• Introducing XDA Computing: Discussion zones for Hardware, Software, and more!    Check it out!
  • Fill out your device list and let everyone know which phones you have!    Edit Your Device Inventory

Will there be a Custom Rom?

Search This thread

alibei

Senior Member
Jan 1, 2017
364
521
22
Deggendorf
@Hlcpereira I don't know if it will help you but in the stock kernel there is CONFIG_NO_DELAY_DMA_UNMAP_CAMERA=y enabled. In your kernel it is missing. I found it out by comparing your defconfig and the extracted one from the stock kernel.
 
Last edited:

Hlcpereira

Recognized Developer
@Hlcpereira I don't know if it will help you but in the stock kernel there is CONFIG_NO_DELAY_DMA_UNMAP_CAMERA=y enabled. In your kernel it is missing. I found it out by comparing your defconfig and the extracted one from the stock kernel.

Welp, that just confirms even more my theory that unless Xiaomi do a proper kernel code release getting tucana back cameras to work won't be possible at all with the camera Q blobs at least. I may try again the Pie blobs again, so let's see (And I really hope this time I don't end up looping without a single tracing of what actually caused it on the logs at least, because this camera bs already have me mad enough)
 

alibei

Senior Member
Jan 1, 2017
364
521
22
Deggendorf
Now I extracted the defconfig from an android 9 kernel and didn't find CONFIG_NO_DELAY_DMA_UNMAP_CAMERA. So maybe your approach with the android 9 blobs will work.
 

alibei

Senior Member
Jan 1, 2017
364
521
22
Deggendorf
I tired to use the pie camera blobs and the pie firmware blobs for the camera.

I got some different error like:
camxoverrideeeprom.cpp:137 DualCameraData_FromPersist_IsValid() persist : access fail /mnt/vendor/persist/camera/rear_dc_cal_wt.bin
camxoverrideeeprom.cpp:137 DualCameraData_FromPersist_IsValid() persist : access fail /mnt/vendor/persist/camera/rear_dc_cal_wt_bak.bin
camxoverrideeeprom.cpp:137 DualCameraData_FromPersist_IsValid() persist : access fail /mnt/vendor/persist/camera/rear_dc_cal_tt.bin
camxoverrideeeprom.cpp:137 DualCameraData_FromPersist_IsValid() persist : access fail /mnt/vendor/persist/camera/rear_dc_cal_tt_bak.bin

I attached the full log.
 

Attachments

  • logcat.zip
    261.2 KB · Views: 12

Hlcpereira

Recognized Developer
mi10 pro has the same camera sensor as the note 10 and they have custom roms

1 - Wrong, they have a similar sensor but not exatcly the same
2 - They are using prebuilt vendor and kernel, while I'm building custom vendor and using a cleanup kernel without Xiaomi ricing or bs. (No I won't go down this path although I have a vendorless and prebuilt kernel tree on my tucana tree repo, but it didn't boot so feel free to make it boot using that tree)
3 - They are a 4.19 kernel device, which is completely different from the kernel 4.14
 

Kosecki99

Senior Member
Dec 9, 2016
173
14
1 - Wrong, they have a similar sensor but not exatcly the same
2 - They are using prebuilt vendor and kernel, while I'm building custom vendor and using a cleanup kernel without Xiaomi ricing or bs. (No I won't go down this path although I have a vendorless and prebuilt kernel tree on my tucana tree repo, but it didn't boot so feel free to make it boot using that tree)
3 - They are a 4.19 kernel device, which is completely different from the kernel 4.14

Henrique, Could you make a custom kernel flashable by twrp to try to port a custom rom?...thank you
 

Hlcpereira

Recognized Developer
I tired to use the pie camera blobs and the pie firmware blobs for the camera.

I got some different error like:
camxoverrideeeprom.cpp:137 DualCameraData_FromPersist_IsValid() persist : access fail /mnt/vendor/persist/camera/rear_dc_cal_wt.bin
camxoverrideeeprom.cpp:137 DualCameraData_FromPersist_IsValid() persist : access fail /mnt/vendor/persist/camera/rear_dc_cal_wt_bak.bin
camxoverrideeeprom.cpp:137 DualCameraData_FromPersist_IsValid() persist : access fail /mnt/vendor/persist/camera/rear_dc_cal_tt.bin
camxoverrideeeprom.cpp:137 DualCameraData_FromPersist_IsValid() persist : access fail /mnt/vendor/persist/camera/rear_dc_cal_tt_bak.bin

I attached the full log.

