[Kernel][Stock1.3][Already included in CM7] Dalingrin's OC kernel [09/01/11]

Status
Not open for further replies.
Search This thread

skwalas

Senior Member
Feb 22, 2011
259
56
I do not believe that the CPU scheduler is causing the lack of deep sleep. I've had reports of this from both the stock and the overclock kernel(stock uses CFS).
BFS got a bad wrap in the Android world early on because BFS was still very new at the time. It has matured greatly since then.
I am wondering if the stock examples are saused by something different, specifically the nook thinking its still charging when not connected. I just found it this morning in that state again (at least the second time I know of) and since it doesn't sleep on charger, obviously no deep sleep. (And before some idiot starts bashing me, no it was NOT on the charger, just acting as if it was: in battery use and in spare parts.

It also still seems strange that on vanilla nightlies, deep sleep works immediately and flawlessly for me, but I lose it without fail every time I put the oc kernel on.

Still awaiting suggestions from anyone on what to try capturing in my oc vs non-oc states here...
 
Last edited:

tsukaza

Senior Member
Aug 31, 2010
282
44
I installed CPU SPY and it creamed my NC, I have to start from 10.1 and went thru the whole process to get it backup. It happened again after I rebuild the NC, so I know it might be CPU SPY doing something weird to the system. The NC is now fine, using the 4.5.25 and OC kernel without CPU SPY.
 

MedLine

Senior Member
Feb 27, 2011
197
29
... deep sleep ... what to try capturing in my oc vs non-oc states here...
maybe like what madcat posted here, to get same thing you can use terminal app:
Code:
su
logcat -d > /mnt/sdcard/logcat.txt
dmesg > /mnt/sdcard/dmesg.txt
cat proc/wakelocks > /mnt/scard/wakelocks.txt
files will be on sdcard. explanation of what numbers mean in the wakelocks file here: http://lkml.org/lkml/2010/8/6/581
 

rhester72

Senior Member
Mar 30, 2011
418
63
I installed CPU SPY and it creamed my NC, I have to start from 10.1 and went thru the whole process to get it backup. It happened again after I rebuild the NC, so I know it might be CPU SPY doing something weird to the system. The NC is now fine, using the 4.5.25 and OC kernel without CPU SPY.

What?

CPU Spy does nothing more than read entries from the /proc filesystem, which is essentially no different than reading a file. It doesn't actually *DO* anything...no backgrounding, no real calculations...it just reads a file, calculates percentages, and prints it to the screen. It's probably the -least- impactful thing you could run.

Rodney
 

tsukaza

Senior Member
Aug 31, 2010
282
44
Every time I install CPY SPY (twice I tested) and reboot the NC will loop forever on the boot animation. It is what I observed so I reported it here. Now the NC is fine without it. I like the app, it is a nice tool.
 

newellj79

Senior Member
Jan 2, 2011
2,370
296
Every time I install CPY SPY (twice I tested) and reboot the NC will loop forever on the boot animation. It is what I observed so I reported it here. Now the NC is fine without it. I like the app, it is a nice tool.

Weird, I don't see how cpuspy could have that effect.
 

tsukaza

Senior Member
Aug 31, 2010
282
44
Anyone having issue using SETCPU on boot? Maybe that is what is killing my NC. Now I set it after it boots up. I will try CPY SPY in a day or two again and see what happens.
 

knaries2000

Senior Member
Apr 13, 2009
172
12
I do not believe that the CPU scheduler is causing the lack of deep sleep. I've had reports of this from both the stock and the overclock kernel(stock uses CFS).
BFS got a bad wrap in the Android world early on because BFS was still very new at the time. It has matured greatly since then.

You may be right, but so far with my testing, cfs is working wonderfully with regards to deep sleep. So it may be worthwhile for the people who have deep sleep issues to try it. It could very well be just a coincidence for me though.

