Error function main 320 uploading image using sahara protocol failed

Any advice?

Now i´m just missing the GSI ROM installation.

Next step is just flashing the «system_a» and «system_b» partitions, right? This can be done from the TWRP as well, i suppose, or needs to be done from the bootloader with «fastboot flash system_a ROMFILENAME.img»?

This was answered last week, please see posts around #340.

When using the command: .fastboot flash boot_b.TWRP-CUBE_X_Neo-20201028.img — write:

There is as space missing between boot_b and ….

GSI ROM from link in guide comes with .xz entension, this is not recognized in TWRP. Renaming to zip doesn´t help. Any clues?

Extract it with 7ZIP and flash the IMG-File (in TWRP you can choose between ZIP or IMG).

So I managed it after two days
What rom would you recommend me. I am mainly interested in stable operation without bugs.

If 5 ghz wifi is not crucial, i recommend crdrom android 11

If 5 ghz wifi is not crucial, i recommend crdrom android 11

If I would like 5g wifi, what kind of rom would you recommend?

If I would like 5g wifi, what kind of rom would you recommend?

The original one. Or maybe LOS 17.1 works with 5 GHz, please search this thread.

I still run on LOS 17.1 and can confirm that 5 GHz WiFi (somehow) works. It has week signal sensitivity and transfer rates are low for 5 GHz. If you won’t move your tablet too far from router, it will work. But do not expect any great WiFi performace. As far as I know this is the problem of all Android 10 based GSI ROMs that were discussed here in this thread. Because of this problem I’m still considering moving back to the original ROM. But do not want to lose the dark mode of Android 10.

I confirm. Definitely dark mode makes far better battery performance for the tablet. Just for this reason even, it’s worth to move to Android 10 or 11

Hi, I have a problem installing rom which is 3gb+ size.
When flashing in twrp, he writes «Size of image is larger than target device».
It has a solution?

Afaik the size of the system partition is 3069 MB, if you want to flash bigger roms, you have to change the partition layout. Or flash a rom without gapps and a nano/pico gapps package afterwards (this normally needs a resize of the system partition file-system).

Side question, how often do these go on sale? I’m very curious but don’t need a table for at least a few months.

Not sure. You’d have to watch sites like Banggood, Gearbest, Geekbuying, & DHGate.

Side note — I think I’ll update the guide to work with the 12/20 firmware since it’s been tested thoroughly.

For anyone wondering about ROMs, I still prefer & recommend qiDroid (Android 10) for stability https://sourceforge.net/projects/treblerom/files/qiDroid/

Another recommendation and trick for Safety NET pass :

The option(s) in phh treble settings don’t work.
The only way to make it work is to flash magisk latest canary build and install MagiskHide Props Config v5 module -> edit device fingerprint — > pick certified finger print -> Samsung -> Samsung Galaxy Tab S7+ Wifi SM-T970 (11) and reboot tablet.

I randomly tried this again yesterday with the newer version crdrom-v307-210513 and now it passed the safety net check (unlike the previous version). Nothing else has changed.

Then your needs have become at least three times as expensive within a year?

More like I realized that I wanted a better tablet than what the X Neo provided. The long fight with figuring how to unlock the bootloader & headaches with GSI roms just pushed me in that direction.

I support anyone that wants to continue tinkering & using it. If there’s ever a new official ROM from ALLDOCUBE, I have no problem with testing it out & making sure it works right. Honestly, I even forgot to update the guide to the most recent version but I haven’t tracked if any new ones are out since then.

If anyone wants to take over updating the guide, I’ll gladly hand over the original DOCX for modification with the stipulation that my work & username not be removed from it (improving it & giving me credit is fine as well). Actually, allowing more contributors to it with sharing in OneDrive would probably work best. I recommend publishing it to PDF for public access though.

More like I realized that I wanted a better tablet than what the X Neo provided. The long fight with figuring how to unlock the bootloader & headaches with GSI roms just pushed me in that direction.

I support anyone that wants to continue tinkering & using it. If there’s ever a new official ROM from ALLDOCUBE, I have no problem with testing it out & making sure it works right. Honestly, I even forgot to update the guide to the most recent version but I haven’t tracked if any new ones are out since then.

If anyone wants to take over updating the guide, I’ll gladly hand over the original DOCX for modification with the stipulation that my work & username not be removed from it (improving it & giving me credit is fine as well). Actually, allowing more contributors to it with sharing in OneDrive would probably work best. I recommend publishing it to PDF for public access though.

Hi CrimsonKnight13,
I’ m extremely happy with the X Neo. Please try to keep us informed if there are any updates.

Hi CrimsonKnight13,
I’ m extremely happy with the X Neo. Please try to keep us informed if there are any updates.

Just checked. Nothing new since the Chinese 20201216 update. Have they abandoned it already?

So thank you for that!

FYI, Eremitein (developer of many GSI roms, e.g. qidroid and crdrom) wrote at XDA, that the build server he used was damaged and all his work is suspended.

There seems to be a recent version of caos11 so maybe that’s been resolved?

Just checked. Nothing new since the Chinese 20201216 update. Have they abandoned it already?

Apparently so. There is no reply to the broken dark theme threads on the official forums — and that should be fairly easy to fix.

Me I thought to try a customrom to get that straightened out and I’m failing at it. Hard.

QFIL hangs indefinitely (I left it over night). When I disconnect the device I get a send reset_state_machine error.
The guide said to watch for a progress-bar but there is none while that goes on. It’s all inactive (like the ‘Download’ button in the screenshot).
No movement anywhere.

MiFlash gives the infamous hello packet error:

[4:00:28 PM  COM3]: serial port COM3 opend successfully
[4:00:28 PM  COM3]:[COM3]:start flash.
[4:00:29 PM  COM3]:cannot receive hello packet,MiFlash is trying to reset status!
[4:00:30 PM  COM3]:try to reset status.
[4:00:32 PM  COM3]:write time out try agian 1
[4:00:34 PM  COM3]:write time out try agian 2
[4:00:37 PM  COM3]:write time out try agian 3
[4:00:39 PM  COM3]:write time out try agian 4
[4:00:42 PM  COM3]:write time out try agian 5
[4:00:44 PM  COM3]:write time out try agian 6
[4:00:47 PM  COM3]:write time out try agian 7
[4:00:48 PM  COM3]:error:The write timed out.
[4:00:48 PM  COM3]:System.Exception: write time out,maybe device was disconnected.
   at XiaoMiFlash.code.Utility.Comm.WritePort(Byte[] send, Int32 offSet, Int32 count)
   at XiaoMiFlash.code.bl.SerialPortDevice.SaharaDownloadProgrammer(Int32 programmertype)
   at XiaoMiFlash.code.bl.SerialPortDevice.XiaomiFlash()     at XiaoMiFlash.code.Utility.Comm.WritePort(Byte[] send, Int32 offSet, Int32 count)
   at XiaoMiFlash.code.bl.SerialPortDevice.SaharaDownloadProgrammer(Int32 programmertype)
   at XiaoMiFlash.code.bl.SerialPortDevice.XiaomiFlash()
[4:00:48 PM  COM3]:no provision exit:COM3

I’ve been trying several QPST/QFIL/9008 Drivers and I get varied responses. This latest one seems promising:

From the attached MiFlash port traces for ufs and emmc (I believe I have the emmc version):

16:30:20: DEBUG: Only 0 bytes written of 227 requested, attempt 100 of 100


     _____                   
    |  ___|                   
    | |__ _ __ _ __ ___  _ __
    |  __| '__| '__/ _ | '__|
    | |__| |  | | | (_) | |   
    ____/_|  |_|  ___/|_| 


16:30:20: {ERROR: WritePort:5103 Could not write to '\.COM10', Windows API WriteFile failed! Your device is probably *not* on this port, attempted 100 times

Googling leads me to believe I’ve bricked my device. Only — I can access the bootloader, recovery and system. From android I can use adb to reboot. That does not sound particularly ‘bricked’ to me?

I did have a lot of trouble to get my device in 9008 mode since I needed to figure out to switch my device-driver from QUSB__BULK to Qualcomm HS_USB QDLoader 9008. Otherwise I’d only get diagnostic modes (900E for EFL mode and 9091 for android).
Then there was the command packet too large phase:

2021-07-26 20:32:52.242    20:32:52: ERROR: function: sahara_rx_data:287 Command packet length 1702240364 too large to fit
2021-07-26 20:32:52.243   
2021-07-26 20:32:52.243    20:32:52: ERROR: function: sahara_main:983 Sahara protocol error
2021-07-26 20:32:52.244   
2021-07-26 20:32:52.244    20:32:52: ERROR: function: main:320 Uploading  Image using Sahara protocol failed

Maybe in between all of that I inflicted a boo-boo on my neo. Any idea what I should try next?

Thanks for doing this tutorial, it’s what lead me to buy the device. No-name china tablets sadly are often abandoned out of the factory but treble allows us to circumvent that. Without a guide like the one you wrote many people (like me) would not even contemplate to try, however.

So thank you for that!

There seems to be a recent version of caos11 so maybe that’s been resolved?

Apparently so. There is no reply to the broken dark theme threads on the official forums — and that should be fairly easy to fix.

Me I thought to try a customrom to get that straightened out and I’m failing at it. Hard.

QFIL hangs indefinitely (I left it over night). When I disconnect the device I get a send reset_state_machine error.
The guide said to watch for a progress-bar but there is none while that goes on. It’s all inactive (like the ‘Download’ button in the screenshot).
No movement anywhere.

View attachment 378927

MiFlash gives the infamous hello packet error:

[4:00:28 PM  COM3]: serial port COM3 opend successfully
[4:00:28 PM  COM3]:[COM3]:start flash.
[4:00:29 PM  COM3]:cannot receive hello packet,MiFlash is trying to reset status!
[4:00:30 PM  COM3]:try to reset status.
[4:00:32 PM  COM3]:write time out try agian 1
[4:00:34 PM  COM3]:write time out try agian 2
[4:00:37 PM  COM3]:write time out try agian 3
[4:00:39 PM  COM3]:write time out try agian 4
[4:00:42 PM  COM3]:write time out try agian 5
[4:00:44 PM  COM3]:write time out try agian 6
[4:00:47 PM  COM3]:write time out try agian 7
[4:00:48 PM  COM3]:error:The write timed out.
[4:00:48 PM  COM3]:System.Exception: write time out,maybe device was disconnected.
   at XiaoMiFlash.code.Utility.Comm.WritePort(Byte[] send, Int32 offSet, Int32 count)
   at XiaoMiFlash.code.bl.SerialPortDevice.SaharaDownloadProgrammer(Int32 programmertype)
   at XiaoMiFlash.code.bl.SerialPortDevice.XiaomiFlash()     at XiaoMiFlash.code.Utility.Comm.WritePort(Byte[] send, Int32 offSet, Int32 count)
   at XiaoMiFlash.code.bl.SerialPortDevice.SaharaDownloadProgrammer(Int32 programmertype)
   at XiaoMiFlash.code.bl.SerialPortDevice.XiaomiFlash()
[4:00:48 PM  COM3]:no provision exit:COM3

I’ve been trying several QPST/QFIL/9008 Drivers and I get varied responses. This latest one seems promising:

From the attached MiFlash port traces for ufs and emmc (I believe I have the emmc version):

16:30:20: DEBUG: Only 0 bytes written of 227 requested, attempt 100 of 100


     _____                  
    |  ___|                  
    | |__ _ __ _ __ ___  _ __
    |  __| '__| '__/ _ | '__|
    | |__| |  | | | (_) | |  
    ____/_|  |_|  ___/|_|


16:30:20: {ERROR: WritePort:5103 Could not write to '\.COM10', Windows API WriteFile failed! Your device is probably *not* on this port, attempted 100 times

Googling leads me to believe I’ve bricked my device. Only — I can access the bootloader, recovery and system. From android I can use adb to reboot. That does not sound particularly ‘bricked’ to me?

I did have a lot of trouble to get my device in 9008 mode since I needed to figure out to switch my device-driver from QUSB__BULK to Qualcomm HS_USB QDLoader 9008. Otherwise I’d only get diagnostic modes (900E for EFL mode and 9091 for android).
Then there was the command packet too large phase:

2021-07-26 20:32:52.242    20:32:52: ERROR: function: sahara_rx_data:287 Command packet length 1702240364 too large to fit
2021-07-26 20:32:52.243  
2021-07-26 20:32:52.243    20:32:52: ERROR: function: sahara_main:983 Sahara protocol error
2021-07-26 20:32:52.244  
2021-07-26 20:32:52.244    20:32:52: ERROR: function: main:320 Uploading  Image using Sahara protocol failed

Maybe in between all of that I inflicted a boo-boo on my neo. Any idea what I should try next?

