Page 5 of 6 FirstFirst 123456 LastLast
Results 81 to 100 of 118
  1. #81  
    Quote Originally Posted by TomD View Post
    Tried "drop_caches" in Terminus with Dev Mode and got"

    "Permission denied"

    Any ideas?
    I used putty and ran this command and it did indeed free up large amounts f memory.

    Did you try with this method?

    Fresh boot using "shutdown -r now"
    root@palm-webos-device:/# free
    free
    total used free shared buffers cached
    Mem: 504680 271640 233040 0 25032 93960
    -/+ buffers/cache: 152648 352032
    Swap: 131064 0 131064

    1 app openbrowser home screen)
    root@palm-webos-device:/#
    root@palm-webos-device:/# free
    free
    total used free shared buffers cached
    Mem: 504680 277764 226916 0 25584 95020
    -/+ buffers/cache: 157160 347520
    Swap: 131064 0 131064

    10 apps open:

    root@palm-webos-device:/# free
    free
    total used free shared buffers cached
    Mem: 504680 356980 147700 0 26152 107452
    -/+ buffers/cache: 223376 281304
    Swap: 131064 0 131064

    10 apps closed:
    root@palm-webos-device:/# free
    free
    total used free shared buffers cached
    Mem: 504680 356980 147700 0 26152 107452
    -/+ buffers/cache: 223376 281304
    Swap: 131064 0 131064

    As we can see some items were left in cache...

    sync; echo 3 > /proc/sys/vm/drop_caches

    root@palm-webos-device:/# free
    free
    total used free shared buffers cached
    Mem: 504680 190680 314000 0 176 28844
    -/+ buffers/cache: 161660 343020
    Swap: 131064 0 131064
    root@palm-webos-device:/#

    Really does work....
    Last edited by corbintechboy; 06/23/2010 at 01:41 PM.
  2. #82  
    It seems to me based on the post above that more than just the 10 apps that were opened had stuff cached. So perhaps it may not be a good idea to clear it in that way unless it doesn't effect the core apps that run in the background such as phone etc. I remember back in the day having to wait like 10 maybe 15 seconds for the phone app to launch when answering a call now it's 1 or 2 seconds if not instant most of the time. Is there a way of listing everything that is in the cache before and after the above post process is performed?
    As requested: for my works on webOS patches and apps. Twitter: @larryboytw Patches: Small icons browser start page, 5x5 launcher. I have an AAS CIS Programming degree. I enjoy working on open source projects and alpha and beta testing.
    http://install.preware.org/ for easy to get up and running for patches and apps.
  3. #83  
    Quote Originally Posted by StoneRyno View Post
    It seems to me based on the post above that more than just the 10 apps that were opened had stuff cached. So perhaps it may not be a good idea to clear it in that way unless it doesn't effect the core apps that run in the background such as phone etc. I remember back in the day having to wait like 10 maybe 15 seconds for the phone app to launch when answering a call now it's 1 or 2 seconds if not instant most of the time. Is there a way of listing everything that is in the cache before and after the above post process is performed?
    I did a test based on your question and it seems the same as far as phone app reply time before or after the command is run.
  4.    #84  
    Quote Originally Posted by corbintechboy View Post
    I used putty and ran this command and it did indeed free up large amounts of memory.
    Actually, it did not. It cleared the buffers and caches, something that would happen automatically the first time a process asked for more memory and there was no free memory outside of the buffers and caches.

    The Linux cache is not where apps are loaded directly (as I have said many times before, apps are not separate processes in webOS), and the storage listed in the buffers and cached lines is always available to malloc() calls in Linux applications (like the malloc() call that would be used by the LunaSysMgr or BrowserServer processes to load new apps).

    So running a command which clears these caches has absolutely no effect on the amount of memory available for loading apps.

    -- Rod
    WebOS Internals and Preware Founder and Developer
    You may wish to donate by Paypal to donations @ webos-internals.org if you find our work useful.
    All donations go back into development.
    www.webos-internals.org twitter.com/webosinternals facebook.com/webosinternals
  5. #85  
    Quote Originally Posted by rwhitby View Post
    Actually, it did not. It cleared the buffers and caches, something that would happen automatically the first time a process asked for more memory and there was no free memory outside of the buffers and caches.

    The Linux cache is not where apps are loaded directly (as I have said many times before, apps are not separate processes in webOS), and the storage listed in the buffers and cached lines is always available to malloc() calls in Linux applications (like the malloc() call that would be used by the LunaSysMgr or BrowserServer processes to load new apps).

    So running a command which clears these caches has absolutely no effect on the amount of memory available for loading apps.

    -- Rod
    Sorry but you are incorrect.