As for my dislike (maybe too strong of a word) for bfs, it is not just from experience. I read some info on how it works a while ago (it may be stale by now), and I didn't really think it's approach will be good overall. Also I read some comparision between bfs and cfs, generally cfs is more efficient although bfs in some situations has less latency. If you have a system with low utilization, it really doesn't matter which scheduler to use. If you have a system that is heavily loaded, cfs will probably work better even with single cpu, since it used a tree to order the task, which results in O(log n) task lookup time, while bfs has a single queue, that requires O(n) lookup time. Also since cfs throughput is better it will clear up the queue/tree faster. The only situation that I think bfs has over cfs is with bursty workloads which is a valid concern for android mainly because of the garbage collector.

Also, I am not sure if the dalvik vm that we use is cgroups aware, if it is, bfs may actually cause random stalls, although I have not really notice it.
 
J

Jsorvik

Guest
Add me to the list that loses deep sleep with the OC. I have had no problem with deep sleep until I added the OC. I wiped cash, dalvik, and fixed permissions all with no success. Are you looking for more logs? IF so, should I do the following:

su
logcat -d > /mnt/sdcard/logcat.txt
dmesg > /mnt/sdcard/dmesg.txt
cat proc/wakelocks > /mnt/scard/wakelocks.txt

Aside from losing the deep sleep, the performance is spectacular! I love the kernel, but I would also love to recapture deep sleep!
 

knaries2000

Senior Member
Apr 13, 2009
172
12
if you don't mind, why don't you give the cfs version of the OC a try. it is in the prev page. It solved my deep sleep issues, I would like to know if it's just a coincidence or not. If you have issues or don't like it, just reflash the original OC kernel and you are back to the way it was.

Add me to the list that loses deep sleep with the OC. I have had no problem with deep sleep until I added the OC. I wiped cash, dalvik, and fixed permissions all with no success. Are you looking for more logs? IF so, should I do the following:

su
logcat -d > /mnt/sdcard/logcat.txt
dmesg > /mnt/sdcard/dmesg.txt
cat proc/wakelocks > /mnt/scard/wakelocks.txt

Aside from losing the deep sleep, the performance is spectacular! I love the kernel, but I would also love to recapture deep sleep!
 
J

Jsorvik

Guest
Yep. . . switching to the cfs version of the OC Kernel does fix the deep sleep issue for me. Just for grins and giggles, I am going to flash back to the BFS version right away and see what happens. . . I will report back here in a few minutes. . .
 

Divine_Madcat

Retired Forum Moderator
Sep 9, 2005
2,940
2,613
USA
Yep. . . switching to the cfs version of the OC Kernel does fix the deep sleep issue for me. Just for grins and giggles, I am going to flash back to the BFS version right away and see what happens. . . I will report back here in a few minutes. . .

But are you sure that it wasn't the subsequent flash and reboot that fixed the issue? I am on the BFS kernel, and have deep sleep, so it leads to me to doubt BFS as being the culprit.
 
J

Jsorvik

Guest
Nope. . . not nearly intelligent enough about the file systems stuff in android to be sure of anything. . . All I can say is that when I flashed the cfs version of the OC, I regained deep sleep. I subsequently reflashed the BFS version of the OC and lost my deep sleep again.

I am not sure what interactions between my set up and the two kernels are causing the difference, but that is my experience so far. If those of you who are more knowledgeable would like me to try something else, just let me know. With good directions, I am at usually at least as smart as a guinea pig and can try to give you what you need!
 
J

Jsorvik

Guest
Updated to N101 and had same results. No deep sleep with OC BFS and deep sleep working with OC CFS. . ..
 
Last edited:

knaries2000

Senior Member
Apr 13, 2009
172
12
My problem with deep sleep wasn't that I couldn't get to deep sleep at all. It was intermitten, some time, it would other times it would not, and I have no idea when. Some times if I disable wifi, and re-enable it it would allow be to sleep again, but later it would stop. After changing to cfs, I was always able to get to deep sleep.

Also I noticed that with cfs, I get much better cpu scaling distribution. with bfs, even with conservative governor, my nook was pretty much using only 1 speed, usually the 2nd highest one (1100mhz), 90+% of the time, and it rarely used the lower rates. with cfs, scaling was more evenly distributed. So battery seems to burn a little faster in BFS as a result. There is something in bfs that is affecting cpu scaling and probably deep sleep. I don't have a theory on what though..


But are you sure that it wasn't the subsequent flash and reboot that fixed the issue? I am on the BFS kernel, and have deep sleep, so it leads to me to doubt BFS as being the culprit.
 

boxcar8028

Senior Member
Mar 15, 2011
955
220
Seattle, WA
Maybe not. . . . .I better check this more carefully. . .

This deep sleep and SOD bug is a tuff one to pin down. playing with the wifi on and off and rebooting seem to help a little but still can't find a true fix yet. my deep sleep is off and on maybe 1-10% not like my htc aria with 80% min all the time.

cm7 n100 OC 6/09 300 600 800 1100 1200
 

odub303

Senior Member
Nov 30, 2007
1,337
464
I do not believe that the CPU scheduler is causing the lack of deep sleep. I've had reports of this from both the stock and the overclock kernel(stock uses CFS).
BFS got a bad wrap in the Android world early on because BFS was still very new at the time. It has matured greatly since then.

FWIW, I have only ever lost deep sleep while using the OC kernel, and ever since replacing the uimage with the CFS uimage, I have not had any sleep issues, surviving through multiple reboots and heavy usage.
 
Status
Not open for further replies.

Top Liked Posts

  • There are no posts matching your filters.
  • 174
    Using this kernel is at your own risk. Before flashing make backups!!!

    Overview
    -In addition to the general CPU overclock the DSP and possibly the L3 bus are overclocked to levels comparable to other higher end OMAP3 SOCs. This should result in a minor speed up in video decoding among other things. Occip pointed out that the L3 bus is set in the boatloader and kernel values are ignored. I still need to investigate this.
    I feel comfortable with the overvolt used in these kernels for a few reasons. Omap cpus automatically reduce voltage based on load and stability. The voltage gating in omap processors is very advanced. Absolute voltages set in the kernel are only a maximum for a given clock rate. The cpu should seldom reach the defined maximum. In addition the highest overvolt used is very small ~0.05v.​

    Common Features for 2.6.32 builds
    -As of 06/09/11 the CPU clock can be customized via this app: http://forum.xda-developers.com/showthread.php?t=1104039
    -Overclockable to 1200mhz and 1100mhz. Must be set with an app like SetCPU unless running CM7. CM7 has overclocking under Cyanogenmod settings. DO NOT use SetCPU profiles with CM7.
    -****Small overvolt**** compared to the stock 3621 specs. This is running at the same voltage as the omap3630 which is very similar to our CPU.
    -1.2ghz is overvolted slightly more compared to 1.1ghz. Resultant voltage is dependent on your particular hardware. It will attempt to dynamically scale to keep stability.
    -Available clock speeds: 300mhz, 600mhz, 800mhz, 1100mhz, 1200mhz
    -Partial fix for strange touchscreen edge behavior brought to you by deeper-blue.
    -Smartass CPU governor. Very similar to the Interactive governor found in my 2.6.29 kernel but with a built in SetCPU profile-like functionality. CPU clock speed is limited to 3-600mhz when the screen is off.
    -Various performance tweaks.​


    Kernel Downloads


    * This kernel is included in CM7 starting with nightly 132 and will be CM 7.1 stable. There is no need to flash a separate kernel unless you are running Stock Froyo 1.2.
    * 2.6.29 releases are discontinued and no longer supported. They will remain hosted for those that might need them.
    * There are separate builds for CM7 and stock 1.2 Froyo
    * 2.6.32 CM7 builds currently do not work with the stable CM7 builds until CM7.1. You need to use CM nightlies until 7.1 is released.

    * All 2.6.32 kernels require an updated u-boot from B&N 1.2 Froyo. If your Nook has "Read Forever" or a green Cyanogenmod logo on the screen when you boot then you have the latest version. If it says something else then you need to update.

