[GUIDE] Dell XPS 15 (9550) Mojave 10.14 / 10.15 Quick Installation

[joevt]

2 hours ago, golimpio said:

 

I've tried it, it gave me 2560x1440 but not at the correct refresh rate and it doesn't look as good as the original from MacOS (without SwitchResX), it's like when you have 2560x1440 without proper HiDPI, it's uncomfortable to look at the screen. I might have done something wrong though.

 

What does Hackintool say about the iGPU? Select the proper ig-platform-id from proper Intel Generation.

 

 

0x19260004 is the Intel Iris Graphics 540. What platform-id were you using before? What CPU do you have?

 

In SwitchResX, double click the resolution to see the timing info. What Pixel Clock, Active, and Scale to numbers is it using?

 

 

 

 

[golimpio]

3 hours ago, joevt said:

What does Hackintool say about the iGPU? Select the proper ig-platform-id from proper Intel Generation.

 

 

0x19260004 is the Intel Iris Graphics 540. What platform-id were you using before? What CPU do you have?

 

In SwitchResX, double click the resolution to see the timing info. What Pixel Clock, Active, and Scale to numbers is it using?

 

 

 

 

 

Before: 0x191b0000 (I've tried most of them)

 

SwitchResX for 0x191b0000:
Pixel Clock: 533.25
Active: 3840 x 2160

Scale: 2560 x 1440

Current:
 

SwitchResX for 0x19260004:
Pixel Clock: 533.25
Active: 3840 x 2160

Scale: 5120 x 2880

 

When I use the correct platform (0x191b0000), I cannot see 2560x1440 as a HiDPI option (the rendering is far inferior).

You can see in the Hackintool screenshot that I have the wrong display set, and this was on purpose to get 2560x1440 HiDPI. For Iris 530, I can only get 2560x1440 without HiDPI.

I wrote about the refresh rate on the other msg, but I think I was wrong, the refresh rate seems to be the same, it's just the way MacOS renders it that is inferior when it's not HiDPI. So far I can get the 2560x1440 resolution for the external monitor on Iris 530, but not the quality. Even setting 5120x2880 on SwitchResX, I still can't see it as HiDPI on the list.

Thanks!

[joevt]

24 minutes ago, golimpio said:

 

Before: 0x191b0000 (I've tried most of them)

 

SwitchResX for 0x191b0000:
Pixel Clock: 533.25
Active: 3840 x 2160

Scale: 2560 x 1440

Current:
 

SwitchResX for 0x19260004:
Pixel Clock: 533.25
Active: 3840 x 2160

Scale: 5120 x 2880

 

When I use the correct platform (0x191b0000), I cannot see 2560x1440 as a HiDPI option (the rendering is far inferior).

You can see in the Hackintool screenshot that I have the wrong display set, and this was on purpose to get 2560x1440 HiDPI. For Iris 530, I can only get 2560x1440 without HiDPI.

I wrote about the refresh rate on the other msg, but I think I was wrong, the refresh rate seems to be the same, it's just the way MacOS renders it that is inferior when it's not HiDPI. So far I can get the 2560x1440 resolution for the external monitor on Iris 530, but not the quality. Even setting 5120x2880 on SwitchResX, I still can't see it as HiDPI on the list.

Thanks!

 

When using 0x191b0000, did you try adding a scaled custom resolution of 5120x2880 in SwitchResX? Create the scaled timing, then restart the computer, and see if the system accepted the scaled resolution.

 

Maybe try overriding some of the flags or settings for the 0x191b0000 framebuffer to match the flags or settings in the 0x19260004 framebuffer. Set FramebufferMemorySize to 57 MB or 0 MB. Select FBLimit4KSourceSize.

 

I have a 6700K with Intel HD Graphics 530 (using iMac17,1). I'll try some tests tomorrow.

 

[Tikakan]

Hi, 

 

So I made a new 10.14.6 bootstick with  clover 2.5 v5103 and the 10.15 EFI Folder from wmchris, result: after installation on a wiped SSD and copying the allocfix.efi into the driver folder it booted. I copied the kexts to L/E, at the moment the system is running flawlessly. 

 

I hope, it stays that way. 

