crazybirdy Posted February 18, 2017 Share Posted February 18, 2017 So be sure your prelinked is rebuilted correctly after permissions are repaired. All right, you're right! Something wrong with prelinkedkernel, now it works fine with BCM4322, and both hex-patch and info-inject work fine with prelinkedkernel. Nice job. 2 Link to comment Share on other sites More sharing options...
Micky1979 Posted February 18, 2017 Share Posted February 18, 2017 All right, you're right! Something wrong with prelinkedkernel, now it works fine with BCM4322 and both hex-patch and info-inject work fine with prelinkedkernel. Nice job. Good! Also your ATI card works this way? ... please test adding an additional personality (I'm searching for bugs) 2) If I put my kexts in /S/L/E (and not in /L/E for obvious reasons) and want to boot without cache, why do I need KernelBooter_kexts set to Yes along -f flag? once your kexts are prelinked you don't need additional magic by the bootloader (the reason you put your kexts in System's dirs ?). -f has nothing to do with this, we are only testing the new patcher that still is not able to patch kexts on the kernel cache, but injecting personalities is the thing that I'm supposing to work with or without -f (loading kernel + kexts, or loading the prelinkedkernel) also now! The final goal is to load/patching kexts on the prelinked.otherwise not all the peoples are able to boot a vanilla installer, recovery hd etc, where you don't have the possibility to add kexts or manually patching them. 2 Link to comment Share on other sites More sharing options...
crazybirdy Posted February 18, 2017 Share Posted February 18, 2017 Good! Also your ATI card works this way? ... please test adding an additional personality (I'm searching for bugs) My NB can only update to 10.11.6, cause of sse4.1 is needed for 10.12. Will test later with the same kexts.plist for both BCM4322 and ATI 4330m. 1 Link to comment Share on other sites More sharing options...
Micky1979 Posted February 18, 2017 Share Posted February 18, 2017 Thanks so mutch! 1 Link to comment Share on other sites More sharing options...
crazybirdy Posted February 18, 2017 Share Posted February 18, 2017 Good! Also your ATI card works this way? ... please test adding an additional personality (I'm searching for bugs) The info-Injector works fine with both BCM4322 and ATI 4330m and patched-AMD4600Controller, the other patch work fine too. The AMD4600Controller patch doesn't work, need to use patched-AMD4600Controller to boot system. I think, it could be the size problem of hex. bdmesg and kexts.plist FYI. bdmesg+kexts.plist.zip 1 Link to comment Share on other sites More sharing options...
Micky1979 Posted February 18, 2017 Share Posted February 18, 2017 Wow thanks. I see that the OSBundleRequired is set to "Safe Boot", this mean that will not work in -f or UseKernelCache=No because the kext will not even be loaded. That remains for when we will able to patch the prelinkedkernel, so I think this is not a problem yet. Thanks again! 2 Link to comment Share on other sites More sharing options...
oliveiro Posted February 19, 2017 Share Posted February 19, 2017 Surely can't be the problem, but IGPEnabler does not exist and does nothing, "/System/Library/PrelinkedKernels/prelinkedkernel" should not exist on a Vanilla installer.... the prelinked is no longer inside "/.IABootFiles/"? Thanks for your answer. You're right, I meant the lesser vanilla installation (basesystem.dmg restore). When I tried the real vanilla, my understanding of IABoot files was too limited, and didn't tried that much to get the drive to simply show as a boot option with r4828. Also when I checked voodooprojects, I didn't find anything related to 2848 commit details (maybe it is a wrong place to look at for it ?), hence my question to try and understand the way it enables nvidiawebdriver with 10.12.3. IGPEnabler was an old option I actually used in my attempts to get a proper Capitan and 3 GPUs setup, didn't realize it was not part of the genuine Chameleon project, sorry for that. I came to the same conclusion of it being a kind of lesser GraphicsEnabler and ended not using it. Did you manually create a folder called Kernels under /S/L in which you copied the appropriate kernel file? Depending on the method used to create the USB installer, the kernel folder and file are not created/copied. As for IGPEnabler, it's from the dark side ; yet another example of copying and rebranding other people's work. It's just a renaming of Chameleon's GraphicsEnabler. Some call this stealing, justifiably so imo. Hi ! thanks too for answering. And no, I didn't copy the kernel file, that was my next thing to try, should the attempt to namely type the prelinkedkernel path at boot fail. But since it worked like this, I didn't try, and came to the conclusion I had missed a step in the installer preparation . You confirmed that. It brings me back to my other question : anybody with a successful setup with at least 2 nvidia GPU (GTX type?), Sierra 12.3 and chameleon r2848 ? Link to comment Share on other sites More sharing options...