    All kernels hosted here:
    http://kan.gd/uic

    md5sum for kernels:
    Code:
    0c1a3f88da68f17a187e7d6535e3ecef  ./090111/update-Froyo1.2-dalingrin-OC-sd-090111.zip
    5673c886a08577d963be7722528a6865  ./090111/update-Froyo1.2-dalingrin-OC-emmc-090111.zip
    a526a38d41a5b7a069ed5e12b8799815  ./063011/update-Froyo1.2-dalingrin-OC-sd-063011.zip
    a8ba40ac44241543be52854c048b3a72  ./063011/update-Froyo1.2-dalingrin-OC-emmc-063011.zip
    94ff341d338ddcd432afa313f03edaa3  ./063011/update-CM7-dalingrin-OC-sd-063011.zip
    aacfb728b2a650ca5194d4ece097f22e  ./063011/update-CM7-dalingrin-OC-emmc-063011.zip
    130fe72bbffa0a660475c1d360de1548  ./061811/update-Froyo1.2-dalingrin-OC-emmc-061811.zip
    bfc4e9e3f70d45324201bd6a14a44bc0  ./061811/update-CM7-dalingrin-OC-sd-061811.zip
    7156b00dffc5a636cd6d7a53e00abe2c  ./061811/update-CM7-dalingrin-OC-emmc-061811.zip
    ca101e3b9a2874e3521c89453352b0ed  ./061311/update-Froyo1.2-dalingrin-OC-sd-061311.zip
    d17e29cc5a3a758b8a445e4d554e9538  ./061311/update-Froyo1.2-dalingrin-OC-emmc-061311.zip
    1ba2195b53a15d31ef9cd23e9fc5342a  ./061311/update-CM7-dalingrin-OC-sd-061311.zip
    0048009800107dc596a3c06d8df33aff  ./061311/update-CM7-dalingrin-OC-emmc-061311.zip
    adb0afe1d5999ee30e773bf112aa6d76  ./060911/update-Froyo1.2-dalingrin-OC-sd-060911.zip
    7d1c53b29f0513543b515e0c37e4c8cb  ./060911/update-Froyo1.2-dalingrin-OC-emmc-060911.zip
    a770009f2807e04f6761bb315309a9a4  ./060911/update-CM7-dalingrin-OC-sd-060911.zip
    fd6634c0a904d8fc92b9e38583cde5ad  ./060911/update-CM7-dalingrin-OC-emmc-060911.zip
    52b593ddabf70cc84b24c479cdb871f1  ./052311/update-Froyo1.2-dalingrin-OC-sd-052311.zip
    43e498da25b62ec0516471ecfb600eb7  ./052311/update-Froyo1.2-dalingrin-OC-emmc-052311.zip
    5732fa2ef134d351ec98635e4beb8128  ./052311/update-CM7-dalingrin-OC-sd-052311.zip
    d6dac86b336e4ca0f9f8daffba9237ea  ./052311/update-CM7-dalingrin-OC-emmc-052311.zip
    a02ebb0350533af211089cde95fb5105  ./051311/update-Froyo1.2-dalingrin-2.6.32-emmc-051311.zip
    1177d17548c65f89962720ba3bbffaa4  ./051311/update-cm7-dalingrin-2.6.32-sdcard-051311.zip
    bfa163bc7ad0e1208997af1dbfb9e33b  ./051311/update-cm7-dalingrin-2.6.32-emmc-051311.zip
    053ad16383b7b2abc859f622927c965e  ./042411/update-CM7-dalingrin-OC-sd-042411.zip.md5
    9d0f3fe6d4781a0a2e7b494523c1db72  ./042411/update-CM7-dalingrin-OC-sd-042411.zip
    7652d13fc5f3d819b986fda7e6d3ee3f  ./042411/update-CM7-dalingrin-OC-emmc-042411.zip.md5
    1569ba396bdac3404f997d7e05faa247  ./042411/update-CM7-dalingrin-OC-emmc-042411.zip
    6a7519ba0ae3a7b4f34e38fdb7fa0d0c  ./042311/update-CM7-dalingrin-OC-sd-042311.zip
    f62579042d610f4fbf739d41f7aebc48  ./042311/update-CM7-dalingrin-OC-sd-042311B.zip
    a241e36f4c6c663190eb4a13b1cc73c4  ./042311/update-CM7-dalingrin-OC-sd-042311A.zip
    19bd148a4a180c1bf6b88559ffc0f9f5  ./042311/update-CM7-dalingrin-OC-emmc-042311.zip
    745960368e2ada7d48db13b37a997d4c  ./042311/update-CM7-dalingrin-OC-emmc-042311B.zip
    4fd0718b4418ccdf56f230df147e7b41  ./042311/update-CM7-dalingrin-OC-emmc-042311A.zip
    b16f61d937e39f6b1fe35463d576a671  ./041911/update-froyo-dalingrin-OC-sd-041911.zip
    3b44f1972c947da700a835d29c654f34  ./041911/update-froyo-dalingrin-OC-emmc-041911.zip
    a335f912d3cb3faffb4366e20fda7654  ./041811/update-CM7-dalingrin-OC-sd-041811C.zip
    41817da1d160733d5ce198cb0139db46  ./041811/update-CM7-dalingrin-OC-sd-041811B.zip
    7b91add1ce7bdd2a6722071aec8098eb  ./041811/update-CM7-dalingrin-OC-emmc-041811.zip
    e77dedc60d8f0af588820af533d6004c  ./041811/update-CM7-dalingrin-OC-emmc-041811B.zip
    0846c996bad671a07c095a783506a139  ./041311/update-CM7-dalingrin-OC-sd-041311.zip
    34a13469b79700c1ae7bcc0ab056abdc  ./041311/update-CM7-dalingrin-OC-emmc-041311.zip
