_SIZE, KBESieSearchEXT, sizeof(KBESieSearchEXT), KBESieReplaceEXT, 1))) {

// this is for 10.12.4 beta and may (or not) be changed.

// to avoid an additional search that takes time (even if little), do that only if SearchAndReplace return 0

Num = Num + SearchAndReplace(Kernel, KERNEL_MAX_SIZE, KBESie4SearchEXT, sizeof(KBESie4SearchEXT), KBESie4ReplaceEXT, 1);

// but if the patch will be confirmed, swap KBESie4* with KBESie* to be faster in newer version of Sierra!

}

else

{

Num ++;

}

DBG_RT(Entry, "==> kernel Sierra: %d replaces done.\n", Num);

}

will speed up the process (not tested, and off course if SearchAndReplace respect the MaxReplaces given)

Micky1979 · January 26, 2017

but ...we have the "decode kernel" function in Clover?

No

this is the same: Get_PreLink() ...should be a good place to dinamically patching the kernel (LC_SEGMENT/LC_SEGMENT_64), since we already iterates through the kernel in that function.

Everythings needs is already in loader.h

bronxteck · January 26, 2017

3989 seems to compile fine using edk 23831 after removing conflicting file EfiShell.h.

Booted fine on haswell integrated graphics laptop with beta 10.12.4 also

used mick script with latest edk rev edit in it

FredWst · January 27, 2017

Hi,

Rev 3990 working fine.

I've a question I install a fresh Sierra 10.12.3, choose my language French, at second stage install goes in English ?

Is-it a Clover bug ?

Fred

arsradu · January 27, 2017

Hi guys,

Has anyone tried 10.12.4 PB1 yet?

Does it boot for you?

I'm on Clover 3974 and can't boot anymore after upgrading to that Beta. Good thing it's on a different partition.

Update:

Nvm that. I managed to boot using the patch posted by Slice above. Million thanks!

update 2: for anyone wondering, Clover 3994 injection now works fine without the patch above. So if you haven't, you should probably update.

Slice · January 27, 2017

this is the same: Get_PreLink() ...should be a good place to dinamically patching the kernel (LC_SEGMENT/LC_SEGMENT_64), since we already iterates through the kernel in that function.

Everythings needs is already in loader.h

Take into account that we don't know where the kernel is loaded. It is Chameleon who knows.

tluck · January 27, 2017

Happy new year! thank you all!

reporting good news...

clover 3994 and edk2 23831 all working good here

for 10.12.3 and 10.12.4 pb1

on Lenovo T420 UEFI and Lenovo T460 UEFI+EmuVariable

Clover General discussion

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites