[DEV] CyanogenMod 10.2 - HTC Droid DNA

This is for developers only. No builds will be posted in this thread.
https://github.com/CM10DNA
Any valid development related discussion can be posted here. Please no posts that aren't directly related to development (i.e. comments, thanks, etc.).

---

I thought I would share an update on how this very experimental code-base is doing.
The code is running again (much of yesterday it was crashing on startup).
I have been running dirty flashing from my CM 10.1 build. It seems like that was safe to do so. I tried doing a clean flash this morning but sadly, my clean flash seems to have broken cell service. No phone or data. Restoring my cm 10.1 backup of data to the same install fixed it. Strange. (This included the change pushed this morning that updates the version of the apns file).
Superuser is now working but may be a little flaky. For example, on my clean install Titanium Backup couldn't acquire root access until after I rebooted.
Camera is still not working. It says "failed to connect to camera". I haven't looked at it at all, but there is a massive commit in the M7 kernel pulling source from the HTC One Google Edition for the camera.
Overall, I'm finding it's either completely broken or working very well. I think that the cellular network issue is going to get resolved upstream. I am afraid that we might be on our own for the camera.

---

crpalmer said:
I thought I would share an update on how this very experimental code-base is doing.
The code is running again (much of yesterday it was crashing on startup).
I have been running dirty flashing from my CM 10.1 build. It seems like that was safe to do so. I tried doing a clean flash this morning but sadly, my clean flash seems to have broken cell service. No phone or data. Restoring my cm 10.1 backup of data to the same install fixed it. Strange. (This included the change pushed this morning that updates the version of the apns file).
Superuser is now working but may be a little flaky. For example, on my clean install Titanium Backup couldn't acquire root access until after I rebooted.
Camera is still not working. It says "failed to connect to camera". I haven't looked at it at all, but there is a massive commit in the M7 kernel pulling source from the HTC One Google Edition for the camera.
Overall, I'm finding it's either completely broken or working very well. I think that the cellular network issue is going to get resolved upstream. I am afraid that we might be on our own for the camera.
Click to expand...
Click to collapse
https://github.com/CyanogenMod/andr...mmit/ec27077d6497c66f52488126ef6b181ef3fbed0d
seems cm has stopped camera building throughout probably prepping for the focal merge
edit: idk if m7 has cam working but it hasn't been detected as a bug
"Current bugs (they are known):
SMS sending will not work (receiving works)
Dialer sub-submenu FC's the dialer
Keyboard FC's when gesture swiping"

---

Not sure how much this helps. But the m7 kernel recently got its camera code updated with the Google edition. http://review.cyanogenmod.org/#/c/46164/
With that so. They are able to use the 4.2 libs http://review.cyanogenmod.org/#/c/46299/

---

Flyhalf205 said:
Not sure how much this helps. But the m7 kernel recently got its camera code updated with the Google edition. http://review.cyanogenmod.org/#/c/46164/
With that so. They are able to use the 4.2 libs http://review.cyanogenmod.org/#/c/46299/
Click to expand...
Click to collapse
we'd probably have to merge their whole drivers/media/video/msm and use their camera binaries at least that's how it works with sense if you want to use their 4.2.2 cam libs

---

JoelZ9614 said:
we'd probably have to merge their whole drivers/media/video/msm and use their camera binaries at least that's how it works with sense if you want to use their 4.2.2 cam libs
Click to expand...
Click to collapse
And their commit conflicts all over the place which is going to make it hard to cherry pick it. This was the commit I referred to earlier today.
On the plus side, the ril issue looks like it will get resolved for us. There is a commit out there for Samsung hardware that adds a couple of RIL events that look like they have been added.

---

crpalmer said:
And their commit conflicts all over the place which is going to make it hard to cherry pick it. This was the commit I referred to earlier today.
On the plus side, the ril issue looks like it will get resolved for us. There is a commit out there for Samsung hardware that adds a couple of RIL events that look like they have been added.
Click to expand...
Click to collapse
What about the commit introducing the HTCCDMAQualcommRIL?
Sent from my HTC6435LVW using xda app-developers app

---

times_infinity said:
What about the commit introducing the HTCCDMAQualcommRIL?
Sent from my HTC6435LVW using xda app-developers app
Click to expand...
Click to collapse
Which commit?

---

crpalmer said:
Which commit?
Click to expand...
Click to collapse
http://review.cyanogenmod.org/#/c/45607/
Hope it helps.
Sent from my HTC6435LVW using xda app-developers app

---

I'll take a look at the radio stuff tomorrow. I have a couple ideas.

---

Deck got data working on the m7 (Sprint), check the commits or wait until tomorrow if he hasn't pushed them yet
Sent from my buttered S3

---

chad0989 said:
I'll take a look at the radio stuff tomorrow. I have a couple ideas.
Click to expand...
Click to collapse
To be clear, it is working fine for me if I dirty flash. I have been running 10.2 builds for 48 hours everywhere from strong wifi to bad 3G in the middle of a forest.
It's only breaking if I do a factory reset before flashing a build. That's why my guess was that it was related to this event being missing:
cyanogenmod.org/#/c/46280/1/libril/ril.cpp
(around line 2240) and that a similar change was going to be required to our libril. My inclination would be to wait and see what happens with the M7 builds (which it sounds are coming along) and see if it just magically fixes itself for us....
Edited to add: sending of text messages wasn't working but looks like they were supposed fixed last night (I haven't built it to test it yet)