Ok, I read the logcat and I can see that other than those that you wrote down it has a lot of dlopen errors too, which are killing the camera service HAL, whether cause of missing blobs or missing symbols.
 

gomorin

Member
Jan 28, 2017
26
2
I've talked to CrDroid developers, and they said that we can make a crowdfunding for them to buy the device and develop CrDroid for us, now I can put 80 dollars but we need more than that. Reply to this message if you are interested in that.
 

alibei

Senior Member
Jan 1, 2017
364
521
22
Deggendorf
For some reason patchelf is not patching com.vidhance.node.eis.so correctly.
In the logcat I'm seeing:
camxosutilslinux.cpp:874 LibMap() dlopen: dlopen failed: "/vendor/lib64/camera/components/com.vidhance.node.eis.so" loaded phdr 0x79460df040 not in loadable segment, on '/vendor/lib64/camera/components/com.vidhance.node.eis.so'

I attached the logcat.
 

Attachments

  • logcat.zip
    261.2 KB · Views: 5

Hlcpereira

Recognized Developer
I saw that you updated your source i will test this with android 9 kernel blobs.

For some reason patchelf is not patching com.vidhance.node.eis.so correctly.
In the logcat I'm seeing:
camxosutilslinux.cpp:874 LibMap() dlopen: dlopen failed: "/vendor/lib64/camera/components/com.vidhance.node.eis.so" loaded phdr 0x79460df040 not in loadable segment, on '/vendor/lib64/camera/components/com.vidhance.node.eis.so'

I attached the logcat.

Yeah. I'm trying to understand what happened, because I did that back on Q bringup for Dipper and it worked just fine. I don't know if it's trying to get from a different path or what, anyway, I'll put the lib patched on ghidra later on to see what is happening on it.
 

alibei

Senior Member
Jan 1, 2017
364
521
22
Deggendorf
Just out of curiosity I deleted:
vendor/lib/camera/components/com.qti.node.watermark.so
vendor/lib64/camera/components/com.qti.node.watermark.so
vendor/lib64/camera/components/com.vidhance.node.eis.so
vendor/lib64/camera/components/com.vidhance.stats.aec_dmbr.so
With that I don't get the error:
dlopen failed: cannot locate symbol "_ZN7android13GraphicBuffer4lockEjPPv" and dlopen failed: cannot locate symbol "__sfp_handle_exceptions"
But I'm getting the error:
Probe eeprom tucana_semco_m24c128x_s5khmx_8plens failed, expected vendor id:8f, read vendor id:3
 

Attachments

  • logcat.zip
    574.2 KB · Views: 1

Top Liked Posts

  • There are no posts matching your filters.
  • 15
    Ok, a bit of updates, mainly because I do feel that you guys deserve.

    First one: I'm going to temporary close the crowdfunding. "Why? The goal wasn't reached yet neither is being one and a half month" I know, yet @Genomey could found a cheap Mi Note 10 on eBay and got it for about 252 USD. I'm going to wait for him get the device on his end and check and if everything is ok I'll send the money to him while he send me the device. "What will happen with the remaining amount?" 63.74 USD (or 347.01 BRL) are from me, I already had it or got it by my own in the meanwhile.

    The total amount donated was of 267.22 USD. (1454.61 BRL).

    Second is: In the meanwhile me and @0201m were working on the device sources and testing (huge shout out for him for helping in testing and troubleshooting) and so far: Kernel boots, everything builds, but yet depending of which blobs/vendor binaries you use the device reboots to recovery. Also when we managed to make it boot, the display service was dead (which means it's stuck on the Mi logo while the system already booted) XD (Not joking at all, me and him almost lost our minds cause of that ;_; ). At least the vibrator is confirmed to be working. So far that's the progress.
    8
    https://forum.xda-developers.com/mi-note-10/how-to/custom-development-t4018813

    In short:there are same some efforts trying to get custom Roms but nothing groundbreaking yet.
    I am working on the device tree right now to compile from source. Not ready yet and don't have the device in hand right now (should arrive soon) s.o. else is trying with treble images.
    7
    Any progress with the ROM? ... looking forward to trying at least one beta

    Still trying to figure out about the main cam sensor. Took those days off due the fact I am a student after all and had a bunch of works from collage to do. As I said to another people: until I get, at least, the main cam working from the not working list (which also includes NFC and proximity sensor), I won't do a public release and that's final. All the other camera sensors are being registered properly FYI.
    7
    @Hlcpereira How's the progress btw ?
    Well. I got the device already, unlocked it too. Main camera, IR and NFC are dead at the moment. Rest is working

    Sent from my MI 8 using XDA Labs
    6
    Update of the situation (again): About the custom ROM start up progress: As it seems to be clear a lead was taken on what's wrong about the display (finally and kudos for @alibei for sharing his finds) things should start go more smoothly. I'm rebasing the sources and double checking everything ATM. And about the device that @Genomey found on eBay, well that didn't go well... He then sent 200 USD to covering up money for me buying a tucana here at Brazil, and so do it, I did start looking already for a device in here since then. I should be be buying definately around this week (that's my hopes at least)