You’re having similar issues to what I had. Since I can’t see my guide document at work, did you miss the part where you have to set it to reset the port or connection in QFIL or MiFlash?

You’re having similar issues to what I had. Since I can’t see my guide document at work, did you miss the part where you have to set it to reset the port or connection in QFIL or MiFlash?

I did try to follow the guide closely. Since I couldn’t get it to work I’ve diverged quite a bit, however.
That’s why I now also use a ‘fresh’ virtual machine to test. Windows 7 32bit vs my current W10 x64. No difference whatsoever unfortunately.

I’m assuming you meant the download configuration settings with that reset part, yeah? Or perhaps not since there is no such setting in MiFlash. Huh. So there is more 3-way button mashing to be done?

Getting into EDL was a right pain also because I never was quite sure if I got there or not. QFIL displaying ‘Diagnostic mode’ might have been a signal to cycle as you wrote so I kept going. It might also have been a signal to switch drivers. Still not sure why this isn’t an issue for anyone else? Is there a way to make sure I’m where I need to be on the device side mode-wise?
There is no mention in the MiFlash part of the guide on cycling btw — is it not needed there? Seeing as it seems to have been added later its probably just an oversight.
On that note: the qfil sections read: «rawprogram_un**spare**.xml» while the MiFlash one has the correct? «rawprogram_un**sparse**.xml».

To illustrate this is what my device looks like without me changing anything:

I then install the drivers from the ‘drivers for 9008’ zipfile (which seem identical to the ones in the chinese rom) and choose the proper one manually with ‘update driver’.
Is there perhaps an order to installing setup.exe and qualcomm_hsusb_drv_1013_64111.exe ? I did reboot several times but I wonder whether it was even necessary.
During installation I’m prompted to choose between WWAN-DHCP and ETHERNET-DHCP. I chose the former since the driver version is higher (NDIS 6.20 vs 5.1) No idea if that is a good choice. Or even necessary for anything? Nothing popped out in my firewall logs…

This gets me:

Which is then available instantly in QFIL (see first picture). No program or machine restart asked for.

Another thing I’m not sure about is the firehose thing. Do I actually need to remove the files or is it sufficient to just pick the right .elf? And do I need to copy both _ddr and _lite? QFIL UFS section does not mention to remove the _lite, so probably not.

On the MiFlash side of things I’ve later tried to choose Chip->Qualcomm (seemed benign enough) and tried the Driver->Install option to get lots of USB drivers (read: unsigned warnings). Neither seemed to do anything, though. It’s not mentioned in the guide either so I didn’t expect much.

If there is anything you can think of for me to test or some information to provide let me know. I’m pretty much out of ideas

AmberKnave Did you jump straight to eMMC? Have you tried UFS as well? Next option is use a USB 2.0 or 3.1 port instead of 3.0 (or different from whatever you’ve used). Also, if you have other USB-C cables, try those, too.

First try went to UFS. I think that got me a ‘Command packet length 1702240364 too large to fit’ error. I’ve seen too many kinds by now to remember which was first. Probably should have asked for help first instead of fiddling myself.
Bootloader says ‘VARIANT — SDM EMMC’ btw. I figured you’d have mentioned looking there if that was reliable. Since I have zero hints on where my problem lies I keep trying to change one thing and then go through all setting-permutations that I can think of.

To be clear on the ‘reset the port or connection in QFIL or MiFlash’ from your previous post:
1. start program
2. shut down tablet
3. (re)connect usb
4. button-mash to get a black-screen
5. repeat 3. and 4. until qfil/miflash shows comXY without any ‘diagnostics’ or my fingers fall off (why isn’t this mode accessible from the bootloader?!)

As for ports my laptop only has so many and while I have a box full of micro usb cables this is my first and only usb-c device.
I might dig up another machine tomorrow — if flashing is even possible with windows XP?

First try went to UFS. I think that got me a ‘Command packet length 1702240364 too large to fit’ error. I’ve seen too many kinds by now to remember which was first. Probably should have asked for help first instead of fiddling myself.
Bootloader says ‘VARIANT — SDM EMMC’ btw. I figured you’d have mentioned looking there if that was reliable. Since I have zero hints on where my problem lies I keep trying to change one thing and then go through all setting-permutations that I can think of.

