patstew
|
2009-09-19
, 23:50
|
Posts: 69 |
Thanked: 53 times |
Joined on Aug 2009
|
#321
|
|
2009-09-19
, 23:56
|
Posts: 271 |
Thanked: 220 times |
Joined on Sep 2009
|
#322
|
Is this definitely the case? The reason I ask is that looking at http://mos.futurenet.com/techradar/c..._12-420-90.jpg and http://en.wikipedia.org/wiki/File:Types-usb_new.svg it looks like the N900 has a micro-B socket. USB-OTG devices are required to have a micro-AB socket (a square one without the bevelled corners). If it doesn't take micro-A plugs, then either it doesn't support host mode, or it doesn't comply to standards, or wikipedia is wrong. http://en.wikipedia.org/wiki/Univers...-AB_Socket_OTG
|
2009-09-20
, 08:09
|
Posts: 2,802 |
Thanked: 4,491 times |
Joined on Nov 2007
|
#323
|
Actually, ext3 is better for flash than having to repair the filesystem every time you run out of power. The journalling is a much smaller burden than fscking the whole fs on boot. So, no, you're mistaken.
|
2009-09-20
, 15:19
|
|
Posts: 5,478 |
Thanked: 5,222 times |
Joined on Jan 2006
@ St. Petersburg, FL
|
#324
|
Running out of power should be an exceptional event, while the journalling is active the entire time the filesystem is mounted. While it's on, every time you change a single bit on the fs the system will have to re-write at least two entire eraseblocks on the flash (one for the journal, one for the actual filesystem data/metadata, depending on the mount options used). Wear-levelling or not, that seems too excessive for my liking.
The Following 2 Users Say Thank You to GeneralAntilles For This Useful Post: | ||
|
2009-09-20
, 15:58
|
Posts: 2,802 |
Thanked: 4,491 times |
Joined on Nov 2007
|
#325
|
This has been discussed endlessly on #maemo, and much smarter people than myself preferred ext3 (including Nokia, it appears).
Observation of flash memory survival rates (nearly 100% in what I've observed) seems to bear this out.
|
2009-09-20
, 16:13
|
|
Posts: 5,478 |
Thanked: 5,222 times |
Joined on Jan 2006
@ St. Petersburg, FL
|
#326
|
Survival over what period and usage pattern? Keep in mind that flash writes impact the entire block device (not partition), so having things like swap, journal, .ash_history etc on the same chip have a cumulative wear effect.
BTW, there was some speculation that there's an extra (faster?) memory device in the RX-51 specifically for swap use, but I've lost track - was that ever confirmed either way? If there is, it could also be used to store the ext3 journal.
|
2009-09-20
, 16:25
|
Posts: 3,319 |
Thanked: 5,610 times |
Joined on Aug 2008
@ Finland
|
#327
|
Survival over what period and usage pattern? Keep in mind that flash writes impact the entire block device (not partition), so having things like swap, journal, .ash_history etc on the same chip have a cumulative wear effect.
The Following 4 Users Say Thank You to attila77 For This Useful Post: | ||
|
2009-09-20
, 17:05
|
Posts: 227 |
Thanked: 53 times |
Joined on Feb 2008
@ Lyon, France
|
#328
|
|
2009-09-20
, 17:06
|
Posts: 2,014 |
Thanked: 1,581 times |
Joined on Sep 2009
|
#329
|
This has been discussed endlessly on #maemo, and much smarter people than myself preferred ext3 (including Nokia, it appears). Observation of flash memory survival rates (nearly 100% in what I've observed) seems to bear this out.
I'll stick to ext3, thanks.
The Following User Says Thank You to Bratag For This Useful Post: | ||
|
2009-09-20
, 18:51
|
Posts: 3,319 |
Thanked: 5,610 times |
Joined on Aug 2008
@ Finland
|
#330
|
Not all users enabled swap space (I didn't), and when it is enabled, I wonder whether it is intensively used in practice.
The Following 3 Users Say Thank You to attila77 For This Useful Post: | ||