 

Best wishes from Vienna, Tikakan 

Edited February 13, 2020 by Tikakan

[Krim404]

23 hours ago, Tikakan said:

Hi, 

 

So I made a new 10.14.6 bootstick with  clover 2.5 v5103 and the 10.15 EFI Folder from wmchris, result: after installation on a wiped SSD and copying the allocfix.efi into the driver folder it booted. I copied the kexts to L/E, at the moment the system is running flawlessly. 

  

I hope, it stays that way. 

  

Best wishes from Vienna, Tikakan 

A big problem with the allocation error is the fact, that the cached kernel expands. AllocFix tries to clean up the memory to allow more free space in one location - but this is limited.

 

When your kernel is too big: try slimming it down.

[golimpio]

On 2/11/2020 at 10:51 PM, joevt said:

 

When using 0x191b0000, did you try adding a scaled custom resolution of 5120x2880 in SwitchResX? Create the scaled timing, then restart the computer, and see if the system accepted the scaled resolution.

 

Maybe try overriding some of the flags or settings for the 0x191b0000 framebuffer to match the flags or settings in the 0x19260004 framebuffer. Set FramebufferMemorySize to 57 MB or 0 MB. Select FBLimit4KSourceSize.

 

I have a 6700K with Intel HD Graphics 530 (using iMac17,1). I'll try some tests tomorrow.

 

 

Thanks for your help. Unfortunately using SwitchResX for a scaled custom resolution of 5120x2880 or changing the framebuffer memory didn't work for me. I can get 1440p resolution, bot low resolution only and not HiDPI.

My lack of knowledge doesn't help me to find the right patch and I don't really know if it's even possible to patch it for this specific case (1440p HiDPI), since it looks like MacOS decides which resolutions will support HiDPI based on the ig-platform-id, and that's why Intel Iris Graphics 540 works out of the box in my case.

Perhaps it might be simple to just try to fix the sleep issue. If I can figure it out why the internal display wakes up to a black screen (not totally black, as I can see the it's turned on).

*EDIT:* Up to 10.13.x it was simple to have 1440p HiDPI and sleep working, I just needed to apply the Floris497 mac-pixel-clock-patch and I'd have HiDPI.

Edited February 15, 2020 by golimpio
Edited since I don't actually remember changing the ig-platform-id before 10.14.x

[Rajveer86]

I'm currently on 10.15.1 and am looking to update to 10.15.3. Apart from a newer version of Clover needed, is the Github repo compatible wtih 10.15.3 or are there any changes that I'd need to make?

[agrafuese]

@wmchris and all,

 

I wanted to report that I've been getting distorted audio when using headphones (I'm not sure why or what triggers this, but it *might* be related to sleep/wake [unconfirmed]). In looking for a solution, I came across the hda-verb helper and an associated script, which seems to have solved the issue as far as my early testing goes. As I understand it, these work alongside CodecCommander.kext, so I'm curious why the tutorial doesn't mention hda-verb and the script. Is another kext (AppleALC, perhaps?) supposed to be doing this job and failing for me? I'm just trying to figure out if using hda-verb/script is redundant. I don't mind using them, but if we have a kext to do this and it is failing, I would like to prioritize fixing it instead of using hda-verb/script.

 

EDIT: This method also solved a problem I was having with combo jack (3-ring) headphones not working at all. 

 

EDIT 2: Not sure if related, but - I am still having trouble with internal microphone not working. Before I used hda-verb/script, I was getting nothing but noise, and now I get no signal at all. Also, I don't get a microphone device recognized when plugging in a headset with a microphone. Does our AppleALC (or something else) need further patching or does something need injection of some kind? This is not an area that I am knowledgeable in.

Edited February 18, 2020 by agrafuese

[golimpio]

6 hours ago, agrafuese said:

@wmchris and all,

 

I wanted to report that I've been getting distorted audio when using headphones (I'm not sure why or what triggers this, but it *might* be related to sleep/wake [unconfirmed]). In looking for a solution, I came across the hda-verb helper and an associated script, which seems to have solved the issue as far as my early testing goes. As I understand it, these work alongside CodecCommander.kext, so I'm curious why the tutorial doesn't mention hda-verb and the script. Is another kext (AppleALC, perhaps?) supposed to be doing this job and failing for me? I'm just trying to figure out if using hda-verb/script is redundant. I don't mind using them, but if we have a kext to do this and it is failing, I would like to prioritize fixing it instead of using hda-verb/script.

 

EDIT: This method also solved a problem I was having with combo jack (3-ring) headphones not working at all. 

 

EDIT 2: Not sure if related, but - I am still having trouble with internal microphone not working. Before I used hda-verb/script, I was getting nothing but noise, and now I get no signal at all. Also, I don't get a microphone device recognized when plugging in a headset with a microphone. Does our AppleALC (or something else) need further patching or does something need injection of some kind? This is not an area that I am knowledgeable in.

 

I think this script was always there (now it's at `.../10.15/Post-Install/Additional Steps/Audio/ComboJack`), the part in the guide that mentions the script is in "Step 7: Fixes / Enhancements / Alternative Solutions / Bugs" (https://github.com/wmchris/DellXPS15-9550-OSX/blob/10.15/Tutorial_10.15_Step7.md), I guess the folder where the script lives might be wrong since it changed along the way.

 

*EDIT:* Fixed the audio related links in the tutorial (Step 7: Fixes / Enhancements / Alternative Solutions / Bugs)

Edited February 19, 2020 by golimpio

[golimpio]

On 2/16/2020 at 11:33 PM, Rajveer86 said:

I'm currently on 10.15.1 and am looking to update to 10.15.3. Apart from a newer version of Clover needed, is the Github repo compatible wtih 10.15.3 or are there any changes that I'd need to make?

 

I've upgraded to 10.15.3 without issues (I had the latest Clover installed at the time). 

[agrafuese]

7 hours ago, golimpio said:

 

I think this script was always there (now it's at `.../10.15/Post-Install/Additional Steps/Audio/ComboJack`), the part in the guide that mentions the script is in "Step 7: Fixes / Enhancements / Alternative Solutions / Bugs" (https://github.com/wmchris/DellXPS15-9550-OSX/blob/10.15/Tutorial_10.15_Step7.md), I guess the folder where the script lives might be wrong since it changed along the way.

 

*EDIT:* Fixed the audio related links in the tutorial (Step 7: Fixes / Enhancements / Alternative Solutions / Bugs)

Thanks for pointing this out. I did a text search in the tutorial for things like hda-verb, combojack, etc before I posted, but nothing came up - now I realize it was because I didn't see that the additional fixes section was on its own page. My bad. Just glad to see I'm taking all the proper steps

[agrafuese]

Update for anyone reading this who may also be curious about the combo jack and headphone distortion fixes:

 

The fix included in this tutorial/repo is great, but the install script that comes with it needs updating/re-writing. The scripting/wording in install.sh should be as follows:

 

#!/bin/bash
echo "This installer requires admin rights. Also make sure you don’t have Time Machine running or any usb stick inserted!"
echo "IMPORTANT: Please remove any previous jack fixes manually before continuing. Also, remove CodecCommander.kext from your EFI kexts folder, and replace with the included VerbStub.kext."
echo "Do you want to continue?"
select yn in "Yes" "No"; do
    case $yn in
        Yes ) break;;
        No ) exit;;
    esac
done

echo "Installing local configuration"
sudo cp ComboJack /usr/bin
sudo chmod 755 /usr/bin/ComboJack
sudo chown root:wheel /usr/bin/ComboJack
sudo cp hda-verb /usr/bin
sudo chmod 755 /usr/bin/hda-verb
sudo chown root:wheel /usr/bin/hda-verb
sudo cp com.XPS.ComboJack.plist /Library/LaunchAgents/
sudo chmod 644 /Library/LaunchAgents/com.XPS.ComboJack.plist
sudo chown root:wheel /Library/LaunchAgents/com.XPS.ComboJack.plist
sudo launchctl load /Library/LaunchAgents/com.XPS.ComboJack.plist
sudo mkdir /usr/local/share/
sudo chmod 755 /usr/local/share/
sudo mkdir /usr/local/share/ComboJack/
sudo chmod 755 /usr/local/share/ComboJack/
sudo cp Headphone.icns /usr/local/share/ComboJack/
sudo chmod 644 /usr/local/share/ComboJack/Headphone.icns
sudo cp l10n.json /usr/local/share/ComboJack/
sudo chmod 644 /usr/local/share/ComboJack/l10n.json
echo
echo "Please reboot! Also, it may be a good idea to turn off \"Use"
echo "ambient noise reduction\" when using an input method other than"
echo "the internal mic (meaning line-in, headset mic). As always: YMMV."
echo
echo "You can check to see if the watcher is working in the IORegistry:"
echo "there should be a device named \"VerbStubUserClient\" attached to"
echo "\"com_XPS_SetVerb\" somewhere within the \"HDEF\" entry's hierarchy."
echo
echo "Enjoy!"
echo
exit 

