View Single Post
peterleinchen's Avatar
Posts: 4,118 | Thanked: 8,901 times | Joined on Aug 2010 @ Ruhrgebiet, Germany
#22
Originally Posted by Mohammed Muid View Post
what shud i do?
Sorry, but I have to say: sell and get yourself an Android (this is the N900 and not a cell phone.)

Originally Posted by MaddogG View Post
Or, you can try swapset. I use it and it works pretty fine.
Thanks a lot for the hint to this package (script).
This seems to me as the best trial for newbies. I checked sources and its nicely done.
Only the compcache I am worried about. Followed the zram thing, but it stopped somehow. I do not know the latest status, so I would like to ask anybody about feedback (otherwise I would propose to disable that?).

Originally Posted by Estel View Post
@peterleinchen
As I've written, way of doing ti via /etc/event.d/ script isn't bad at all, but it got one drawback - boot time is considerably slower, as Maemo need to enable eMMC swap first, then, via Your script, enable microSd swap, and disable eMMC swap.
No, as enabling swap is very fast, it doesnt consume boot time. Further I do not need to disable eMMC swap, as I start SD swap with higher priority. Last I do not disable eMMM swap, as it will not be used anymore at all (lower priority, it would only be used when my swap overloads). Just a few 4-5 MB are left on eMMC which were used during time between ke-recv and my script.

As - at this point - some thing are already swapped (most likely, You run event.d scripts after starting hildon-desktop), they need to be moved from eMMC swpa to microSD swap.
Again no, as I start my script like
start on started ke-recv
and then wait until SD and/or eMMC are mounted.
As desribed above, tme swap ram is not moved back. At least not immediately, but slowly during runtime.


As for using microSD swap together with eMMC swap, I don't recommend it, even with higher priority for microSD swap - it's still "in use", which creates (less, but still), I/O rade with things from optFS.
Just a No, as explained above.
Or at least what I understood with my limited Linux skills.


Also, I'm quite sure, that using loopback file for swap is considerably slower than native swap - as it's with ED loopback image or native partition.
This is a full Yes from your experiences with ED. But I do not know that in detail for swap space.
Again, limited Linux skills.

Please see this post (my explanations) as discussion base.
I think it is a nice way to do and it helps to move I/O load away from eMMC.

I think about merging my approach (using of swap file also possible) with the swapset script (starting it earlier, this is started on hildon-desktop) ...
 

The Following 2 Users Say Thank You to peterleinchen For This Useful Post: