.24 is rock solid for me, but i have noticed the wifi disconneting it self at times
hopefully asus will address that soon

Originally Posted by knoxploration

much of Tien Phan's time seems to be taken up with monitoring and responding to complaints for other Asus products on sites like NewEgg, Amazon and Best Buy, while numerous of us here have received no response after reaching out about our Transformer problems.

Yep, that is True, surely Asus realise how many of the XDA people are top influencers of other peoples future buying decisions.

Tien posted his direct email address in one post ... cl-tien@asus.com

Has everybody with issues made their retailer aware of the problem directly, you must do this, make an issue of this with thr retailers and Asus will listen, if there is anought noise a retailer will think twice about stocking the products.

Asus UK have been useless in all my contact with them. The last one suggests an RMA, which as we know is not going to help.

Originally Posted by jtrosky

Are you stock or do you use a custom kernel? It's amazing how some can get such high uptimes - I can't even get 72 hours, due to one reason or another!! :(

Sent from my Transformer TF101 using Tapatalk 2

Stock and not (and never was) rooted. It is as out-of-the-box

I wonder if it has something to do with the preparation that I did for the update (cold boot, wipe, wipe, cold bood, update, wipe, cold boot) (not sure if this is relevant but... that's what I did I'm one of the few with no problems)

.24 is stable for my use

---

I have not had a SOD nor RR since .24.

Originally Posted by oldcpu

I have not had a SOD nor RR since .24.

What kind of uptime are you at?

Sent from my Transformer TF101 using Tapatalk 2

Smoking gun logcat

---

The saga continues. This morning I turned on the tablet - I normally leave it charging overnight so its fully charged by morning. I checked email etc then loaded up the latest Google+. Opened and closed it a couple of times and suddenly the screen goes blank and only desktop is visible with no icons. Fortunately I had smart taskbar running (great tool) and was able to navigate to the list of all applications and run logcat. I captured the log of the system in that state.

The log clearly shows that the root cause was the openGL renderer running out of memory.
I have hundreds of these errors in my log file.
While in that state I was randomly clicking around the screen, and somehow when I hit the top left corner, where I have the google search icon positioned the desktop sprang back to life. I loaded up logcat again and saw an interesting line.

" 05-28 07:53:54.327 492 492 W HardwareRenderer: Mountain View, we've had a problem here. Switching back to software rendering.
"

Not sure who's message that is but it looks like something triggered failsafe software rendering.
Excerpt from the log is below. I have also filed a report via the Google+ app so that Google also gets this info. Still no reply from ASUS. We have completely proven documented reports of the problem yet no response. Pathetic.


====================================

05-28 07:53:53.967 492 492 D OpenGLRenderer: GL error from OpenGLRenderer: 0x505

05-28 07:53:53.967 492 492 E OpenGLRenderer: OpenGLRenderer is out of memory!

05-28 07:53:53.967 492 492 D OpenGLRenderer: GL error from OpenGLRenderer: 0x505

05-28 07:53:53.967 492 492 E OpenGLRenderer: OpenGLRenderer is out of memory!

05-28 07:53:53.967 492 492 D OpenGLRenderer: Flushing caches (mode 1)

05-28 07:53:54.057 17706 17706 D TextView: Constructor - Got Res id for appearance for textColorPrimaryInverse

05-28 07:53:54.057 17706 17706 W ResourceType: Skipping entry 0x106006a in package table 0 because it is not complex!

05-28 07:53:54.057 17706 17706 D TextView: Constructor - Got appearance for textColorPrimaryInverse

05-28 07:53:54.057 17706 17706 D TextView: Constructor -- Got mEditTextBackgroundColor

05-28 07:53:54.117 17706 17708 D dalvikvm: GC_CONCURRENT freed 230K, 5% free 6743K/7047K, paused 2ms+2ms

05-28 07:53:54.297 492 492 E libEGL : eglMakeCurrent:684 error 3009 (EGL_BAD_MATCH)

05-28 07:53:54.297 492 492 E HardwareRenderer: eglMakeCurrent failed EGL_BAD_MATCH

05-28 07:53:54.297 492 492 D NvOsDebugPrintf: NvRmChannelSubmit failed (err = 196623, SyncPointValue = 17686473)

05-28 07:53:54.317 17706 17706 D libEGL : loaded /system/lib/egl/libEGL_tegra.so

05-28 07:53:54.327 492 492 W HardwareRenderer: Mountain View, we've had a problem here. Switching back to software rendering.

