Originally Posted by e334

http://irclog.netripper.com/htc-linux/2011/11/23/

Wistilt2 mentions it also.. I'm inclined to say it isn't beneficial though...

They're talking about that 30fps limit which was broken on the OG EVO. They were thinking we had a similar artificial limit...

1. Thank You, Thank You, Thank You to all those continuing to develop these for the Touch Pro 2!!!

2. any idea when 3.4 will be included in the autobuilds?

3. Is there a link to what's new in 3.4 for us less informed techno geeks?

Originally Posted by thegof

1. Thank You, Thank You, Thank You to all those continuing to develop these for the Touch Pro 2!!!

2. any idea when 3.4 will be included in the autobuilds?

3. Is there a link to what's new in 3.4 for us less informed techno geeks?

1. You're welcome.

2. Nope. Probably when it's ready.

3. If you read the git commits, you should be able to get the jist. I don't think there's a whole lot of "new stuff" as far as RHOD-specific code... Just new kernel code that they're trying to get working on the RHOD.

htc-msm-3.4

---

I emailed WisTilt2 this morning, and hopefully 3.4 will get added to the autobuild soon.

It's in pretty good shape now, but initially we had a few serious issues with this kernel.

- When it came out, there was a serious problem with the memory manager in the upstream 3.4 kernel that flat out broke the chroot system call (not just for us, but other users as well). If you have ever taken a peak at our initramfs, we use chroot (or rather switch_root) to maneuver onto our rootfs. We were stuck spinning our wheels until around 3.4.2 made an appearance on kernel.org and this problem was patched.

- We had another problem with the patches that instruct the vanilla linux kernel to use the android wake_locks infrastructure. Not sure who's to blame here, but I am inclined to think it's kernel.org again. At any rate, for the first time ever I saw frequent SODs on the 3.x kernels. I think we fixed this also.

- Finally, USB: I think we are victims of our own success here. To get adb, usb tethering etc, to work with android, we use android-specific patches on the vanilla linux kernel. As we progress to newer and newer kernels, so do these android patches. The problem is that ever since 3.0, google has stopped supporting gingerbread-like usb configurations in their kernel patches. So we've been dragging the usb layer from .39 all the way up to 3.3 -> this way we were successful in using the new kernels with gingerbread. This however is a HUGE hack and not good in any way shape or form. Finally something in the new 3.4 kernel prevents us to use the drivers from .39 and we were forced upgrade our usb layer, and together with it bring in ICS geared USB patches.

The result: with some work, we can get adb to work. However gingerbread is looking for active USB connections in the wrong place and even if you get ADB going, GB is unaware of an active usb connection so you can't enable other perks like USB tethering.

Not sure what the solution is to this last problem -> probably need to hack at frameworks in GBX but the motivation for this is somewhat low, since at that point we would need to be distributing a separate build to use specifically with 3.4.

Therefore finally, the current bug-list for 3.4:
1. Kexec is broken
2. USB: We can get ADB (instructions on this to follow), however USB notification doesn't work, and consequently can't enable USB tethering. <--- see here how to take advantage of usb functionality on htc-msm-3.4
3. Bluetooth -> haven't looked at the BT patches yet, but hey BT never really works OK anyways.

On the bright side -> you will be using a kernel that's the bleeding edge in the linux world. Also, this might be placebo, but I think the new usb drivers manage the serial clocks better and perhaps power-management seems slightly better (interested in hearing how other folks think it compares).

edit: I thought my model number was in my sig- 210

Here is my experience with 3.3.6. Since 3.4 seems to be the order of the day, I'm not initially including a log file. Please tell me if one is still desired, and I will make it asap.

Thanks again for the work.

What I'm seeing so far:

Seems fairly stable. 3.3.6 has a lot more free ram at any given time compared to .27 (20-30mb free compared to 6-12). But i also notice that ram gets "cleaned out" more frequently. If I am using the news and weather widget and follow a story into a browser then close the browser the widget needs to reload. This didn't happen under .27. Not bad, per se. I am assuming this memory management is responsable for the phone feeling snappier in general. Edit: Multi-tasking is impossible with this kernal. The phone is quick, but it is becuase it never has more then one app open and usable at any given time. If I want to swtich between the browser and gmail, one of them has to close for the other to open.

Screen backight didn't work at first-it was stuck at full. Had to boot back to WinMo and disable "auto backlight". Next boot into Android and the screen backlight settings now work. Although, the minimum setting (backlight icon greyed out on the power widget) which used to give a usable amount of backlight indoors, now turns the backlight off-leaving the screen on but dark. And the minimum setting with the icon turned on is too bright, using way too much battery power. The auto setting never gets it right-didn't in really .27 either-always too bright(but did adjust). I don't think the auto setting is doing anything, actually. Edit: If I let the device time out on the back light and immediately press the power button to turn it back on, the sensor will detect the ambient light and adjust the brightness. Though, curiously, outside in the sun the back light is dim. Inside in dim lighting the back light is bright...this is backwards behavior, I would think.

Wifi connects fine to my WPA2 network. Though it seems to use much more power then under .27. In .27 leaving wifi turned on at home used no more batery then leaving it off. Now under 3.3.6, leaving wifi on at home causes my battery to drain about 40%+ faster.

BT. Hit or miss. Jabra Cruiser. When I can get the phone to connect, it works as it should-audio and calls. But I can't always get the phone to pair. Then I can't always get it to connect. And once the connection is lost (ie, walking out of range), I need to start the whole trial an error process all over again to get it to connect. So it works well when it works. But it is a pain in the arse.

The steps to connect with the most success are:
Turn on Jabra BT, start pairing mode
Turn on BT on phone
Imediately make it discoverable
Imedately search for devices
Then quickly press Jabra to connect
I added immediately/quickly to the steps above becuase any delay and the two won't pair or won't connect. I need to go through the above *everytime* I want to use BT.

Wifi tehtering: Haven't tried it yet.

Camera: Still images are fine. I might even say that low light performace is a bit better. Occasionaly the phone will freeze when taking photos, though.

"Reboot" option, first time I tried to reboot the phone it hung on the process and wouldn't do a darn thing until I pushed the reset button.

Charging and battery. Under .27 getting the phone to charge was a futile battle. I had to plug/unplug/turn on/turn off for serveral cycles before the phone would start charging. For the most part, using 3.3.6, the phone starts to charge with out hassle in about 10-30 seconds after plugging it in. Battery life seems fine, equal to .27- as long as I don't use wifi.

GPS. It works. But even using the Leo injector app, sometimes a lock can take as long as 10 minutes! And it has never locked in under a minute.

As I said, if 3.3.6 is still on the table, and you want my logs, please don't hesitate to ask.

Thanks!!!

Autobuild seems to be offline

Originally Posted by nickleby

Autobuild seems to be offline

Yea, jonpry said something similar earlier on IRC...

htc-msm-3.4 update: We managed to fix USB Tethering, and sounds like WisTilt2 will do his best to get the kernel on the autobuild shortly.

Please follow these instructions to take advantage of the usb functions on this kernel.

As always, when submitting reports please include complete GetLogs output.

Originally Posted by arrrghhh

Yea, jonpry said something similar earlier on IRC...

Autobuild is back baby

Are the kernels newer than 3.3.6 optimized for Nand GBX0*?