    If we look a bit closer we will see with using the -m option:

    Nothing in memory fresh boot with "shutdown -r now"

    root@palm-webos-device:/# free -m
    free -m
    total used free shared buffers cached
    Mem: 492 268 224 0 23 89
    -/+ buffers/cache: 155 337
    Swap: 127 0 127

    Now 10 apps:

    root@palm-webos-device:/# free -m
    free -m
    total used free shared buffers cached
    Mem: 492 342 150 0 25 105
    -/+ buffers/cache: 211 281
    Swap: 127 0 127

    Apps simply closed:

    root@palm-webos-device:/# free -m
    free -m
    total used free shared buffers cached
    Mem: 492 306 185 0 25 107
    -/+ buffers/cache: 173 319
    Swap: 127 0 127

    After clearing cache:

    root@palm-webos-device:/# free -m
    free -m
    total used free shared buffers cached
    Mem: 492 186 306 0 0 28
    -/+ buffers/cache: 158 334
    Swap: 127 0 127


    Cache is stored in local memory. So by removing cache we free cached items in physical memory.
  6.    #86  
    Quote Originally Posted by corbintechboy View Post
    Sorry but you are incorrect.

    If we look a bit closer we will see with using the -m option:

    Nothing in memory fresh boot with "shutdown -r now"

    root@palm-webos-device:/# free -m
    free -m
    total used free shared buffers cached
    Mem: 492 268 224 0 23 89
    -/+ buffers/cache: 155 337
    Swap:

    Now 10 apps:

    root@palm-webos-device:/# free -m
    free -m
    total used free shared buffers cached
    Mem: 492 342 150 0 25 105
    -/+ buffers/cache: 211 281
    Swap:

    Apps simply closed:

    root@palm-webos-device:/# free -m
    free -m
    total used free shared buffers cached
    Mem: 492 306 185 0 25 107
    -/+ buffers/cache: 173 319
    Swap:

    After clearing cache:

    root@palm-webos-device:/# free -m
    free -m
    total used free shared buffers cached
    Mem: 492 186 306 0 0 28
    -/+ buffers/cache: 158 334
    Swap:


    Cache is stored in local memory. So by removing cache we free cached items in physical memory.
    Sigh. We will need to agree to disagree on this one it seems, as you are simply not reading the words I am writing here.

    If you wish to continue to try and convince me you are correct, please post the kernel code which the lowmemnotify kernel module uses to calculate the real free memory (hint: it's the second line printed by the free command, the values in which do not change appreciably) and show how the memory that just happens to be marked as caches or buffers enters into that equation.

    -- Rod
  7. #87  
    Quote Originally Posted by rwhitby View Post
    Sigh. We will need to agree to disagree on this one it seems, as you are simply not reading the words I am writing here.

    If you wish to continue to try and convince me you are correct, please post the kernel code which the lowmemnotify kernel module uses to calculate the real free memory (hint: it's the second line printed by the free command, the values in which do not change appreciably) and show how the memory that just happens to be marked as caches or buffers enters into that equation.

    -- Rod
    Fair enough.

    I could be convinced I am wrong by someone taking a Sprint Pre and getting the TMC error and executing the above command and trying again without a reboot.

    If this does not work, I will apologize to all and back away with the understanding that I was wrong.

    I really think this could possibly fix the issue.
  8. #88  
    Quote Originally Posted by corbintechboy View Post
    Fair enough.

    I could be convinced I am wrong by someone taking a Sprint Pre and getting the TMC error and executing the above command and trying again without a reboot.

    If this does not work, I will apologize to all and back away with the understanding that I was wrong.

    I really think this could possibly fix the issue.
    I can confirm as a 3rd party... that I have gotten TMC, NOT executed that code. And was able to try again without reboot using the wonderful code in the Alpha UberKernel w/ compcache. So whatever Rod and crew are doing with adjustments to the lowmemnotify are working......

    Not saying I have close to the knowledge of either of you.... but the Kernel mods being done, are greatly fixing and adjusting the way webos reports TMC.
  9. #89  
    It would be nice if corbintechboy's attempt led to something.

    What problem I experience is not with TMC errors coming up after I've opened 3-5 apps, or browsers, or 3D games, but that after a few hours use on the stock Pre, closing all Apps leads to no reclamation of memory, disallowing 3D games entirely, or sometimes even the browser.

    From my expectation, as soon as the app is closed it should release all memory, or be reclaimed by the garbage collector soon. However it seems that once the TMC has begun blocking games it never reclaims memory until a restart. I've gone at least a day with this being the case.

    I, conveniently, have no proof data-wise of this occurring. But like OldSkool, the Kernels seem to help.
  10. #90  
    Quote Originally Posted by alex.dobeck View Post
    It would be nice if corbintechboy's attempt led to something.