crazybirdy Posted February 19, 2017 Share Posted February 19, 2017 Wow thanks. I see that the OSBundleRequired is set to "Safe Boot", this mean that will not work in -f or UseKernelCache=No because the kext will not even be loaded. That remains for when we will able to patch the prelinkedkernel, so I think this is not a problem yet. Thanks again! ok, but it can't patch prelinkedkernel for AMD4600Controller here, too. Link to comment Share on other sites More sharing options...
Micky1979 Posted February 19, 2017 Share Posted February 19, 2017 ok, but it cann't patch prelinkedkernel for AMD4600Controller here, too. Yes, there is no one line of code to allow that yet! Thanks for your answer. You're right, I meant the lesser vanilla installation (basesystem.dmg restore). When I tried the real vanilla, my understanding of IABoot files was too limited, and didn't tried that much to get the drive to simply show as a boot option with r4828. No problems, but I only wanted to know if were buggie loading the Vanilla installer. My goal is to load Vanilla stuff as mutch as possible. Also when I checked voodooprojects, I didn't find anything related to 2848 commit details (maybe it is a wrong place to look at for it ?), hence my question to try and understand the way it enables nvidiawebdriver with 10.12.3. I'm not part of the Chameleon Team, only writing some code, that once is tested, and if complete, I send to the mantainers.... and if He like maybe publish it. NVDAStartUpWeb patch is just, as I said, a tiny patch that replace the path where to look in the IOreng for the nvda_drv argument. It brings me back to my other question : anybody with a successful setup with at least 2 nvidia GPU (GTX type?), Sierra 12.3 and chameleon r2848 ? No idea 2 Link to comment Share on other sites More sharing options...
crazybirdy Posted February 19, 2017 Share Posted February 19, 2017 Yes, there is no one line of code to allow that yet! ok, that's fine, patched kext works fine. May I ask you about NVDAStartup.kext panic of 10.12? I have the same issue of NVDAStartup.kext panic since 10.12 as yellowpebble when boot with Clover, but work fine with Enoch 2839, etc without any panic. http://www.insanelymac.com/forum/topic/320033-rebooting-kernel-panic-screen-for-installing-sierra/ Do you have the same issue with your NVidia GT610? I think, maybe something is wrong with nvidia.c of Clover. Lots of people have the same issue with old nvidia card, and always need to use NVDAStartup.kext.310.42.25f01(10.11.6) to work with 10.12.x when they use Clover. So do I. Link to comment Share on other sites More sharing options...
Micky1979 Posted February 19, 2017 Share Posted February 19, 2017 ok, that's fine, patched kext works fine. May I ask you about NVDAStartup.kext panic of 10.12? I have the same issue of NVDAStartup.kext panic since 10.12 as yellowpebble when boot with Clover, but work fine with Enoch 2839, etc without any panic. http://www.insanelymac.com/forum/topic/320033-rebooting-kernel-panic-screen-for-installing-sierra/ Do you have the same issue with your NVidia GT610? I think, maybe something is wrong with nvidia.c of Clover. Lots of people have the same issue with old nvidia card, and always need to use NVDAStartup.kext.310.42.25f01(10.11.6) to work with 10.12.x when they use Clover. So do I. Hi crazybirdy. Sorry I'm not sure but NVDAStartupWeb looks also for supported chips family, maybe is that the cause? Note: I've managed to restore the NVRAM support in Enoch just some moment ago. Now I need to finalize the code.. 4 Link to comment Share on other sites More sharing options...
