[hlte][2014/NOV/20][W03] Yank555.lu kernel v1.1b

Re: [hlte][2014/SEP/10][deCAF] Yank555.lu kernel v1.0f

Postby yank555 » 10 Sep 2014, 22:36

Well, 1.0f seems to bring issues ... use with care, if you feel like testing, else rollback to 1.0e for now...

JP.

Sent from my Nexus 5 running official SlimKAT 7.0 on Yank555.lu v1.0 kernel.
Nexus 5 (aka. Hammerhead) 32Gb
Note 3 SM-N9005 32Gb Proudly eFused
SGS3 GT-I9300 32Gb
HTC Sensation XE
HTC HD2
TF300TG 32Gb

Image
Credits FAdrums !
User avatar
yank555
-----------
 
Posts: 3520
Joined: 11 Mar 2013, 20:51
Has thanked: 924 times
Been thanked: 4608 times

Re: [hlte][2014/SEP/10][deCAF] Yank555.lu kernel v1.0f

Postby capnkrunch » 11 Sep 2014, 05:26

Just got another one. Seems different from the others. Once again I was on chrome but this time my screen got all mangled, but it still seemed like it was responding to touch. I hit the power button at which point it froze and rebooted. Still on lite e with w03 25.

Maybe it would be worth just rebasing even further back. The last preunified build was 1/30 and there are a number of mdss related commits on that day. Maybe just checkout the last commit on the 29th and go from there. Will03 might need to make some changes the rom side too. Sure you lose a lot of under the hood stuff but they're not really enhancements if they hurt stability.

UPDATE: literally right after I submitted this I realized I forgot the last_kmsg and got a reboot. I think that one was what 1.0f was trying to fix (there's a pingpong timeout message). Here's both for good measure. I'm going to try switching browsers for now since every single reboot I've had on deCAF was while using Chrome.
Attachments
last_kmsg2.zip
(12.12 KiB) Downloaded 712 times
last_kmsg.zip
(90.83 KiB) Downloaded 686 times
capnkrunch
 
Posts: 22
Joined: 03 Jul 2014, 19:41
Has thanked: 2 times
Been thanked: 5 times

Re: [hlte][2014/SEP/10][deCAF] Yank555.lu kernel v1.0f

Postby yank555 » 11 Sep 2014, 11:03

capnkrunch wrote:Just got another one. Seems different from the others. Once again I was on chrome but this time my screen got all mangled, but it still seemed like it was responding to touch. I hit the power button at which point it froze and rebooted. Still on lite e with w03 25.

Maybe it would be worth just rebasing even further back. The last preunified build was 1/30 and there are a number of mdss related commits on that day. Maybe just checkout the last commit on the 29th and go from there. Will03 might need to make some changes the rom side too. Sure you lose a lot of under the hood stuff but they're not really enhancements if they hurt stability.

UPDATE: literally right after I submitted this I realized I forgot the last_kmsg and got a reboot. I think that one was what 1.0f was trying to fix (there's a pingpong timeout message). Here's both for good measure. I'm going to try switching browsers for now since every single reboot I've had on deCAF was while using Chrome.


I have an ulite (as in ultralight) where the rebase goes back further, but it will give crappy framerates on boot, but will start working fine once you've booted up and turned the screen off once.

The problem is, CAF didn't just get added sometime in January, they changed to kernel base in january, CAF and mdss commits go back very far back in 2013 (stopped looking sometime in octobre 2013).

If you want, I can upload the ulite, but I didn't help with these crashes on my phone and since it's crappy framerate at boot made me keep it private so far...

JP.

EDIT : Both logs are pingpong timeout...
Nexus 5 (aka. Hammerhead) 32Gb
Note 3 SM-N9005 32Gb Proudly eFused
SGS3 GT-I9300 32Gb
HTC Sensation XE
HTC HD2
TF300TG 32Gb

Image
Credits FAdrums !
User avatar
yank555
-----------
 
Posts: 3520
Joined: 11 Mar 2013, 20:51
Has thanked: 924 times
Been thanked: 4608 times

Re: [hlte][2014/SEP/10][deCAF] Yank555.lu kernel v1.0f

Postby capnkrunch » 11 Sep 2014, 16:18

yank555 wrote:I have an ulite (as in ultralight) where the rebase goes back further, but it will give crappy framerates on boot, but will start working fine once you've booted up and turned the screen off once.

The problem is, CAF didn't just get added sometime in January, they changed to kernel base in january, CAF and mdss commits go back very far back in 2013 (stopped looking sometime in octobre 2013).

If you want, I can upload the ulite, but I didn't help with these crashes on my phone and since it's crappy framerate at boot made me keep it private so far...

JP.

EDIT : Both logs are pingpong timeout...