To be clear on the ‘reset the port or connection in QFIL or MiFlash’ from your previous post:
1. start program
2. shut down tablet
3. (re)connect usb
4. button-mash to get a black-screen
5. repeat 3. and 4. until qfil/miflash shows comXY without any ‘diagnostics’ or my fingers fall off (why isn’t this mode accessible from the bootloader?!)

As for ports my laptop only has so many and while I have a box full of micro usb cables this is my first and only usb-c device.
I might dig up another machine tomorrow — if flashing is even possible with windows XP?

SUCCESS!

Not sure which part pulled through in the end but from previous com-port related messages I’m guessing it’s the last one:

1. Reconnect USB often and in different slots
2. Use adb reboot edl (because button mashing is for suckers)
3. QFIL: Tools -> Get Flash Information
4. Delete AppDataRoamingQualcommQFILQFIL.config and redo QFIL settings

Point 4 seems particularly relevant as that file didn’t change since yesterday (modified/last accessed) timestamp and contained a wrong portnumber.
Not sure if it’s really relevant considering both logs and the statusbox showed different and recent ones.
Nevertheless it started working within one try of me deleting that…

Also I recommend using Tools->GetFlashInformation to get an educated guess on emmc vs ufs and see if the connection in edl works without any chance of interrupting anything critical.
Could have saved myself some time waiting for a flash that never came and a progressbar that never moved.

Onwards to get chinese developer privileges!

AmberKnave excellent! I’m glad you figured it out. The whole process can hit big headaches like you’ve seen.

Yeah I can see that. I’m still riding my success-high so I’m not bothered that I got stuck on the fastboot stage ^^

PS E:2106AllDCubeplatform-tools> .adb devices
List of devices attached
MYSNHERE        device

PS E:2106AllDCubeplatform-tools> .adb reboot bootloader
PS E:2106AllDCubeplatform-tools> .fastboot getvar unlocked
< waiting for any device >
## CTRL-C here
PS E:2106AllDCubeplatform-tools> .fastboot devices -l
PS E:2106AllDCubeplatform-tools> .fastboot devices

adb works, rebooting works, fastboot does not. no device-listing no getvar.

Good time to escape to the land of dreams to escape the fact that there are quite a few more steps ahead me thinks.

Thanks for your help so far!
(I’ll be sure to bother you again tomorrow)

AmberKnave be sure to check the USB devices in Device Manager. You might need to install the right fastboot driver.

That did it, thanks

Small typo in the guide:

2. Run .fastboot flash boot_a . TWRP-CUBE_X_Neo-20201028.img & verify

There is a space between ‘.’ and ‘TWRP’. So no copy+paste.

Quick questions about magisk:
do I flash this before or after a rom?
can I flash/install it via adb/fastboot like a rom?
Between:

Magisk-ef9d077c-phh.zip
Magisk-uninstaller-20200323.zip
Magisk-v20.4.zip

Do I need all of them and is there an order to flashing?

Thanks

That did it, thanks

Small typo in the guide:

2. Run .fastboot flash boot_a . TWRP-CUBE_X_Neo-20201028.img & verify

There is a space between ‘.’ and ‘TWRP’. So no copy+paste.

Thanks. I’ll have to update it with the last released Chinese firmware version, too. I’ve just been lazy about doing so.
Many on this thread have provided differing fixes but I’ve lost track. I’ll gladly provide collaboration privileges if anyone wants in on adding more to it and/or making it nicer to look at.

Quick questions about magisk:
do I flash this before or after a rom?
can I flash/install it via adb/fastboot like a rom?
Between:

Magisk-ef9d077c-phh.zip
Magisk-uninstaller-20200323.zip
Magisk-v20.4.zip

Do I need all of them and is there an order to flashing?

Thanks

I think 20.4 should be safe. Do some testing with it & if that gives you problems, go with the phh one (might be a newer version of that from PHH, too). Only use the uninstaller if you are having issues with Magisk.

I also recommend checking on the latest Magisk releases (stable, beta & canary) to see which one works best at this time.

Thanks. I’ll have to update it with the last released Chinese firmware version, too. I’ve just been lazy about doing so.
Many on this thread have provided differing fixes but I’ve lost track. I’ll gladly provide collaboration privileges if anyone wants in on adding more to it and/or making it nicer to look at.

