In fact, if you have root access, a "$su #echo all > /proc/yaffs #cat /proc/yaffs" would be optimal. This should give a fair amount of debugging info for system, userdata, and cache. If you post that up here, I should be able to give it a shot.
On Mon, Jan 12, 2009 at 3:54 PM, Joel Knighton <[email protected]>wrote: > Okay, someone who can replicate this problem, can you perform a "cat > /proc/yaffs" and then post the output here. Curious to see the YAFFS > debugging info. > > > On Mon, Jan 12, 2009 at 3:45 PM, Jean-Baptiste Queru <[email protected]>wrote: > >> >> When I tried to reproduce it a few months ago I think that I was able >> to reproduce it without such a shortcut, but I might be wrong. >> >> JBQ >> >> On Mon, Jan 12, 2009 at 1:42 PM, Stoyan Damov <[email protected]> >> wrote: >> > >> > You might want to check whether this is related to having a shortcut >> > of the app on the home screen. I have a hunch it is. >> > >> > Cheers >> > >> > On Mon, Jan 12, 2009 at 11:39 PM, Jean-Baptiste Queru <[email protected]> >> wrote: >> >> >> >> There is a bug somewhere (it's assigned to me for investigation) where >> >> the system process keeps apk files open after they get unlinked in >> >> some scenario close to what you mention (install, launch, uninstall), >> >> which can then trigger the yaffs2 leak bug. >> >> >> >> JBQ >> >> >> >> On Mon, Jan 12, 2009 at 1:35 PM, Stoyan Damov <[email protected]> >> wrote: >> >>> >> >>> Thanks. >> >>> >> >>> A little bit more info about that bug - I *am* experiencing it again. >> >>> It *is* related to re-installs of one and the same application over >> >>> and over again. >> >>> I reinstalled my app maybe ~20 times today and slowly my ~70 went to >> 63 MB. >> >>> My app is 2MB so I have to have ~68MB but I don't. I noticed the >> >>> browser took 2MB and deleted them. >> >>> However, the free memory increased by ONE MB. >> >>> >> >>> WTF is going on here? >> >>> >> >>> Hurray! :) I did the battery pull and apparently I've hit the right >> >>> moment to pull the battery. >> >>> My memory increased from 64 to 69 MB. Now, how the device pulled that >> >>> off is a mystery to me because my app is 2MB (perhaps the ~70 MB are >> >>> close to ~71) but what the heck, the good thing is that the bug is >> >>> indeed *this* one, and not another which I'm the only one >> >>> experiencing! >> >>> >> >>> Problem "solved", THANKS to everybody! >> >>> >> >>> Cheers >> >>> >> >>> >> >>> On Mon, Jan 12, 2009 at 11:06 PM, Jean-Baptiste Queru <[email protected]> >> wrote: >> >>>> >> >>>> Second hand information about the battery trick: yaffs2 has some >> >>>> sanity-checking code that can detect and recover from unlinked files, >> >>>> but that code is only run when the filesystem wasn't cleanly >> >>>> unmounted. >> >>>> >> >>>> JBQ >> >>>> >> >>>> On Mon, Jan 12, 2009 at 1:02 PM, Stoyan Damov < >> [email protected]> wrote: >> >>>>> >> >>>>> On Mon, Jan 12, 2009 at 10:55 PM, Dianne Hackborn < >> [email protected]> wrote: >> >>>>>> >> >>>>>> Another place to look -- there is a filesystem bug that can >> sometimes happen >> >>>>>> where unlinked files are not recovered. Here is the comment from >> an >> >>>>>> engineer who knows more about it: >> >>>>>> >> >>>>>> "They can easily tell by looking at the number of unlinked files >> for the >> >>>>>> user partition in /proc/yaffs. If that number is very large, then >> they can >> >>>>>> reboot the device, wait a few second after they see the android, >> then pull >> >>>>>> the battery again. That should make the number of unlinked files >> drop back >> >>>>>> down. If that number isn't very large, then it is probably >> something else." >> >>>>> >> >>>>> I read about this on the net -- I thought it was some sort of a dark >> >>>>> joke or something -- apparently not :) >> >>>>> I did pull the battery though - nothing (good) happened. >> >>>>> >> >>>>> This developer you're talking about - can he elaborate on how the >> >>>>> "battery pull trick" actually works -- I'm genuinely interested. >> >>>>> >> >>>>>> >> >>>>>> Unfortunately it looks like only root cat read /proc/yaffs (though >> that >> >>>>>> seems a little overly restrictive). However you can try the >> pulling the >> >>>>>> battery trick and see if that helps. >> >>>>>> >> >>>>>>> >> >>>>>>> Well, the over-the-air patch @#$%ed root access so I can't look >> anywhere. >> >>>>>> >> >>>>>> The /data/local directory is owned by the shell user, so you don't >> need root >> >>>>>> for that -- just "cd /data/local" and look at what is there. There >> is a >> >>>>>> chance that some temp .apk files have been left there from "adb >> install", or >> >>>>>> some other files created by other shell sessions. >> >>>>> >> >>>>> I already reset the phone but if I encounter this again I'll check >> >>>>> there (+ I'll have root this time :) >> >>>>> >> >>>>> Thanks, >> >>>>> Stoyan >> >>>>> >> >>>>> > >> >>>>> >> >>>> >> >>>> >> >>>> >> >>>> -- >> >>>> Jean-Baptiste M. "JBQ" Queru >> >>>> Android Engineer, Google. >> >>>> >> >>>> > >> >>>> >> >>> >> >>> > >> >>> >> >> >> >> >> >> >> >> -- >> >> Jean-Baptiste M. "JBQ" Queru >> >> Android Engineer, Google. >> >> >> >> > >> >> >> > >> > > >> > >> >> >> >> -- >> Jean-Baptiste M. "JBQ" Queru >> Android Engineer, Google. >> >> >> >> > > > -- > Joel Knighton > -- Joel Knighton --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Android Developers" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/android-developers?hl=en -~----------~----~----~----~------~----~------~--~---