Essentially, I have removed any scripting that involved files or caches related to /S/L/E, since we don't generally use that type of configuration in this repo. I also added lines to create the /usr/local/share/ and /usr/local/share/ComboJack folders, as these lines were missing previously, resulting in some of the files not copying to their proper locations, and therefore the listener script was not working properly.

 

*NOTE: I have not included scripting to automatically remove CodecCommander.kext or copy VerbStub.kext in the EFI kexts folder, because it is much safer for the user to do that manually themselves, and it is easy enough to do so. So, I added notes to remind the user to do so instead. Also, I realize that the bit about "remove previous jack fixes" is vague, but hopefully users can figure that out themselves. There are just too many variables to write a script that would satisfy all possible use case scenarios and configurations.

 

 

*ALSO NOTE: I am not "officially" requesting that these changes be made to the repo, out of respect for any possible reasons why someone would argue against them. I am simply suggesting that these changes be reviewed in order to bring the install.sh script up to date with the way that the repo is structured. The old script was confusing and didn't work properly OOTB.

[Rajveer86]

18 hours ago, golimpio said:

 

I've upgraded to 10.15.3 without issues (I had the latest Clover installed at the time). 

 

Thanks, I downloaded the Github repo, replaced the Clover build from your zip below and the update went fine

 

On 2/8/2020 at 3:37 AM, golimpio said:

CLOVER.zip

 

[jwijesundra]

did a clean install to 10.15.3

struggled with the memory allocation issue, couldn't even get the installer to boot, even after setting slide value.

 

what worked in the end was to use version 1 of the memory fix along with allocfix dot efi 

 

i.e. in CLOVER/drivers/UEFI

put AllocFix.efi       

and OsxAptioFixDrv-64.efi   

removed

OsxAptioFix3Drv-64.efi

(kept the slide as well, not sure if it was necessary)

 

it works most of the time but sometimes during the scrolling text the machine reboots by itself but usually will work fine after reboot. Does anyone know why this happens or how I can diagnose?

 

Besides that all seems to be working well.

 

Got scared a sec after enabling file vault and the mac boot option disappeared. Turned out by default clover hides preboot so need to edit config plist (on clover configurator go to section gui -> hide volume, and remove preboot)

Also had trouble booting windows using clover. Windows would just crash. Solution was to disable legacy boot and allow only UEFI. The other weird thing was I installed latest clover 5104, but before this fix the splash screen showed an older version (5098). After the fix, it showed 5104. Any ideas why? The repo is 5102 so not sure where 5098 came from.

 

Edited February 24, 2020 by jwijesundra

[agrafuese]

19 hours ago, jwijesundra said:

it works most of the time but sometimes during the scrolling text the machine reboots by itself but usually will work fine after reboot. Does anyone know why this happens or how I can diagnose?

 

Is it a kernel panic when this happens? - i.e. when it finally does boot into macOS do you get a crash error report window saying that there was an issue, and additional info? If so, you might want to copy paste the info it gives you so we can see where the problem is coming from. I had a similar issue where I would get a reboot during startup and it was a kernel panic caused by one of the kexts.

 

Also, what BIOS version are you running? Latest (1.13.1) ? If not, you may want to update. I just recently updated myself, and it unexpectedly fixed a certain issue I was having (I am not on Catalina, and it wasn't an issue you are having, but I figured I would mention this just in case).

[jwijesundra]

no kernel panic - don't get any message saying there was an issue.

