contents of usb permenently in ram

[666philb]

i'm posting this hoping someone can shed some light on what happened ..

i'm running puppy live on a macbook pro ... no harddrive and no savefile ..just live. current uptime 32days . i can't reboot the thing as it has a faulty graphics chip that comes away from the motherboard!! if it goes off i have to dismantle it and put it in the oven and bake for 8 mins (the previous uptime was similar but when drunk one night i rebooted and needed to bake the thing again!)... it's had so many bakes now 10+ ...that if it goes down again that may be it.

so that's why i'm running completely live.
it's quite amazing ... puppy splits the 4gb ram , half for ram usage and the other half acts like a savefile that needs to be managed.

i have an 8gb sd card for storage but either the sdcard is corrupt or the motherboard is damaged from so much baking, so i can't trust the card. but i need to use it as a storage device as i have limited ram .. and this is where it gets odd.

i had a googleearth.sfs loaded from the sdcard and it worked fine even though it was on the untrusty card. the sd card then got knocked and the whole filesystem became read only & un-mountable. there was music ...pets ...pictures etc on the card aswell.
i couldn't change the sdcard back to read/write so i took the un-mountable card out and formatted in another computer.

when i put the blank/formatted card back in the mac, ALL files that where previously on the card where there and available ... music tracks played, films played !! it must have copied the entire contents to ram.
it didn't recognise the card as empty until i unmounted the (not there) googleearth.sfs off the card.

i can understand that the googleearth.sfs if unloaded would still be in ram, but does anyone have any idea why all the contents of the sdcard (music ,films etc) would be retained in ram?

[mikeb]

Rox would keep the previous contents visible until you make it refresh...thats what comes to mind. ie they were gone but Rox was a bit behind the times.

mike

so is this like a bakers dozen?

[666philb]

hi mikeb,

defo not a rox thing, the files were retained in memory. i could play a film or tune from the sdcard... even though the sdcard had been formatted!!

[mikeb]

In that case not a clue.

The sfs would be copied to tmpfs in the initrd I believe if loaded on the fly.... the rest dragged in...who knows....

/initrd/mnt/tmpfs and tmpfs2 are the places..

mike

only other rox thought ...if rox is open in a mounted folder (NFS, drive etc ) then unmounting is not possible.... perhaps it hung onto files in true posix style...ie a file is still there until released in the ether.

[amigo]

Formatting a drive doesn't really remove any data. It simply lays down 'markers' in certain locations. So, if you reformat it with the same filesystem as before, there is a good chance that the files will still be there, intact.

[mikeb]

Formatting a drive doesn't really remove any data. It simply lays down 'markers' in certain locations. So, if you reformat it with the same filesystem as before, there is a good chance that the files will still be there, intact.

would the theory on that then be the old valid pointers to the data were still in a buffer until the reformatted drive was truly reloaded/rescanned?

a curious can of worms. I have recovered data via reformatting/partitioning back to as was (quick format) before today.

mike

[666philb]

thanks for the replys .... kind of throws some light onto what happened ... even though i formated the drive ..... the pointers to all the files were still in the computers memory and therefor i could still play the deleted files.

i believe it was to do with the sfs_loading mechanism as opposed to rox that caused the weirdness.

it was very refreshing running puppy completely live for 5 months (no save), unfortunately i had to finally turn the macbook off, and it may well never turn back on again
i'm now back on my main computer with a savefile and even though it's a faster comp ... it actually feels a little slower in general .... as not everything is being completely done in ram.

[dogle]

I experienced some very vexing weirdness a few months ago .... which may or may not be related to your issues. (I run 431 in RAM from multisession disk most of the time).

After a lot of head-scratching and bad language, it was sorted; the trigger in my case seems to have been RFI spikes being picked up via my cabling rats-nests, and the underlying cause an old CMOS coin battery which had become too weak to hold things in place. Whatever, replacing the battery fixed it.

I suggest that you might, in the absence of any better ideas, just try substituting a new CMOS backup battery and see what happens (don't be fooled by a voltmeter check on the old one, which may mislead you if it is not really up to scratch)