    25610f0b29c944762b642aec495d4193  ./041211/update-froyo-dalingrin-OC-sd-041211.zip
    b3efe8ac1a514e43157eb32cbefce791  ./041211/update-froyo-dalingrin-OC-emmc-041211.zip
    bac72586c1fa1a4c374f3cc5ad0c1cbf  ./041211/update-CM7-dalingrin-OC-sd-041211.zip
    44e2984a61eca20cbde595da7e66f9f5  ./041211/update-CM7-dalingrin-OC-emmc-041211.zip
    b8e5425a7288a5b43e39c5f455eeb0d7  ./041111/update-CM7-dalingrin-OC-sd-041111.zip
    7181812d69ecdc80f32fc0b3f5c1181f  ./041111/update-CM7-dalingrin-OC-emmc-041111.zip
    def332f172d6db5822c436900d36b65a  ./040511/update-CM7-dalingrin-OC-sd-040511.zip
    385234a8e7f7da3212171d37eab1d95e  ./040511/update-CM7-dalingrin-OC-emmc-040511.zip
    bae26ee57c51ac48d687a7a3f4d1efb9  ./040411/update-froyo-dalingrin-OC-sd-040411.zip
    8f99790621b3bc0fef4a7d5f3319c009  ./040411/update-froyo-dalingrin-OC-emmc-040411.zip
    200be792e966453e21bbec0d6b10a288  ./040411/update-eclair-dalingrin-OC-sd-040411.zip
    872dbe19f539cec5e9e58e6b36aede35  ./040411/update-eclair-dalingrin-OC-emmc-040411.zip
    53e66c6328a72ed500c6d47bd9c48ea6  ./040411/update-CM7-dalingrin-OC-sd-040411.zip
    abe51daa708ecdafb74d286dd21ad2d8  ./040411/update-CM7-dalingrin-OC-emmc-040411.zip
    684c0a009c4503e7ad8111e4d2889d93  ./033111/update-froyo-dalingrin-OC-sd-033111.zip
    e75019d04668208cbe81e5aa8daf5ed5  ./033111/update-froyo-dalingrin-OC-emmc-033111.zip
    61c35771ff5ef267ca47fbae36d901df  ./033111/update-eclair-dalingrin-OC-sd-033111.zip
    a3da36043ea8efac3b35c5f82aa24d3d  ./033111/update-eclair-dalingrin-OC-emmc-033111.zip
    34bd1736b35f400a936b5adfdd5d061c  ./033111/update-CM7-dalingrin-OC-sd-033111.zip
    2bccf595bbec4010724ede6365a4644d  ./033111/update-CM7-dalingrin-OC-emmc-033111.zip
    fba87ec0d18411e72b5718fde1607e9e  ./032811/update-froyo-dalingrin-OC-sd-032811.zip
    6601902dbc03ea67ee97eda1a59c9fcd  ./032811/update-froyo-dalingrin-OC-emmc-032811.zip
    f133b2c452382f04f8b50dfb790527c7  ./032811/update-eclair-dalingrin-OC-sd-032811.zip
    692330cc8b13ce1b041e2a87c633168e  ./032811/update-eclair-dalingrin-OC-emmc-032811.zip
    6bacc09cf04e7edf78eefcc6fb728838  ./032811/update-CM7-dalingrin-OC-sd-032811.zip
    4c9102c3f2a6f253c81c777029ba7980  ./032811/update-CM7-dalingrin-OC-emmc-032811.zip
    13c03150b72938fa15f43bd795969dfe  ./031611/update-CM7-dalingrin-OC-sd-031611.zip
    b481b9f729e465e258e4fdb85dc1700d  ./031611/update-CM7-dalingrin-OC-emmc-031611.zip
    4ddd47b2633b86004f970faa53aba671  ./031311/update-CM7-dalingrin-OC-sd-031311.zip
    c534673e003946a6ed108f60bb075d55  ./031311/update-CM7-dalingrin-OC-emmc-031311.zip
    e54bdeefac03912e4948b753ce73276b  ./031111/update-CM7-dalingrin-OC-sd-031111.zip
    67857a797ecb8219ad4ff3b5e5581d88  ./031111/update-CM7-dalingrin-OC-emmc-031111.zip
    cbf70b6556ccc58690257026c224a56b  ./031011/update-froyo-dalingrin-OC-sd-031011.zip
    cfb083c3ec0816b893d43cade27e3b4e  ./031011/update-froyo-dalingrin-OC-emmc-031011.zip
    90aaab882957111328503573f9ffd16c  ./030811A/update-froyo-dalingrin-OC-sd-030811A.zip
    55c282c3928abadd0486bc04cbfd2e74  ./030811A/update-froyo-dalingrin-OC-emmc-030811A.zip
    71df4553e3986719750dde72d3da5e96  ./030811A/update-CM7-dalingrin-OC-sd-030811A.zip
    91637380e3ec853c8506e50e72be07ff  ./030811A/update-CM7-dalingrin-OC-emmc-030811A.zip
    25ba50b8865315b44fc8b98d382b5447  ./022511/update-HC-dalingrin-OC-sd-022511.zip
    f08c443ed730b147e57ac224fdd521eb  ./022511/update-HC-dalingrin-OC-emmc-022511.zip
    3243664a945b82fde9895bdfd59db701  ./022511/update-Froyo_CM7-dalingrin-OC-sd-022511.zip
    0bf22eface5c313c783b627d7a85d475  ./022511/update-Froyo_CM7-dalingrin-OC-emmc-022511.zip
    6efa7c62b5edf1d313996c13bed3d224  ./022511/update-eclair-dalingrin-OC-sd-022511.zip
    b2dfb032aae911e81708df8f39a22381  ./022511/update-eclair-dalingrin-OC-emmc-022511.zip

    Updates:
    [9/01/11] - Update to 1.3 source drop and other odds and ends
    [6/30/11] - Add voltage control, add InteractiveX governor, fix typos, change in audio wake lock
    [6/18/11] - Fix ondemand gov, USB host support(including xbox gamepad), bluetooth wake lock, improved wifi
    [6/13/11] - Remove BFS due to sleep issues. Backport 2.6.39 ondemand governor.
    [6/09/11] - Support for customizing the CPU overclock. Kernel ships defaulted to OC up-to 1.2ghz but can now be changed via an app. Minor change in SDIO(wifi) release behavior.
    [5/23/11] - BFS scheduler. BFQ I/O scheduler, SLQB allocator, Tiny RCU, removed smartass and ondemand, optimized KCFLAGS, fix video issues for people with modified u-boot, and more...
    [5/13/11] - 2.6.32 builds for CM7 beta and B&N 1.2 Froyo
    [4/04/11] - Minor update to all kernels: utf8 filesystem support and advanced routing for Cisco VPN connections.
    [3/31/11] - All kernels except Honeycomb: VPN fixed(hopefully).
    -DSP fix with overclock: http://forum.xda-developers.com/showpost.php?p=12564117&postcount=1017

    [3/28/11] - All kernels execpt Honeycomb: Cifs and tun modules included. Netfilter/IPtables tested insofar as Droidwall worked
    -CM7 and Froyo: Reverted the 2.6.29.6 update because of I/O performance problems found by Divine_Madcat

    [3/16/11] - CM7: 2.6.32 DSP driver backport for video playback support.
    [3/13/11] - CM7: Bluetooth/sysfs race condition fix. Fix spotted upstream by verygreen
    [3/11/11] - Bluetooth added to CM7 kernel
    [3/8/11] - Attempt #3 to get proper VPN support...TUN support in kernel and removed module.
    [3/8/11] - For CM7: internal storage mounting. For froyo and CM7: Interactive governor, various backports, attempt #2 for Netfilter w/NAT, tun/tap module.
    -Be sure to enable the interactive governor in Cyanogenmod settings->Performance->CPU for CM7
    -And use something like SetCPU for Froyo

    [2/25/11] - Performance tweaks. Netfilter, iptables, and tuntap support in kernel. Few fixes
    [2/23/11] - All kernels updated including eclair. Eclair now at feature parity. Kernel no longer overclocks on boot. Increased speaker and headphone volume. All kernels CWM flashable.
    [2/18/11] - Cache line check race condition. CM7 support.
    [2/6/11] - New Honeycomb(v03+) and Froyo kernels. Touchscreen workarounds from Deeper-blue. Various optimizations.
    [2/5/11] - New Honeycomb kernel to work with v03+
    [2/2/11] - Update with Honeycomb info
    [1/30/11] - Froyo 1ghz kernel posted
    [1/24/11] - Froyo test kernel
    -1.1ghz froyo test kernel. It has not been extensively tested. Please report results.
    [1/23/11] - 1.1ghz kernel added
    -This kernel is stable on my Nook after 10 Quadrant runs. Still boots everytime as well.
    -Renamed the 1ghz kernel to match new 1.1ghz kernel

    [1/23/11] - New 1ghz kernel added
    -This kernel is an experiment it uses more Omap3630 settings than the previous kernel which seems to make it stable for me.
    -I don't expect this kernel will be stable for everyone but for those that could get
    950mhz to work, you may be able to get this 1ghz kernel working.
    -Only short term stability testing so far. YMMV


    Install Instructions:
    All kernels are CWM flashable.
    If not using CWM, unzip the contents of the update and push uImage onto /boot partition. Then you will need to copy the system folder.
    Code:
    if flashing to sdcard:
         adb shell mount -t vfat /dev/block/mmcblk1p1 /emmc
         adb push uImage /emmc/uImage
         adb remount
         adb push system /system
    
    if flashing internal eMMC
         adb shell mount -t vfat /dev/block/mmcblk0p1 /emmc
         adb push uImage /emmc/uImage
         adb remount
         adb push system /system
    
    Reboot Nook!

    If you have problems with force closing apps or poor performance, clear the dalvik cache and regular cache.
    DO NOT use the Rom Manager option to wipe data and cache. You do not want to wipe data. Just boot into CWR
    and wipe cache, then under advanced wipe dalvik cache.
    Note that the first boot after clearing the caches will take longer than normal.

    Yes....This color is hideous.
    19
    well, its been about 3hrs since I log out of latitude and Im happy to report that my nook has been deep sleeping all this time. according to cpu spy, deep sleep 95% and the 5% was shared between 300 and 600 :D

    I think this is the trick to long deep sleep. Im going to turn on live wallpaper and see what happens. its going to be interesting to see how it goes over night. so far so good.

    Just to add: I've never had problems with deep sleep and I've also never signed into latitude.


    About PBD:
    Its on my to-do list but it isn't at the top yet =(
    It seems to have gotten better for most folks who have been rebooting regularly so I'm working on other things for now.

    Things on my to-do list:
    -Merge OMAP SoC optimizations into the CM7 framework(this helps us and lays the foundation for OMAP4 support in CM7)
    -Figure out why our Wifi driver sucks and is causing SODs for some people
    -Media optimizations, including Flash hw overlay support and perhaps 720p DSP decoding(media optimizations are dependent on the OMAP SoC optimizations first)
    -Merge OC kernel into CM7
    -Look into PBD again
    ...etc
    19
    I will no longer be releasing this kernel separately from CM7. It will be merged into CM7 as the default kernel. I will keep this thread alive for Stock 1.2 users.