No need to upload the ulite if it's still crashing. About CAF, it seems like nonhammerhead snapdragon 800 devices all use CAF as a base. Though the only one I've seen that's independent of CM is Omni's find7 kernel. Why they need a CAF base, I don't know. I wonder if it's possible to just take the Samsung source and work from there or use the the aosp hammerhead code as a base and add the samsung code to that.
capnkrunch
 
Posts: 22
Joined: 03 Jul 2014, 19:41
Has thanked: 2 times
Been thanked: 5 times

Re: [hlte][2014/SEP/10][deCAF] Yank555.lu kernel v1.0f

Postby yank555 » 11 Sep 2014, 18:32

capnkrunch wrote:No need to upload the ulite if it's still crashing. About CAF, it seems like nonhammerhead snapdragon 800 devices all use CAF as a base. Though the only one I've seen that's independent of CM is Omni's find7 kernel. Why they need a CAF base, I don't know. I wonder if it's possible to just take the Samsung source and work from there or use the the aosp hammerhead code as a base and add the samsung code to that.


While starting on the Samsung source may be an idea, starting on Hammerhead is mostly not doable (effort not reasonable).

But I have no clue about the ROM-side implication, though, having to use stock libs, no idea...

JP.

Sent from my Nexus 5 running official SlimKAT 7.0 on Yank555.lu v1.0 kernel.
Nexus 5 (aka. Hammerhead) 32Gb
Note 3 SM-N9005 32Gb Proudly eFused
SGS3 GT-I9300 32Gb
HTC Sensation XE
HTC HD2
TF300TG 32Gb

Image
Credits FAdrums !
User avatar
yank555
-----------
 
Posts: 3520
Joined: 11 Mar 2013, 20:51
Has thanked: 924 times
Been thanked: 4608 times

Re: [hlte][2014/SEP/11][deCAF] Yank555.lu kernel v1.0f+

Postby yank555 » 11 Sep 2014, 21:50

While 1.0f+ is 'back' to mdss as in 1.0e, either fauxsound update or intellithermal seem to cause issues with the vibrator ...

Will look into it, but not tonight, too late zand too tired ;)

If it's an issue, stick with 1.0e for now.

JP.

Sent from my Nexus 5 running official SlimKAT 7.0 on Yank555.lu v1.0 kernel.
Nexus 5 (aka. Hammerhead) 32Gb
Note 3 SM-N9005 32Gb Proudly eFused
SGS3 GT-I9300 32Gb
HTC Sensation XE
HTC HD2
TF300TG 32Gb

Image
Credits FAdrums !
User avatar
yank555
-----------
 
Posts: 3520
Joined: 11 Mar 2013, 20:51
Has thanked: 924 times
Been thanked: 4608 times

Re: [hlte][2014/SEP/10][deCAF] Yank555.lu kernel v1.0f

Postby capnkrunch » 12 Sep 2014, 04:01

yank555 wrote:While starting on the Samsung source may be an idea, starting on Hammerhead is mostly not doable (effort not reasonable).

But I have no clue about the ROM-side implication, though, having to use stock libs, no idea...

JP.

Sent from my Nexus 5 running official SlimKAT 7.0 on Yank555.lu v1.0 kernel.

Hmmm... the reason I mentioned hammerhead was because I think it uses the AOSP branches of qcom_media audio and display so I wondered if that could be used since the kernel drivers are already working. I wonder if some of Samsung's TW crap is in the stock libs and that's why everyone uses CAF. I feel like everything would be so much easier if Samsung used git like Sony does instead of just releasing tarballs; not to mention how much of a pain it is that you need to download a 1+gb file for the 100 or so mb that the kernel source is at their painfully slow speeds. I've been working on porting Omni to hlte and CM's kernel is a mess that I don't think can be fixed with a couple reverts and merging code from a tarball is just impractical (there were over 1000 files with conflicts and most had multiple conflicts and a lot, especially in the drivers, are difficult to resolve). It's like they're saying "we release this because we legally have to but we're not going to make it easy for you). Actually, it got me wondering if maybe the mdss crashes are due to a merge derp but that seems unlikely if the oneplus one has the same issue. That being said I've gained an appreciation for the cm devs porting so many Samsung devices. Really, it's pretty amazing we get AOSP at all buggy as it may be.