    What problem I experience is not with TMC errors coming up after I've opened 3-5 apps, or browsers, or 3D games, but that after a few hours use on the stock Pre, closing all Apps leads to no reclamation of memory, disallowing 3D games entirely, or sometimes even the browser.

    From my expectation, as soon as the app is closed it should release all memory, or be reclaimed by the garbage collector soon. However it seems that once the TMC has begun blocking games it never reclaims memory until a restart. I've gone at least a day with this being the case.

    I, conveniently, have no proof data-wise of this occurring. But like OldSkool, the Kernels seem to help.
    Your problem sounds like a memory leak.

    Running the command above would be a band aid fix at best. I cannot produce any error on my Pre (Verizon) by opening many apps to check out my theory.

    Really, this does not even have to be a memory leak at all. This could still be a cache memory issue that just a result of full memory and no room to launch anything. May not be a band aid fix after all.

    If you can, get your device in that state and hook the Pre to your computer and try the command above (sync; echo 3 > /proc/sys/vm/drop_caches). I would be very interested in seeing some results from that command. Even if this turned out to be something that needed ran daily, an app could be made without much issue I am sure.
  11. #91  
    Doing some tinkering here and I am trying to get a TMC error.

    At 98 "things" open (browsers and whatnots) the phone simply reboots. 98 does indeed seem to be the magic number here.

    No way to test my theory.
  12.    #92  
    Quote Originally Posted by corbintechboy View Post
    Doing some tinkering here and I am trying to get a TMC error.

    At 98 "things" open (browsers and whatnots) the phone simply reboots. 98 does indeed seem to be the magic number here.

    No way to test my theory.
    What kernel are you running?

    -- Rod
    WebOS Internals and Preware Founder and Developer
    You may wish to donate by Paypal to donations @ webos-internals.org if you find our work useful.
    All donations go back into development.
    www.webos-internals.org twitter.com/webosinternals facebook.com/webosinternals
  13. #93  
    I have observed the following so far in 2 or 3 tests with 3D games. I have not tested with other stuff yet. Also note the tests were performed after compcache and the change mentioned previously were added to the uber kernel testing version. And repeated with the update(s) since then.

    My device consistently boots up with the same amount of memory/swap used give or take 1 or 2 MB. This test gets done the most since I daily reboot along with irregular reboots when required for an update or turning on/off dev mode as needed for development purposes.

    Fresh boot no open apps, load any 3D game, then open govnah, record the memory/swap, close govnah, compare to record from boot, play the game for say an hour give or take, open govnah, record the memory/swap, close govnah, compare to the previous amounts, close 3D app, open govnah, record memory/swap, close govnah. So far memory has always come up with less used than after boot but more swap than after boot. The amount varies each time but only slightly like the slight variation after boot. I assume the reason this is observed is because the memory requirement of the 3D game cleanup is performed to clear non-essential stuff present in memory from boot. To which I have no idea if later on that stuff is loaded back into memory automatically or not. I will probably do this same process but instead load up 10 or 15 apps (not any 3D games) and see what occurs.

    I'm not sure I understand much of the details for compcache so I have no idea if when it is on if I am crossing the limit for it to be used or how that even occurs. But even before turning compcache on I tested to see if the few things I know caused TMC would cause TMC and because of the fix previously discussed I no longer got TMC in those situations. In fact since then and with compcache on (set to 32MB) I have only once gotten TMC and that was from playing assassin's creed as previously described. I have been testing the 3D games I have to see if others result the same. After playing each of the 4 3D games I own for over 2 hours each and rebooting between each the only one I observe a significant increase in memory usage is assassin's creed. Outside of that I have had monopoly lock up on me (the app not device) a lot and I have to swipe it closed and open it again and hope that the auto save wasn't but a turn behind (which it seems it auto saves just about every turn). Though I have no idea why the lock up happens and it isn't consistent either since my first full game from beginning to end no lock ups but after that lock ups occurred in varied frequency from say 10 turns after the app loaded to several times that.
    As requested: for my works on webOS patches and apps. Twitter: @larryboytw Patches: Small icons browser start page, 5x5 launcher. I have an AAS CIS Programming degree. I enjoy working on open source projects and alpha and beta testing.
    http://install.preware.org/ for easy to get up and running for patches and apps.
  14. #94  
    Quote Originally Posted by rwhitby View Post
    What kernel are you running?

    -- Rod
    .... Only Kernels that should cause reboot would be F102 and F104 or some other custom kernel, right? Any other kernel "SHOULD" not allow reboot at least not at the cause of lack of memory. But those kernels (F102 and F104) eliminate TMC soooooo the test is completely inconclusive?