05-28 07:53:54.407 17706 17706 D libEGL : loaded /system/lib/egl/libGLESv1_CM_tegra.so

05-28 07:53:54.467 17706 17706 D libEGL : loaded /system/lib/egl/libGLESv2_tegra.so

05-28 07:53:54.517 17706 17706 D OpenGLRenderer: Enabling debug mode 0

---------- Post added at 09:04 AM ---------- Previous post was at 08:37 AM ----------

While searching for a way to disable OpenGL rendering I came across this:
http://www.anandtech.com/show/5310/s...dwich-review/2

This confirms why the instability is triggered by only certain apps. Both GMail and Google+ apps use OpenGL and this is triggering this bug. So far I don't see any way to disable OpenGL system wide. There is a way to force enable OpenGL in settings but not to disable it :-/

I did find a way to disable it in the stock browser:

to turn off OpenGL rendering:
1. Open your browser
2. Enter "about:debug" into the URL bar
3. The browser is now in debug mode, press menu button and go to "Settings"
4 go to "Debug" (visible when in debug mode)
5. disable OpenGL rendering

I will try running that and see if its better. If anyone knows how to disable OpenGL systemwide that might be a good workaround for now.

Originally Posted by devjazz

The log clearly shows that the root cause was the openGL renderer running out of memory.
I have hundreds of these errors in my log file.

Could you please check your logs for all messages related to pid/tid 492? Being a low-numbered pid, it's either a system service (SurfaceFlinger, perhaps?) or a widget loaded early in the runtime startup procedure.

---EDIT---
Found it, it's in frameworks/base/core/java/android/view/HardwareRenderer.java (class HardwareRenderer.GlRenderer):

        private void fallback(boolean fallback) {
            destroy(true);
            if (fallback) {
                // we'll try again if it was context lost
                setRequested(false);
                Log.w(LOG_TAG, "Mountain View, we've had a problem here. " 
                        + "Switching back to software rendering.");
            }
        }

This rules out SurfaceFlinger.

288 hrs (12 days) up time....no changes in performance or behavior.

Originally Posted by jtrosky

What kind of uptime are you at?

I typically prefer to switch off my Transformer 3 times per day (while transiting to/from work and when going to bed at night). I concede I have been negligent in doing this switch off religiously, and unintentionally let the transformer run for two lengthy periods of time, once for greater than 100 hours and more recently greater than 150 hours. Most other times I shut down after 8 to 10 hours of operation. Every shutdown since .24 update has been initiated by me.

Interesting. Under what condition is the fallback code being invoked? It seems like my system was stuck in that state until I clicked search by accident which is when it woke up. BTW shortly after my message the entire system froze up and I had to cold boot it so that seemed to be a temporary lease on life.

It seems having opengl memory allocated on a system wide basis is a recipe for disaster. Doesn't android have some fuse to prevent an app from leaking excess system wide resources like other OS's do?

It would be good if there was a way to disable OpenGL at least on a per app basis. If so I could turn it off for these apps and it would bypass this issue. From my reading it seems that an app requests support for OpenGL in the manifest. ICS provides "force GPU rendering" in developer options but nothing to disable GPU rendering. Is there a tool or way to disable this flag in an app's manifest?

Originally Posted by sbiriguda

Could you please check your logs for all messages related to pid/tid 492? Being a low-numbered pid, it's either a system service (SurfaceFlinger, perhaps?) or a widget loaded early in the runtime startup procedure.

---EDIT---
Found it, it's in frameworks/base/core/java/android/view/HardwareRenderer.java (class HardwareRenderer.GlRenderer):

        private void fallback(boolean fallback) {
            destroy(true);
            if (fallback) {
                // we'll try again if it was context lost
                setRequested(false);
                Log.w(LOG_TAG, "Mountain View, we've had a problem here. " 
                        + "Switching back to software rendering.");
            }
        }

This rules out SurfaceFlinger.