Re: [Announce] CSSU Testing thread
Quote:
Estel - what do you mean by "swap in use"? i've seen several places mentioning swap usage - can you reference me to relevant discussion board (can it help with the problem i'm having?) Anyway like you said - controlling the delay time will be a great thing. Thanks. |
Re: [Announce] CSSU Testing thread
nice
already have it, work perfectly |
Re: [Announce] CSSU Testing thread
@impeham - I wanted to send You PM with answer, but for some reason, I'm unable to PM You (some stupid forum rules limitation, I think) - so, I'm answering here, despite it being off-topic.
Generally, inside Your N900, You have 2 different storage 'devicesm - 256 MB NAND (super fast, at least compared to Your every day flash devices) - where most important (in terms of access times and read/write speeds) system files reside, and 32 GB eMMC, which is "regular" flash storage, and files from optified packages rest there, (symlinked, so system see them as sitting on same storage). On eMMC, You also have - by default - swap partition (if You don't know what swap is, You can google with ease, it's well-documented thing). Now, SWAP is used pretty often - given the amount of RAM we have (256 MB). Flash-based storages doesn't handle many different input-output requests at once (especially true about write operations), and having opt and swap on same storage medium (physically) result in many conflicting I/O requests. Fortunately, on N900 You have another storage medium available - microSD. By creating linux SWAP partition there (using linux machine, linux live CD, or N900 itself, if You feel comfortable with partitioning via terminal commands), enabling it and disabling eMMC swap partition, You get rid of such I/O collisions. For details, see: http://wiki.maemo.org/Swap_on_microSD (You'll find also information on using both eMMC and microSD swap with same priorites - scratch that part, it's outdated, and assumption about performance gain by increased bandwidth - aka RAID0 philosophy - doesn't consider I/O collisions on eMMC, so it's wrong on basics). Generally, if You have swap type partition on Your microSD, You need to place script like tht: Code:
description "swapoff eMMC, swapon MMC" Another thing is swap fragmentation - occurring, after total amount of data written to swap partition (during one session) exceed total size of swap partition (this occur, due to write-erase characteristic of flash storage media) - it can be 'fought' by simply disabling and enabling swap again, but to measure when it's needed, You need simple iostat-based script, which parts of it bundled also inside Your (mentioned already) startup script it /etc/event.d/. Ho ever, it's quite too wide topic for this post, being already *epic* off-topic - thinking about that, I may create separate thread summarizing around-swap knowledge, as it's now spread around whole forum. My augmented version of shadowjk's measuring (anti swap-fragmentation) script, may be also of interst o some, so I'll post it there as well. /Estel |
Re: [Announce] CSSU Testing thread
Estel - thanks a lot for the explanation - i'll create another partition on the memory card for swap and check if it solves the performance issue.
The fragmentation fix script will also be most appreciated :) |
Re: [Announce] CSSU Testing thread
Quote:
|
Re: [Announce] CSSU Testing thread
Bug:
"End Call" button not displayed for phone interface when device rotation is locked locked to portrait. It looks as if the "End Call" button is trying to display below the screen. |
Re: [Announce] CSSU Testing thread
Quote:
|
Re: [Announce] CSSU Testing thread
+1 on this.A solid wiki entry would help many people including noobs in this matter like me :)
|
Re: [Announce] CSSU Testing thread
any news about next update???
|
Re: [Announce] CSSU Testing thread
Quote:
// Edit - it's nice, that You changed it to less demanding form, but it's still kinda useless question. Uless getting answer like WIWBD is Your fetish. |
All times are GMT. The time now is 18:11. |
vBulletin® Version 3.8.8