i remeber reading somewhere about a magentic sensor that is used to detect when the back is removed and unmount the sdcard...

Now that I've mentioned the name, anyone who wants further information can look it up.

Two things:
1. The 256mb rootfs is formatted to Ubifs. This filesystem does not need a fsck tool, and seems to be /very/ resilliant when it comes to corruption - Seriously, I haven't ever had it fail.
The optfs, on the other hand, is just ext3, and it /is/ fairly succeptable to corruption. Maemo will attempt to fsck it upon boot, but it doesn't always work, resulting in the "five dots" reboot loop.
Mydocs and the SD card aren't needed for boot.

Now, Backupmenu only uses files on the rootfs, so it's a lot more resilliant to optfs corruption: Wipe the optfs, it'll still work.


2. Fusi, yes, there's a magnetic sensor. When the back cover is removed, the SD card is unmounted(so it's safe to remove).

Joerg mentioned in other topic, that cutting of power from microSD card slot, wehn card is in use, can - in rare occasions - hardware damage card. And he didn't mean dead loop of waiting for initialization but something pure-hardwarish about way that data is recorder and electrones Sorry, can't find this post right now.

What's the point is that removing battery = removing power from SD slot also (not big deal, cause unless you made hardware trick by sticking magnet to sensor, your card is already unmounted/hard powered off when your backcover is removed), but what's even more important, internal flash is also vulnerable to same hardware damage.

On the other (probably third at this time ;P ) hand, joerg also mentioned - in hotswap battery method - that bme shut down system in <1 second if battery is removed, i.e. control (third) pin is off. That's probably limit possible problems to single file damage, or filesystem error fixable by fsck at worse condition. If BME never fails in its duty, hardware damages are off-limit.

Don't ask me if same thing happen during use of HEN - as far as i know BME is stopped at this time, so i presume that unless we got working bme open source replacement, this may be extremely dangerous behavior during hostmode. But feel free to correct me, if any other entity monitor / hardware monitor watch for this in absence of BME. (still i think this is last defense line, cause stopping BME allow for hotswap)

Last theoretical divagation - if dme can be swapped out during heavy I/O bug so the phone reboots due to watchdog, who knows if BME isn't vulnerable to same swapout problem, -> in some VERY rare occasion of swap horror can not be able to shut down machine on time.

That's all theory (still, hard-proofed). In practice, If you don't feel suicidal and don't try to kill your device on purpose, it's almost impossible to damage device doing that. Almost. Although i fail to see any rationale for using this, when holding power button for solid 8-10 seconds (sorry again, don't remember exact number, may be also 15) launch hardware-controlled power down (in theory should be able to power down device always, like holding power button for 10 seconds in every laptop - in practice i heard about 1, literally 1 occasion, when even power button wasn't able to power down device and battery removal was a must - that's why on wiki is info about ''last resort'' method of removing battery)

// Edit

Was writing on exact same time as Robbie1ST, so sorry for cross-examining the same problems in few parts of my post. and thank You Robbie for mentioning NAND flash filesystem, that's also very important factor that i missed.

couple of things... shortcircuit is unlikely but possible as while removing the battery slowly u mite shortcircuit it

when the N900 is shutdown with the power off button it sdoes an internal shutdown which wont happen if u pull the battery out

if u have a custon kernal installed u mite get the message unable to load your custom kernel settings..

hope it helps.... cheers

Wrong!

each time you remove the battery while phone is ON Nokia kills a kitten

Maemo5 doesn't run fsck on ext3 filesystems.

It's not a very good idea in many circumstances involving flash memory.

I think - I THINK; do not assume I'm right - there's something in the scripts that run at boot that runs fsck if some conditions are met. I'm too newb to know what each bit of most of those scripts actually means though, and I don't even remember them clearly as the last time I looked was a long time ago - so don't trust me in that regard.

Meanwhile, though, I want to point out that as far as I know, Maemo, if it does fsck opt, does so very rarely. However, there is an option in the power kernel to enable either always or "when-necessary" fsck-ing. I doubt the power-kernel would have come with its own options to do the same thing if the default behavior, whatever it was, wasn't deemed insufficient.