Draft: samsung-milletwifi: new device
Wiki page: https://wiki.postmarketos.org/wiki/Samsung_Galaxy_Tab_4_8.0_Wi-Fi_(samsung-milletwifi)
Device currently just bootloops. Not certain way right now, need to look into getting some sort of logging happening.
However, I'm also unclear if using heimdall
is the right approach here or if I should use lk2nd
instead, as other devices with this chipset use lk2nd
as well, and I already added support for this device to it. Is there a suggested approach here?
Also, there is upstream work for matisse, which is what I was basing some of this work on since it's such a similar device, but I don't think that's ready to use instead of the LineageOS downstream.
Merge request reports
Activity
added 10 commits
-
a244dcd8...d757a368 - 9 commits from branch
postmarketOS:master
- 374f94b4 - samsung-millet: new device
By Bryant Mairs on 2023-09-26T08:45:36
-
a244dcd8...d757a368 - 9 commits from branch
I'm getting errors where the dtb can't be found in the kernel sources I linked:
20:07:43.286035 == Using boot-deploy to finalize/install files == ==> kernel: device-tree blob operations ERROR: Unable to find msm8226/msm8226-sec-milletwifieur-r05.dtb in the following locations: - /boot/dtbs* - /usr/share/dtb/ ==> initramfs: creating boot.img ERROR: Unable to find msm8226/msm8226-sec-milletwifieur-r05.dtb in the following locations: - /boot/dtbs* - /usr/share/dtb/ 20:07:43.588377 boot-deploy completed in: 0.30s 20:07:43.588473 exit status 1 20:07:43.588883 boot-deploy failed 20:07:43.590212 mkinitfs completed in: 1.03s
How do I debug this? It looks like the kernel build is generating dtbs that don't match the naming I expect. I should be able to look at the output build directory to find these, yeah?
By Bryant Mairs on 2023-09-27T07:49:22
The dtbs are a little hard to decipher in those directories. I worked with Skye in the chatroom to try to extract a DTB from the LineageOS 18.1 images you can find, but they don't seem to be in there.
Another approach that could make sense is to build the DTB manually as part of the msm8226 mainline effort since it should be very similar to the
matisselte
work that's already going on there.Not certain which is the best path forward.
By Bryant Mairs on 2023-09-27T07:56:37
added 9 commits
-
0639a31e...438471fe - 8 commits from branch
postmarketOS:master
- cc10a19e - samsung-millet: new device
By Bryant Mairs on 2023-09-28T18:47:37
-
0639a31e...438471fe - 8 commits from branch
Got some help in chat:
deviceinfo_bootimg_qcdt
&deviceinfo_dtb
are incompatible. So that was the cause of my boot issue above. However, flashing the generated image from this just results in a boot loop. My computer doesn't recognize it via USB (BTW `udevadm monitor` is a pretty convenient way to check for USB connections/disconnections live!), so the kernel isn't booting up even that far. Tried the first debugging step of turning offDEBUG_VT
in kconfig, but the boot looping is still happening.For reference, there is a TWRP image and an LineageOS image that does work. They differ in the cmdline options a big (as seen by
pmbootstrap bootimg_analyze
:TWRP:
console=null androidboot.hardware=qcom msm_rtb.filter=0x37 buildvariant=eng
LineageOS:
androidboot.hardware=qcom user_debug=23 msm_rtb.filter=0x37 zcache.enabled=1 zcache.compressor=lz4 buildvariant=userdebug
I'm using the LineageOS DTBs here, as it's the kernel I found for that, so I assume that's correct. I wonder if zcache or lz4 could be an issue. Might have more success with the TWRP options maybe?
By Bryant Mairs on 2023-09-29T18:11:12
mentioned in merge request !4495 (merged)
By Bryant Mairs on 2023-10-19T16:22:38
Closed in favor of !4495 (merged)
By Bryant Mairs on 2023-10-19T16:37:29
mentioned in merge request !5533 (closed)
By Eloy Degen on 2024-08-25T20:08:05