    By default CM7 will use a 32 bit framebuffer as it always has. If you prefer 16 bit for performance, like what is in this OC kernel, then you will be able to switch it via Nook Color Tweaks.

    Once this update is merged it will contain a few other changes. The biggest change other than the kernel is the GPU drivers and libraries. These have been updated to a much newer and supposedly faster release. The wifi driver has also been updated.

    In a separate commit I will also be moving us back to lcd density 160. I've stuck with 161 for a long time now due to a bug in the Android Market. Fortunately, the new Market has fixed the issue and we can now move back to a 160. This will allow more apps to better scale to our display.
    16
    I honestly don't know what else can be done to improve Bluetooth. As of now we've backported 2.6.35+ bluetooth drivers and software to use on the Nook. Most of the important parts are controlled by a closed source proprietary firmware. Not much I do there.

    For those that use Bluetooth and are comfortable with Android and Linux, you could try different bluetooth firmware/script. This is a .bts file.
    15
    I honestly don't know what else can be done to improve Bluetooth. As of now we've backported 2.6.35+ bluetooth drivers and software to use on the Nook. Most of the important parts are controlled by a closed source proprietary firmware. Not much I do there.

    For those that use Bluetooth and are comfortable with Android and Linux, you could try different bluetooth firmware/script. This is a .bts file.

    I want to stress that last point.

    When the kernel went from 2.6.29 to 2.6.32, the bluetooth driver was completely re-implemented (that is, bluetooth was enabled, not once, but two totally different ways for you people..) using a backport of the 2.6.35 driver. So rather than using a simple bluetooth hci driver, CM7.1 uses a "shared transport" driver-- a newer, fresher way of doing things that requires a userspace daemon running ("uim") that interfaces with its counterpart in the kernel("kim") which manages traffic coming from 2ndary bluetooth, FM radio, and (if nook had it) GPS drivers-- because all three can be on one low-cost wireless chip. Which is pretty amazing actually. So since they're sharing the same chip, they need a "traffic cop" to communicate to the rest of the computer, and that's what the kim (kernel init manager) /uimd (user init manager daemon?) shared transport is about.

    Now in the nook's case, there is no antenna for the FM radio (until someone adds one), and no GPS hardware at all. The nook has a WL1271 in case anyone cares. But just as the wifi (which is also on that chip) has its own firmware that is installed at boot via "wlan_loader" (see /init.encore.rc), the bluetooth chip does too-- and basically it's initialized via a "bts" (bluetooth script I think) file provided by Texas Instruments that is run at boot by kim at uim's request, if I remember correctly.

    To quote myself:

    There are several versions of TI's bluetooth initialization script, called the "bts" script. You can get them here:

    https://gforge.ti.com/gf/project/wilink_drivers/

    The .bts files (such as TIInit_7.2.31.bts) may be placed in:

    /system/etc/firmware

    and will automatically be loaded when needed. Try different versions and experiment to see if some work better others. (Watch the log and dmesg to verify that the bts script is running properly)

    These .bts files are custom-written by TI to initialize the bluetooth hardware, and I think that would include setting up power levels and stuff that might be affecting signal strength and all the complaints everyone has. As there are several versions of this file (oddly many sharing the same name), it is still up to *you* to try them all and see which works best for the encore. Who knows, you may stumble onto one that extends the signal by a mile. You can even bat your eyes and petition TI to whip up one custom for the nook. So go on, be a hero.

    Anyway, that's where things stand with that. TMI? Well, you asked.

    One last thing, from Dalingrin's list:

    -Merge OMAP SoC optimizations into the CM7 framework(this helps us and lays the foundation for OMAP4 support in CM7)

    I really want to stress for everyone how much work this is, and that everyone owes Dal some beers or something. I know how much work this is because I went through it once several months back (way before .32 was ported, so the tree went stale and it all ended up being worthless), and it was a monster of a project. But it will be many hours well spent-- the Nook COLOR is based on the OMAP3 system (OMAP3621 if you care)-- but when Dal gets this done and accepted into CM7, it will make the CM7 framework OMAP4-friendly for new, as-yet-unknown devices in the years to come...

    So scroll back up, and thank him again.