    Just inferring from what I have managed to learn.
  15.    #95  
    Quote Originally Posted by OldSkoolVWLover View Post
    .... Only Kernels that should cause reboot would be F102 and F104 or some other custom kernel, right? Any other kernel "SHOULD" not allow reboot at least not at the cause of lack of memory. But those kernels (F102 and F104) eliminate TMC soooooo the test is completely inconclusive?

    Just inferring from what I have managed to learn.
    Yep. You are correct. That's why I'm asking the question.

    Too many opinions are posted here without performing controlled experiments and then reporting the results of those experiments including all relevant information ... this is a "Technical Analysis" thread that has strayed quite considerably from technical analysis reporting.

    -- Rod
    WebOS Internals and Preware Founder and Developer
    You may wish to donate by Paypal to donations @ webos-internals.org if you find our work useful.
    All donations go back into development.
    www.webos-internals.org twitter.com/webosinternals facebook.com/webosinternals
  16.    #96  
    Quote Originally Posted by StoneRyno View Post
    Fresh boot no open apps, load any 3D game, then open govnah, record the memory/swap, close govnah, compare to record from boot, play the game for say an hour give or take, open govnah, record the memory/swap, close govnah, compare to the previous amounts, close 3D app, open govnah, record memory/swap, close govnah. So far memory has always come up with less used than after boot but more swap than after boot. The amount varies each time but only slightly like the slight variation after boot.
    Thanks for the detailed report.

    One thing to note is that it is to be expected that if you go into swap, then the information that is in swap will not automatically move back to RAM, so you will expect to see the swap go up and not come back down again until something happens to clear the swap space. But note also if you open lots of stuff, and then close it, and have no memory leaks, then the total of ram used plus swap used should be the roughly the same before and after irrespective of the specific values of ram used and swap used (which will differ depending on which information stayed in ram and which was moved to swap).

    Can you report on total memory usage numbers (ram plus swap) for the same scenarios?

    -- Rod
    Last edited by rwhitby; 06/23/2010 at 09:00 PM.
    WebOS Internals and Preware Founder and Developer
    You may wish to donate by Paypal to donations @ webos-internals.org if you find our work useful.
    All donations go back into development.
    www.webos-internals.org twitter.com/webosinternals facebook.com/webosinternals
  17. #97  
    Quote Originally Posted by rwhitby View Post
    What kernel are you running?

    -- Rod
    Well, I dropped my first Pre and broke the screen and had to get a replacement. On this one I am bone stock (other then root access).
  18. #98  
    Quote Originally Posted by rwhitby View Post
    Yep. You are correct. That's why I'm asking the question.

    Too many opinions are posted here without performing controlled experiments and then reporting the results of those experiments including all relevant information ... this is a "Technical Analysis" thread that has strayed quite considerably from technical analysis reporting.

    -- Rod
    thanks, figured I was on the right line there. Like I said earlier not a linux expert, but I am pretty good as a general IT trouble shooter in general. Thanks for confirming, makes me feel a lil better about being an alpha tester on the new kernel and govnah. If I understand the concept of what I am testing I feel I can better diagnose my own end user experience.

    sorry to contribute to a lil of the off topic, but it's good to know my basic understanding is in line with what should happen.
  19. #99  
    Quote Originally Posted by corbintechboy View Post
    Well, I dropped my first Pre and broke the screen and had to get a replacement. On this one I am bone stock (other then root access).
    so you are "testing" on a new device.... 90+ cards it has to be a plus.... How long have you had the phone? A stock kernel shouldn't be rebooting before TMC.

    once again the above and below thoughts do not come from my own direct knowledge, but what I have learned from reading and asking questions.

    once again my limited knowledge would tell me that the reboot is not memory related OR your command line circumvents the TMC error (similar how F102 and F104 do)... Maybe not in the same way but still removing or bypassing the factory safety net of TMC.
  20. #100  
    Quote Originally Posted by OldSkoolVWLover View Post
    so you are "testing" on a new device.... 90+ cards it has to be a plus.... How long have you had the phone? A stock kernel shouldn't be rebooting before TMC.

    once again the above and below thoughts do not come from my own direct knowledge, but what I have learned from reading and asking questions.

    once again my limited knowledge would tell me that the reboot is not memory related OR your command line circumvents the TMC error (similar how F102 and F104 do)... Maybe not in the same way but still removing or bypassing the factory safety net of TMC.
    This phone is an insurance replacement. The phone arrived about a week ago brand new in box.

    The reboot issue happens whether after a fresh restart or not. I may try another Kernel don't really know yet. Some say the phone is slow and not to par without a better Kernel. I seem to have a feeling in the other direction I feel the phone is fast and works great.

    There is no real world situation where I would have 90+ apps open, so that issue is a non issue.

    Yes, it is indeed the Pre+.
Page 5 of 6 FirstFirst 123456 LastLast

Posting Permissions