usb disk, in reverse?

[Chris Murphy]

On Dec 10, 2013, at 3:54 AM, Patrick O'Callaghan <pocallaghan at gmail.com> wrote:

> 
> On Mon, Dec 9, 2013 at 12:39 PM, Tim <ignored_mailbox at yahoo.com.au> wrote:
> Chris Murphy:
> >> Offhand I'm not thinking why a TV would need mount media as
> >> read-write.
> 
> 
> Patrick O'Callaghan:
> > In principle, it would allow the TV to flag shows that have already
> > been seen, or even delete shows under user command. Not that I know of
> > any that actually do this of course.
> 
> Or remember the user-corrected rotation of photos that have been viewed.
> 
> Indeed, and no doubt other features would come to mind, e.g. remember current position in video so the user can pick it up on another viewing device, reorganize videos into folders, configure per-folder sorting criteria, etc. etc. We could make TVs as complicated as phones (remember when phones were simple?)

Yes, I agree all of those features are good ideas. I'd just expect, due to the present state of affairs and past experience with embedded firmware modifying FAT16/FAT32/exFAT media, that the device will eventually corrupt the file system. And if that happens with exFAT, you're almost certainly SOL. If it happens with FAT32, there's a better chance of repairing the file system but it also depends how aggressively the device updates both FATs.

Granted, if they're implementing a simple file system this poorly, perhaps they'd implement a more sophisticated file system poorly also. But that just means the manufacturers are essentially incompetent in this particular area, which is my main argument.

> 
> The irony is that many of these devices already run Linux under the covers, you just can't get to it.

That's interesting. In which case I'd like to think they're using the linux kernel's FAT implementation, which shouldn't induce the corruption problems I've mentioned. It could become corrupt by removing media while writes are still occurring, which is normally in the category of user error. But in the context of a consumer device like a TV, in my view it's actually a design flaw that a user could induce such corruption. Losing the data that's unwritten is understandable, but corrupting the ability to access what was long ago successfully written, isn't.

I wonder if any of these linux running products happen to have a kernel with other file system support. Documentation might not mention it but maybe ext3 or ext4 formatted media would also work (even if not officially supported).


Chris Murphy