EDIT: I was thinking, if your ulite kernel was still crashes maybe the cause is ROM side. I never had crashes on preunified CM and it may have just been luck but I used it for at least a month straight (plus assorted other times for less than that). If rebasing the kernel back that far didn't fix them maybe it's because will03 needs to rebase/revert his repos further back as well. Maybe just try a build with everything fully rebased to before 1/30. I think even as an experiment and not an official release it would give you a lot of important information.
capnkrunch
 
Posts: 22
Joined: 03 Jul 2014, 19:41
Has thanked: 2 times
Been thanked: 5 times

Re: [hlte][2014/SEP/11][deCAF] Yank555.lu kernel v1.0f+

Postby yank555 » 12 Sep 2014, 07:11

capnkrunch wrote:Hmmm... the reason I mentioned hammerhead was because I think it uses the AOSP branches of qcom_media audio and display so I wondered if that could be used since the kernel drivers are already working. I wonder if some of Samsung's TW crap is in the stock libs and that's why everyone uses CAF. I feel like everything would be so much easier if Samsung used git like Sony does instead of just releasing tarballs; not to mention how much of a pain it is that you need to download a 1+gb file for the 100 or so mb that the kernel source is at their painfully slow speeds. I've been working on porting Omni to hlte and CM's kernel is a mess that I don't think can be fixed with a couple reverts and merging code from a tarball is just impractical (there were over 1000 files with conflicts and most had multiple conflicts and a lot, especially in the drivers, are difficult to resolve). It's like they're saying "we release this because we legally have to but we're not going to make it easy for you). Actually, it got me wondering if maybe the mdss crashes are due to a merge derp but that seems unlikely if the oneplus one has the same issue. That being said I've gained an appreciation for the cm devs porting so many Samsung devices. Really, it's pretty amazing we get AOSP at all buggy as it may be.

EDIT: I was thinking, if your ulite kernel was still crashes maybe the cause is ROM side. I never had crashes on preunified CM and it may have just been luck but I used it for at least a month straight (plus assorted other times for less than that). If rebasing the kernel back that far didn't fix them maybe it's because will03 needs to rebase/revert his repos further back as well. Maybe just try a build with everything fully rebased to before 1/30. I think even as an experiment and not an official release it would give you a lot of important information.


Oh the issue is not about reverting further, the kernel base changed between pre unified and unified, if you go back to 1/30 in the current kernel repo you will have a totally different kernel than the pre-unified hltexx. The current kernel is not a evolution of the hltexx !

And yes I confirm, hltexx was ROCK SOLID !

A solution could be to make the hltexx kernel boot on the new bootloader. Will already did an updated ROM based on hltexx, but it only boots on the old bootloader, obviously.

I've had a look, but since it's a totally different kernel base, it's like diffing 2 tarballs :(

Btw, Samsung releases tarballs because it has to, they wouldn't release anything if they could. Almost 2 years back, they promised to use git 'by the end of the year' ... nice story, maybe they were talking about Sony :D

JP.

Sent from my 'proudly eFused' Note 3 (n9005) running W03 deCAF SlimKAT v26 on Yank555.lu deCAF W03 v1.0f+ full kernel.
Nexus 5 (aka. Hammerhead) 32Gb
Note 3 SM-N9005 32Gb Proudly eFused
SGS3 GT-I9300 32Gb
HTC Sensation XE
HTC HD2
TF300TG 32Gb

Image
Credits FAdrums !
User avatar
yank555
-----------
 
Posts: 3520
Joined: 11 Mar 2013, 20:51
Has thanked: 924 times
Been thanked: 4608 times

Re: [hlte][2014/SEP/11][deCAF] Yank555.lu kernel v1.0f+

Postby Xavihernandez » 12 Sep 2014, 13:22

Hi @yank555

Here another reboot while sending SMS.
Attached last_kmsg.
Let me know if you do not need last_image for the moment.

I'm gonna flash v1.0f+ asap.
Attachments
2014-09-12_14.17-kmsg.zip
(87.05 KiB) Downloaded 722 times
Device: Samsung SM-N9005 (Galaxy Note 3)
ROM: CM12 by Temasek (v3.3)
Kernel: yank555.lu v1.1c-beta6 for CM12
Modem: N9005XXUBMI6
Xavihernandez
 
Posts: 118
Joined: 22 Jan 2014, 15:58
Has thanked: 96 times
Been thanked: 16 times

Re: [hlte][2014/SEP/11][deCAF] Yank555.lu kernel v1.0f+

Postby yank555 » 12 Sep 2014, 14:47

Well I'm always interested in logs, as maybe in one I may see something that might point me in the right direction ...

But if you see this :

Code: Select all
[11901.850081] 108a0: 000000[11901.866559] [1:       mdss_fb0:  519] Kernel panic - not syncing: Pingpong Timeout


Pretty 1-2 pages before the end, it's the usual mdss crash that's happenig to so many of us ...

So far, I've not yet found anything pointing me in the right direction :(

JP.
Nexus 5 (aka. Hammerhead) 32Gb
Note 3 SM-N9005 32Gb Proudly eFused
SGS3 GT-I9300 32Gb
HTC Sensation XE
HTC HD2
TF300TG 32Gb

Image
Credits FAdrums !
User avatar
yank555
-----------
 
Posts: 3520
Joined: 11 Mar 2013, 20:51
Has thanked: 924 times
Been thanked: 4608 times

PreviousNext

Return to Kernels

Who is online

Users browsing this forum: No registered users and 11 guests

cron