 

i'm running 1.11.2 relatively new but not the latest.

 

thanks for the advice!

[agrafuese]

Wild guess here, but you could try setting slide back to 0 and see if that helps reliability, since you're already using the allocfix/Aptio drivers. I'm only aware of the allocation error coming up and stopping the boot-up process, not automatically rebooting, but I'm far from an expert with that sort of thing. In fact, the one time I tried to put in a slide value, it only made things worse. So for now, I just get random allocation errors every once in a while, and I do a ctl+alt+del to try again. Works fine on the next reboot 99% of the time. 

[Krim404]

6 hours ago, jwijesundra said:

no kernel panic - don't get any message saying there was an issue.

 

i'm running 1.11.2 relatively new but not the latest.

 

thanks for the advice!

this is absolutely normal. if there is not enough space freed or something went wrong, it just tries again. only an issue if this happens multiple times in a row.

[agrafuese]

@wmchris In regards to not having enough space, are we talking about RAM? Or some kind of cache file? Is there anything that a user could do before rebooting that would increase the reliability of this "resource" for the following boot process? - i.e. can anything be emptied, purged, reset, etc before rebooting macOS (or Windows, even)? Sorry if I sound like an idiot. This is all well above my understanding, so I am just trying to learn. I would google it, but I don't know what the proper terminology is.

[Krim404]

19 hours ago, agrafuese said:

@wmchris In regards to not having enough space, are we talking about RAM? Or some kind of cache file? Is there anything that a user could do before rebooting that would increase the reliability of this "resource" for the following boot process? - i.e. can anything be emptied, purged, reset, etc before rebooting macOS (or Windows, even)? Sorry if I sound like an idiot. This is all well above my understanding, so I am just trying to learn. I would google it, but I don't know what the proper terminology is. 

ram, yes, but lower memory. you cant to anything here as this is populated by the bios, the enabled options, the used devices and so on.

[agrafuese]

Thanks for explaining, makes sense now.

[jwijesundra]

On 2/26/2020 at 5:07 PM, wmchris said:

this is absolutely normal. if there is not enough space freed or something went wrong, it just tries again. only an issue if this happens multiple times in a row.

glad to hear!

[golimpio]

I've been testing some changes I've found at https://github.com/xxxzc/xps15-9550-macos, which allowed me to install the latest VoodooPS2Controller.kext and VoodooI2C.kext. So far (2 days) it has worked well.

 

Basically, I've removed "Rename ECDV to EC" from CLOVER/config.plist and added a fake EC device at ACPI/SSDT-PCI0.dsl

 

The whole thing can be seen in details at the github above, if anyone is interested in testing it.

[agrafuese]

@golimpio Are you using I2C as well as VoodooInput? Or did you stop using VoodooInput?

 

EDIT:

I am asking because I tried using the latest versions of VoodooI2C & VoodooI2CHID, and I got panics after waking from sleep. It's a shame, because those kexts make the trackpad sooooo much smoother than VoodooInput. The newest version of VoodooInput shoots the mouse all over the screen when I click. It's terrible. 

 

So, I have a question about this:

 

Quote

I've removed "Rename ECDV to EC" from CLOVER/config.plist and added a fake EC device at ACPI/SSDT-PCI0.dsl

 

Does it solve the panics caused by I2C? Or were you not experiencing them in the first place?

Edited March 11, 2020 by agrafuese

[golimpio]

7 hours ago, agrafuese said:

@golimpio Are you using I2C as well as VoodooInput? Or did you stop using VoodooInput?

 

EDIT:

I am asking because I tried using the latest versions of VoodooI2C & VoodooI2CHID, and I got panics after waking from sleep. It's a shame, because those kexts make the trackpad sooooo much smoother than VoodooInput. The newest version of VoodooInput shoots the mouse all over the screen when I click. It's terrible. 

 

So, I have a question about this:

 

 

Does it solve the panics caused by I2C? Or were you not experiencing them in the first place?

I’m using the latest I2C and stopped using VoodooInput. I wasn’t using the latest I2C version before, so I wasn’t experiencing the panics. I should have tested it before make the changes, but unfortunately I didn’t. So I can’t confirm that the changes I’ve done really solved the panic issue.