Page 3 of 4 FirstFirst 1234 LastLast
Results 41 to 60 of 69
Like Tree16Likes
  1. #41  
    I forgot to say, you need to go to

    etc/init.d/swaphack.sh

    and comment out the line:

    "echo 6000 > /proc/sys/vm/dirty_expire_centrics"

    make it say:

    Code:
    #echo 6000 > /proc/sys/vm/dirty_expire_centrics
    Or else it will override the setting in sysctl.conf

    again, this will all be in the patch
  2.    #42  
    Quote Originally Posted by rmausser View Post
    I forgot to say, you need to go to

    etc/init.d/swaphack.sh

    and comment out the line:

    "echo 6000 > /proc/sys/vm/dirty_expire_centrics"

    make it say:

    Code:
    #echo 6000 > /proc/sys/vm/dirty_expire_centrics
    Or else it will override the setting in sysctl.conf

    again, this will all be in the patch
    I wonder if this was why my phone spontaneously rebooted when my sister tried calling me this morning...
    Palm IIIc -> Sony CLIÉ T650C -> Sony TJ-37 -> Palm TX -> Palm Centro -> Palm Pre Bell -> Palm Pre Plus Bell/Verizon Hybrid -> HP Veer -> HP Pre 3 NA -> BlackBerry Classic -> BlackBerry Priv

    It's a Late Goodbye, such a Late Goodbye.

    Need OEM Palm Pre parts? See here
  3. AMR-1's Avatar
    Posts
    258 Posts
    Global Posts
    261 Global Posts
    #43  
    Gosh I wish we had access to the Devs in the Palm Dev forums that designed this stuff. Would be neato to see what they have tried and experimented with to fix these issues. We already know they are listening/watching since they have previously incorporated some of the compcache patch code into the stock kernels.
  4. #44  
    Quote Originally Posted by ToniCipriani View Post
    I wonder if this was why my phone spontaneously rebooted when my sister tried calling me this morning...
    Quote Originally Posted by ToniCipriani View Post
    I wonder if this was why my phone spontaneously rebooted when my sister tried calling me this morning...
    No, I think the reboot is from something else, im getting them occasionally.

    I think its having vfs Cache Pressure at 20. The kernel runs out of memory. Thats the only reason the device would reboot. Anything else would just cause a luna restart.

    I put it to 40 for the Pre Minus fix, but people on the android forum were talking about setting it to 5, so I thought 20 would be ok.

    I guess its too low.
  5. #45  
    Heres what I am considering changing:

    vm.swappiness = 10
    vm.vfs_cache_pressure = 20 considering 40 instead
    vm.dirty_expire_centisecs = 1200
    vm.dirty_writeback_centisecs = 500
    vm.dirty_background_ratio = 70 Considering 30 instead
    vm.dirty_ratio = 90 Considering 50 instead
    vm.min_free_kbytes = 1024
    vm.max_map_count = 65536
    vm.overcommit_memory = 1 tried 0, it made the system laggy
    vm.overcommit_ratio = 70 considering 10 instead
    vm.zone_reclaim_mode = 2
    kernel.shmax = 268435456
    kernel.shmall = 16777216
    kernel.shmmni = 65576
    sched_compat_yield = 1
    vm.lowmem_reserve_ratio = 1024

    But I have to try every change at once, and then let the device run for a whole day to really see the issues. Then I have to try them in combinations, because one can affect the other.

    It already took me 7 months on the Pre- to get to around these settings, its a slow effort of weeding out the issues/seeing if things are better or worse, checking ram usage in Govnah.

    Wash/Rinse/Repeat
    Last edited by rmausser; 11/30/2011 at 02:36 PM.
  6. #46  
    Quote Originally Posted by rmausser View Post
    comment out the line:

    "echo 6000 > /proc/sys/vm/dirty_expire_centrics"
    Weird, without the line commented out, but with your changes in sysctl.conf:

    Code:
    root@HPPre3:/etc/init.d# cat /proc/sys/vm/dirty_expire_centisecs
    1200
    Looks to me like the changes in sysctl.conf were read correctly during reboot, and the modfication in /etc/init.d/swaphack.sh was not needed on my Pre 3 (though it would not hurt either).

    Max
  7. #47  
    So, I see that mau's patch for this is now in preware - before I installed, I wanted to check and make sure it's compatible with Uberkernel, and if so, do we need to change the compcache settings in Govnah before/after installing?