Apart from those 2 typos I find the guide to be well formatted and clear. Don’t fix what ain’t broken.
Going by how no one else raised any of my issues they are probably not something to be considered in the guide either.
Perhaps make it a rule to only add something if its brought up more than once?
Feel free to use my screenshots if it helps, though.

As an update:

E:2106AllDCubeplatform-tools>fastboot flash system_a "..caos-v310-210721-arm64-bgZ.img"
Invalid sparse file format at header magic
Sending sparse 'system_a' 1/6 (524208 KB)          OKAY [ 11.494s]
Writing 'system_a'                                 OKAY [  0.000s]
Sending sparse 'system_a' 2/6 (519136 KB)          OKAY [ 12.680s]
Writing 'system_a'                                 OKAY [  0.000s]
Sending sparse 'system_a' 3/6 (522476 KB)          OKAY [ 12.182s]
Writing 'system_a'                                 OKAY [  0.000s]
Sending sparse 'system_a' 4/6 (522310 KB)          OKAY [ 11.697s]
Writing 'system_a'                                 OKAY [  0.000s]
Sending sparse 'system_a' 5/6 (524001 KB)          OKAY [ 12.004s]
Writing 'system_a'                                 OKAY [  0.001s]
Sending sparse 'system_a' 6/6 (286198 KB)          OKAY [  6.931s]
Writing 'system_a'                                 OKAY [  0.001s]
Finished. Total time: 75.134s

E:2106AllDCubeplatform-tools>fastboot flash system_b "..caos-v310-210721-arm64-bgZ.img"
Invalid sparse file format at header magic
Sending sparse 'system_b' 1/6 (524208 KB)          OKAY [ 11.562s]
...

Which predictably resulted in a corrupted tablet. So I’ll try crdrom-v308-210521-arm64-bvZ next and see where that takes me

Otherwise this is probably the time to start messing with partitions, isn’t it?

ONWARDS TO GLORY!

Apart from those 2 typos I find the guide to be well formatted and clear. Don’t fix what ain’t broken.
Going by how no one else raised any of my issues they are probably not something to be considered in the guide either.
Perhaps make it a rule to only add something if its brought up more than once?
Feel free to use my screenshots if it helps, though.

I’ll look through your screenshots & see what can be added or help improve the guide.

Generally speaking, I’m glad that my struggles (many hours of anger & angst) have helped many. My original intent was to document everything I went through & add what people had discovered with my guide.

As an update:

E:2106AllDCubeplatform-tools>fastboot flash system_a "..caos-v310-210721-arm64-bgZ.img"
Invalid sparse file format at header magic
Sending sparse 'system_a' 1/6 (524208 KB)          OKAY [ 11.494s]
Writing 'system_a'                                 OKAY [  0.000s]
Sending sparse 'system_a' 2/6 (519136 KB)          OKAY [ 12.680s]
Writing 'system_a'                                 OKAY [  0.000s]
Sending sparse 'system_a' 3/6 (522476 KB)          OKAY [ 12.182s]
Writing 'system_a'                                 OKAY [  0.000s]
Sending sparse 'system_a' 4/6 (522310 KB)          OKAY [ 11.697s]
Writing 'system_a'                                 OKAY [  0.000s]
Sending sparse 'system_a' 5/6 (524001 KB)          OKAY [ 12.004s]
Writing 'system_a'                                 OKAY [  0.001s]
Sending sparse 'system_a' 6/6 (286198 KB)          OKAY [  6.931s]
Writing 'system_a'                                 OKAY [  0.001s]
Finished. Total time: 75.134s

E:2106AllDCubeplatform-tools>fastboot flash system_b "..caos-v310-210721-arm64-bgZ.img"
Invalid sparse file format at header magic
Sending sparse 'system_b' 1/6 (524208 KB)          OKAY [ 11.562s]
...

Which predictably resulted in a corrupted tablet. So I’ll try crdrom-v308-210521-arm64-bvZ next and see where that takes me

Otherwise this is probably the time to start messing with partitions, isn’t it?

ONWARDS TO GLORY!

Some of the ROM images do that but can still be bootable. You’ll have to test them out.

Once you flash any system image, expanding the system partition is highly recommended in order to flash any TWRP-based ZIPs like gapps.