jsl Posted February 22, 2017 Share Posted February 22, 2017 Thanks for your answer. You're right, I meant the lesser vanilla installation (basesystem.dmg restore). When I tried the real vanilla, my understanding of IABoot files was too limited, and didn't tried that much to get the drive to simply show as a boot option with r4828. Also when I checked voodooprojects, I didn't find anything related to 2848 commit details (maybe it is a wrong place to look at for it ?), hence my question to try and understand the way it enables nvidiawebdriver with 10.12.3. IGPEnabler was an old option I actually used in my attempts to get a proper Capitan and 3 GPUs setup, didn't realize it was not part of the genuine Chameleon project, sorry for that. I came to the same conclusion of it being a kind of lesser GraphicsEnabler and ended not using it. Hi ! thanks too for answering. And no, I didn't copy the kernel file, that was my next thing to try, should the attempt to namely type the prelinkedkernel path at boot fail. But since it worked like this, I didn't try, and came to the conclusion I had missed a step in the installer preparation . You confirmed that. It brings me back to my other question : anybody with a successful setup with at least 2 nvidia GPU (GTX type?), Sierra 12.3 and chameleon r2848 ? Yes, Intel HD 4600 (internal) + GTX-980 (PCIe slot-1) + GTX-780 (PCIe slot-2) are working together by Enoch r2848 or Clover r4012 without problem. However this setup is not my favorite because it's quite unstable !!! I guess your problem is due to 550Ti, because according to my experience for MSI 550Ti GPU which need GE=Yes by Chameleon and failed by Clover. But GTX-980/780/680 need GE=No by Chameleon to enable their QE/CI correctly. In my hackintosh GE=Y will enable Intel HD 4600 only and disable the other GTX-980/780/680. My favorite setup is Intel HD 4600 + AMD-290X (or AMD-7950) + GTX-780 (or GTX-980) because it's efficient and very stable. 1 Link to comment Share on other sites More sharing options...
oliveiro Posted February 22, 2017 Share Posted February 22, 2017 Yes, Intel HD 4600 (internal) + GTX-980 (PCIe slot-1) + GTX-780 (PCIe slot-2) are working together by Enoch r2848 or Clover r4012 without problem. I guess your problem is due to 550Ti, because according to my experience for MSI 550Ti GPU which need GE=Yes by Chameleon and failed by Clover. But GTX-980/780/680 need GE=No by Chameleon to enable their QE/CI correctly. In my hackintosh GE=Y will enable Intel HD 4600 only and disable the other GTX-980/780/680. Hey! thanks for answering. With ElCap I actually had to do the same : GE=Yes to enable HD4600, and SkipNvidiaGfx=Yes to avoid injecting anything from Chameleon about the GTX. (but here GTX550Ti works fine with ElCap and no injection, QE-CI ok) I had to set the HD4600 as primary in Bios. Do you need the same setting ? With Sierra, the same works if I stick to nvidia official, but as soon as I try nvidiaweb, there's graphic corruption. Tried different chameleon boot options to no avail. So I believe it's related to nvidiaweb driver, which they say includes "Beta support" only for some configs. So I tried changing smbioses, that failed too. The good news I'd say, is GraphicsEnabler can be set to yes or no, GTXs are correctly configured by macOS, and Metal is supported. I'll probably wait and see till 10.12.4 and new web driver. Link to comment Share on other sites More sharing options...
jsl Posted February 22, 2017 Share Posted February 22, 2017 Hey! thanks for answering. With ElCap I actually had to do the same : GE=Yes to enable HD4600, and SkipNvidiaGfx=Yes to avoid injecting anything from Chameleon about the GTX. (but here GTX550Ti works fine with ElCap and no injection, QE-CI ok) I had to set the HD4600 as primary in Bios. Do you need the same setting ? With Sierra, the same works if I stick to nvidia official, but as soon as I try nvidiaweb, there's graphic corruption. Tried different chameleon boot options to no avail. So I believe it's related to nvidiaweb driver, which they say includes "Beta support" only for some configs. So I tried changing smbioses, that failed too. The good news I'd say, is GraphicsEnabler can be set to yes or no, GTXs are correctly configured by macOS, and Metal is supported. I'll probably wait and see till 10.12.4 and new web driver. Yes, for enable both internal and extra Nvidia or AMD GPU in my Z77, Z87, or Z97 MB need set Intel HD 4000 or 4600 as primary display in BIOS. At 10.12.3 or higher NvidiaWebDriver can NOT be Enabled for 2 GTX GPUs at the same time esp by Clover ! I'll get reboot immediately. 1 Link to comment Share on other sites More sharing options...
oliveiro Posted February 22, 2017 Share Posted February 22, 2017 but I only wanted to know if were buggie loading the Vanilla installer. My goal is to load Vanilla stuff as mutch as possible. Well I would not state it's buggy, but I couldn't see the USB or HDD installer as an option in the chameleon menu. It might have been a mess I created with UEFI and partition tables options. NVDAStartUpWeb patch is just, as I said, a tiny patch that replace the path where to look in the IOreng for the nvda_drv argument. Ah ok thanks ! is the path relative or something like "AppleACPIPCI/PEG0@1/IOPP/PEGP@0" or "AppleACPIPCI/RP05@1C,4/IOPP/PXSX@0" ? Yes, for enable both internal and extra Nvidia or AMD GPU in my Z77, Z87, or Z97 MB need set Intel HD 4000 or 4600 as primary display in BIOS. At 10.12.3 or higher NvidiaWebDriver can NOT be Enabled for 2 GTX GPUs at the same time esp by Clover ! I'll get reboot immediately. Interesting. I don't use clover (all my attempts were slooooow... maybe because I have lot of drives...), did you try enoch/chameleon on 10.12.3 +webdrivers ? does it reboot too ? thanks ! Link to comment Share on other sites More sharing options...
jsl Posted February 22, 2017 Share Posted February 22, 2017 Well I would not state it's buggy, but I couldn't see the USB or HDD installer as an option in the chameleon menu. It might have been a mess I created with UEFI and partition tables options. Ah ok thanks ! is the path relative or something like "AppleACPIPCI/PEG0@1/IOPP/PEGP@0" or "AppleACPIPCI/RP05@1C,4/IOPP/PXSX@0" ? Interesting. I don't use clover (all my attempts were slooooow... maybe because I have lot of drives...), did you try enoch/chameleon on 10.12.3 +webdrivers ? does it reboot too ? thanks ! It can enable NvidiaWebDriver by Enoch r2848 at 10.12.3, but not stable as posted before. In your case I'll try GE=Y or GE=N to boot either nvda_drv=1 or nvda_drv=0 to get the triple screens as stable as possible. 1 Link to comment Share on other sites More sharing options...
oliveiro Posted February 22, 2017 Share Posted February 22, 2017 It can enable NvidiaWebDriver by Enoch r2848 at 10.12.3, but not stable as posted before. In your case I'll try GE=Y or GE=N to boot either nvda_drv=1 or nvda_drv=0 to get the triple screens as stable as possible. When you write "not stable" do you mean system stability or display stability ? i did try GE and nvda_drv, all 4 scenarii, GE=yes nvda_drv=0 is stable and performing well Link to comment Share on other sites More sharing options...
jsl Posted February 22, 2017 Share Posted February 22, 2017 When you write "not stable" do you mean system stability or display stability ? i did try GE and nvda_drv, all 4 scenarii, GE=yes nvda_drv=0 is stable and performing well When NvidiaWebDriver was enabled for both GTX-980 and GTX-780 system easily crashed on play video or flash media with Firefox. Link to comment Share on other sites More sharing options...
gabryele Posted February 27, 2017 Share Posted February 27, 2017 what's the last package version to use chameleon on SIerra without problems? is first page updated? It's not human read about 200 pages Link to comment Share on other sites More sharing options...
Micky1979 Posted March 1, 2017 Share Posted March 1, 2017 Guys, finally we have the NVRAM + kernel and kexts patcher (working on the prelinkedkernel). Moved the old and new code from drivers.c to load.c in the appropriate DecodeMacho/segments functions Later I'll give you a boot file to test! 3 Link to comment Share on other sites More sharing options...