    -- Sent from my Palm Pre3 using Forums
    If you like my patches or have found them to be helpful to you, please consider donating here for the time I put into making them.
    - All my programming is done in my spare time, so any donations help motivate me to spend more of my limited time on webOS.
    eny2002 likes this.
  8. eny2002's Avatar
    Posts
    66 Posts
    Global Posts
    68 Global Posts
    #48  
    ^^ I was wondering exactly that...
    ????
  9. #49  
    Works with any kernel.

    You dont need to change compcache settings but Id still put it to Disabled in Govnah and save that setting as a profile just in case.
  10. #50  
    Quote Originally Posted by rmausser View Post
    Works with any kernel.

    You dont need to change compcache settings but Id still put it to Disabled in Govnah and save that setting as a profile just in case.
    Trying out new patches (Sys Ctrl Optimization and Disable Compcache/enable linux Swap) now. I would really like to know what is going on with the load average of my Pre 3. After applying patches, I did a full device restart. Load avg started out at 12 7 3 or something. With just e-mail and calendar (+Govnah obviously) the latter two crept up to around 8 and 5, while the first one gradually came down all the way to 8! Turning on the phone after being idle for about 45 mins gave me 15(!) 9 7. After a while with screen on I am now at 8.2 8.9 7.2. Are the load avgs really supposed to jump up and down like that, especially with the phone more or less idle?

    Good thing to see load avgs come down though. Prior experience is they had a tendency to mostly creep upwards and maybe that eventually caused the lag to worsen.
  11. #51  
    Quote Originally Posted by tomoqv View Post
    Trying out new patches (Sys Ctrl Optimization and Disable Compcache/enable linux Swap) now. I would really like to know what is going on with the load average of my Pre 3. After applying patches, I did a full device restart. Load avg started out at 12 7 3 or something. With just e-mail and calendar (+Govnah obviously) the latter two crept up to around 8 and 5, while the first one gradually came down all the way to 8! Turning on the phone after being idle for about 45 mins gave me 15(!) 9 7. After a while with screen on I am now at 8.2 8.9 7.2. Are the load avgs really supposed to jump up and down like that, especially with the phone more or less idle?

    Good thing to see load avgs come down though. Prior experience is they had a tendency to mostly creep upwards and maybe that eventually caused the lag to worsen.
    The kernel (and therefore processor) has to work more aggressively to rid the memory of leaks and dirty pages with the settings I made in SysCtl.

    So thats probably what you are seeing at idle.
  12. pdch's Avatar
    Posts
    80 Posts
    Global Posts
    480 Global Posts
    #52  
    Um, I have been running my pre3 for over two weeks now and have not had to reboot it once. It hasn't even stuttered when I have multiple apps open copying from one to another. Are these problems you guys are talking about here something that shows up over time or am I just REALLY lucky?
  13. #53  
    Quote Originally Posted by rmausser View Post
    The kernel (and therefore processor) has to work more aggressively to rid the memory of leaks and dirty pages with the settings I made in SysCtl.

    So thats probably what you are seeing at idle.
    So the settings tell the kernel to more aggressively take care of memory leaks when it is less busy handling other tasks, is that correct? If so, how does that affect battery drain? I am also wondering why a luna restart doesn't seem to clear as much memory/swap as a full restart?
    Last edited by tomoqv; 12/01/2011 at 04:04 PM. Reason: Further question
  14. #54  
    Yes. It works to clear out memory leaks more aggressively.

    You might get some minor extra battery drain.

    A Luna restart just restarts the visual part of webos, which is actually a website hosted on Ur phone. Everything you see on webos is actually a big website hosted in a web browser that is fullscreen.

    Luna restart just refreshes the webpage.

    Device restart restarts all of linux.

    -- Sent from my Palm Pre3 using Forums
  15. #55  
    Quote Originally Posted by rmausser View Post
    Yes. It works to clear out memory leaks more aggressively.

    You might get some minor extra battery drain.

    A Luna restart just restarts the visual part of webos, which is actually a website hosted on Ur phone. Everything you see on webos is actually a big website hosted in a web browser that is fullscreen.

    Luna restart just refreshes the webpage.

    Device restart restarts all of linux.