---

crpalmer said:
And their commit conflicts all over the place which is going to make it hard to cherry pick it. This was the commit I referred to earlier today.
On the plus side, the ril issue looks like it will get resolved for us. There is a commit out there for Samsung hardware that adds a couple of RIL events that look like they have been added.
Click to expand...
Click to collapse
mm actually now that i look at it drivers/media/video/msm has been left basically untouched in our source updating to what the m7 has shouldn't be too much of a problem

---

JoelZ9614 said:
mm actually now that i look at it drivers/media/video/msm has been left basically untouched in our source updating to what the m7 has shouldn't be too much of a problem
Click to expand...
Click to collapse
I tried a cherry-pick before and it looked very bad. But, actually, a lot of it looks like it could be related to the comments that HTC strips from the source when the release that I added back (because it makes it easier to patch from all non-HTC sources).
If you want to take a crack at this, try doing this before cherry-picking their commit:
git show c81741a8337c2342d856ffeac0cc087452729290 drivers/media/video/msm include/media/msm_camera.h | patch -p1 -R
That will strip all the comments back out again and get rid of all those conflicts. The conflicts then look very simple (outside of arch/arm/mach-msm, which is where it gets much harder).

---

crpalmer said:
I tried a cherry-pick before and it looked very bad. But, actually, a lot of it looks like it could be related to the comments that HTC strips from the source when the release that I added back (because it makes it easier to patch from all non-HTC sources).
If you want to take a crack at this, try doing this before cherry-picking their commit:
git show c81741a8337c2342d856ffeac0cc087452729290 drivers/media/video/msm include/media/msm_camera.h | patch -p1 -R
That will strip all the comments back out again and get rid of all those conflicts. The conflicts then look very simple (outside of arch/arm/mach-msm, which is where it gets much harder).
Click to expand...
Click to collapse
the board-monarudo-camera.c doesn't need to be updated, at least i didn't update it in my m7 kernel port
but i can't seem to get the kernel to build
drivers/gpu/msm/adreno_snapshot.c: In function 'snapshot_rb':
drivers/gpu/msm/adreno_snapshot.c:628: sorry, unimplemented: inlining failed in call to 'parse_ib': recursive inlining
drivers/gpu/msm/adreno_snapshot.c:588: sorry, unimplemented: called from here
make[3]: *** [drivers/gpu/msm/adreno_snapshot.o] Error 1
make[2]: *** [drivers/gpu/msm] Error 2
make[1]: *** [drivers/gpu] Error 2
make[1]: *** Waiting for unfinished jobs....
i don't believe it's related to any edits i made tho heres my commit https://github.com/Joelz9614/android_kernel_htc_dlx/commit/daa90f2bf8e64a8ad6f4f497ca412bc0a77ab7ef

---

JoelZ9614 said:
the board-monarudo-camera.c doesn't need to be updated, at least i didn't update it in my m7 kernel port
but i can't seem to get the kernel to build
drivers/gpu/msm/adreno_snapshot.c: In function 'snapshot_rb':
drivers/gpu/msm/adreno_snapshot.c:628: sorry, unimplemented: inlining failed in call to 'parse_ib': recursive inlining
drivers/gpu/msm/adreno_snapshot.c:588: sorry, unimplemented: called from here
make[3]: *** [drivers/gpu/msm/adreno_snapshot.o] Error 1
make[2]: *** [drivers/gpu/msm] Error 2
make[1]: *** [drivers/gpu] Error 2
make[1]: *** Waiting for unfinished jobs....
i don't believe it's related to any edits i made tho heres my commit https://github.com/Joelz9614/android_kernel_htc_dlx/commit/daa90f2bf8e64a8ad6f4f497ca412bc0a77ab7ef
Click to expand...
Click to collapse
Searching in the browser doesn't have any matches for adreno_snapshot...
Maybe try git checkout -b tmp HEAD~1 and make sure that builds?

---

Looks like CM just merged an updated audio HAL along with the legacy HAL, both of which incorrectly detect our device for now. Building with a fix now and if all is well I'll submit upstream.
Edit: It's going to cause a bunch of routing issues again also until some other stuff gets merged in. Best to just revert their change for now for builds.

---

chad0989 said:
Looks like CM just merged an updated audio HAL along with the legacy HAL, both of which incorrectly detect our device for now. Building with a fix now and if all is well I'll submit upstream.
Edit: It's going to cause a bunch of routing issues again also until some other stuff gets merged in. Best to just revert their change for now for builds.
Click to expand...
Click to collapse
Aside from the routing issues, this may already be fixed for us thanks to the m7:
http://review.cyanogenmod.org/#/c/46473/

---

@chad0989, FYI, I reverted the commit to use legacy alsa early this morning and haven't noticed any problems (play music on Bluetooth, voice calls with and without wired headset) using that build.
You may want to give it a try and see if you're ready to revert it.
Edit: actually, I just noticed that they had also enabled legacy audio in msm8960-common but since reverted that. So, I guess I was still running the legacy mode...

---

crpalmer said:
@chad0989, FYI, I reverted the commit to use legacy alsa early this morning and haven't noticed any problems (play music on Bluetooth, voice calls with and without wired headset) using that build.
You may want to give it a try and see if you're ready to revert it.
Edit: actually, I just noticed that they had also enabled legacy audio in msm8960-common but since reverted that. So, I guess I was still running the legacy mode...
Click to expand...
Click to collapse
They aren't allowing new HAL to build yet (see https://github.com/CyanogenMod/android_hardware_qcom_audio-caf/blob/cm-10.2/Android.mk) But we will need that flag when they allow the new HAL to build. I added it preemptively to hopefully avoid breakage. The issue was that the legacy HAL they pushed hadn't included some changes for our CSD client and platform detection. Looking at what was committed last night/today it should work now though.