ErmaC Posted March 1, 2017 Author Share Posted March 1, 2017 Guys, finally we have the NVRAM + kernel and kexts patcher (working on the prelinkedkernel). Moved the old and new code from drivers.c to load.c in the appropriate DecodeMacho/segments functions Later I'll give you a boot file to test! Congratulations!!! ErmaC 2 Link to comment Share on other sites More sharing options...
Micky1979 Posted March 1, 2017 Share Posted March 1, 2017 Ok boot file attached. kexts patcher As I said the kext patcher can now patch prelinked kexts by adding the kext name you are intrested in to kexts.plist. Since last tests I modified the kexts.plist structure and now you have to put your custom patches under "KextsPatches": kernel patches As for kexts you can patch the kernel in kernel.plist under "KernelPatches": Remind you that MatchOS, MatchBuild and Comment are optional keys. NVRAM Embedded in the boot file, there is a modified version of FileNVRAM.kext, no module. To dump the nvram you have to use NVRAM Dumper Pro.app v1.3+ (attached): (making this app to be used in Clover the idea to add back the nvram support in Enoch ) Embedded kext is 64bit only and I have a fork of meklort's github repository where I'll add changes soon and make a pull request to Him. So please report back. (I'm testing it also with real nvram in Clover.. and apparently works well, plus the driver add some missing stuff , but for that I've to check more) Note Olds FileNVRAM.dylib and KernelPatcher.dylib prevent the internal kernel patcher to do their things but still allow the kexts patcher. FileNVRAM.dylib if loaded cause the internal nvram stuff in a off state. Why? Just avoiding conflicts, so you can decide to use one or another, but for this tests mentioned modules shoud be removed. FileNVRAM.kext if already present in Extra/Extensions, /S/L/E or /L/E should be removed.The one from the bootloader should win against it. Delete it anyway. Now you can use the NVDIA Web Drivers by adding the relative arg when pressing the relative option in the NVIDIA Web Drivers Manager, or add it manually. Old patch I made is unnecessary and deprecated, deleted! kernel.plist and kexts.plist attached as example to follow. Other old option are still there. Please report back! boot r2851 nvram+kextsPatcher.zip kexts.plist.zip kernel.plist.zip NVRAM Dumper Pro.app_v.1.3.zip 7 Link to comment Share on other sites More sharing options...
Micky1979 Posted March 2, 2017 Share Posted March 2, 2017 thanks! Link to comment Share on other sites More sharing options...
Shaneee Posted March 2, 2017 Share Posted March 2, 2017 Replaced the boot file on my EFI and OS X won't boot now. Stuck loading the graphics (IOBluetooth Error). My GT 430 just needs GE=Yes no web drivers needed. Edit: Fixed by replacing the boot file using Windows with r2848 Link to comment Share on other sites More sharing options...
gabryele Posted March 2, 2017 Share Posted March 2, 2017 Please, I need help with web driver Nvidia: MacOs does not load driver nvidia. Must I select some setting (nv_disable=1 not work for me)? in the download page http://www.insanelymac.com/forum/files/file/71-enoch/I read - Ability to load Nvidia Webdriver (Sierra) (by Micky1979) but how can we load Nvidia Webdriver? with default installation of Enoch, the webdriver doesn't work. In Clover to load nvidia driver, I had to change config file in this way: <key>SystemParameters</key><dict><key>InjectKexts</key><string>YES</string><key>InjectSystemID</key><true/><key>NvidiaWeb</key><true/></dict> I prefer use Enoch+Chameleon, but I can't load the graphic driver. May be a problem of FileNVRAM? http://www.insanelymac.com/forum/topic/231075-chameleon-23svn-official-pkg-installer-binaries/page-194?do=findComment&comment=2283157 or is a problem of version? I have installed the 10.12.1. Do I need to run something post OS installation (for fakesmc ad example) or Enoch has all we need? Link to comment Share on other sites More sharing options...
Recommended Posts