    -- Sent from my Palm Pre3 using Forums
    I have got about 5 spontaneous resets today, which I rarely if ever experienced prior to installing these recent performance patches. The weird thing is that I went through more than 24 hours with no glitches at all right after installing the patches. And suddenly, after about 30+ hours I get 5 resets within about 6 hours. I'll keep testing. Overall performance seems better with the patches.
  16.    #56  
    Quote Originally Posted by rmausser View Post
    Heres what I am considering changing:

    vm.swappiness = 10
    vm.vfs_cache_pressure = 20 considering 40 instead
    vm.dirty_expire_centisecs = 1200
    vm.dirty_writeback_centisecs = 500
    vm.dirty_background_ratio = 70 Considering 30 instead
    vm.dirty_ratio = 90 Considering 50 instead
    vm.min_free_kbytes = 1024
    vm.max_map_count = 65536
    vm.overcommit_memory = 1 tried 0, it made the system laggy
    vm.overcommit_ratio = 70 considering 10 instead
    vm.zone_reclaim_mode = 2
    kernel.shmax = 268435456
    kernel.shmall = 16777216
    kernel.shmmni = 65576
    sched_compat_yield = 1
    vm.lowmem_reserve_ratio = 1024

    But I have to try every change at once, and then let the device run for a whole day to really see the issues. Then I have to try them in combinations, because one can affect the other.

    It already took me 7 months on the Pre- to get to around these settings, its a slow effort of weeding out the issues/seeing if things are better or worse, checking ram usage in Govnah.

    Wash/Rinse/Repeat
    Just sent these to my phone... gonna keep on trying till I run out of cake.
    Palm IIIc -> Sony CLIÉ T650C -> Sony TJ-37 -> Palm TX -> Palm Centro -> Palm Pre Bell -> Palm Pre Plus Bell/Verizon Hybrid -> HP Veer -> HP Pre 3 NA -> BlackBerry Classic -> BlackBerry Priv

    It's a Late Goodbye, such a Late Goodbye.

    Need OEM Palm Pre parts? See here
  17. #57  
    Hey, rmausser, I am subscribing to this thread to follow your developments.

    I just chanced upon this and the described symptoms seem very much like what I am experiencing with my Veers (described in the Veer forum section: http://forums.precentral.net/hp-veer...rebooting.html).

    Are your fixes going to be applicable to WebOS 2.1.2 as well?

    I hope you are going to make the patches relatively easy for the non-tech crowd. I am somewhat tech-literate, but after I followed all messages in this thread so far, I did get lost in the details.

    All the best, Robert.
  18.    #58  
    Ran for an entire day. A few lags noticed, but there's a significant drop in the amount of slider events missed (i.e. sliding my phone closed and the phone doesn't realize that). The slider issue only seems to happen when phone is in a lag, though.
    Palm IIIc -> Sony CLIÉ T650C -> Sony TJ-37 -> Palm TX -> Palm Centro -> Palm Pre Bell -> Palm Pre Plus Bell/Verizon Hybrid -> HP Veer -> HP Pre 3 NA -> BlackBerry Classic -> BlackBerry Priv

    It's a Late Goodbye, such a Late Goodbye.

    Need OEM Palm Pre parts? See here
  19. #59  
    Quote Originally Posted by tomoqv View Post
    I have got about 5 spontaneous resets today, which I rarely if ever experienced prior to installing these recent performance patches. The weird thing is that I went through more than 24 hours with no glitches at all right after installing the patches. And suddenly, after about 30+ hours I get 5 resets within about 6 hours. I'll keep testing. Overall performance seems better with the patches.
    No more spontaneous resets for the past three days. I don't know what got into my phone this past Saturday... Patches seem to work quite well. I guess I don't notice much of a lag now, even though after 1-2 days of use it starts to feel less responsive, prompting me to restart the device. One strange thing has surfaced, though, but it sounds very odd that it should be related to any of the patches: After a device restart, I need to re-enter the password for my home WiFi. Never needed to do that before.
  20. #60  
    Do the two patches (Sys Ctrl Optimization and Disable Compcache/enable linux Swap) both need to be installed; or can one just have the Disable Compache installed.

    I ask because when both were installed, battery drain was a little too much for me.
    However, just having Disable Compcache installed brings battery back to acceptable levels. Cannot tell objectively if just having that patch installed is improving the phone; but experientially, it seems to be running fine.
    This space for rent or lease. Inquire within.
Page 3 of 4 FirstFirst 1234 LastLast

Posting Permissions