[lkml]   [2000]   [Feb]   [16]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: Of removable devices
In <200002161445.JAA14399@tsx-prime.MIT.EDU> Theodore Y. Ts'o (tytso@MIT.EDU) wrote:
TT> From: Francis GALIEGUE <>
TT> Date: 16 Feb 2000 12:57:17 +0100

TT> "Khimenko Victor" <> writes:

TT> First, sorry for yelling earlier. I'm conscious of problems with PC
TT> floppies. Calling block_fsync() regulary is not even a solution, I'm
TT> aware of this as well - too much overhead for not so much. But a
TT> newbie is not as stupid to eject a floppy when he sees the LED is on
TT> (well, at least I hope so :)

TT> Right. That was the basic design behind supermount, as I understand
TT> it. Users have been well programmed that if the LED light is off, it's
TT> safe to pop the disk out. Because Linux normally has different rules,
TT> this violates the principal of least surprise.

TT> So the question is whether it's possible to do something relatively
TT> painless which allows the right thing to happen under most
TT> circumstances. A key observation is that the sort of programs you
TT> expect naive users to use: i.e., cp, office suite software,
TT> etc. generally tends to open a file for writing, write it out
TT> completely, and then close the file. So if you can automatically
TT> dismount the filesystem after the close, and then automatically mount
TT> the filesystem after something tries to touch the dentry, it works 99.9%
TT> of the time. Is it perfect? No! Is it better than hoping that users
TT> will adapt to Linux? Well, that's a religious question, but the
TT> traditional answers is that users are (a) resistant to change and (b)
TT> highly stubborn, and so it's better to try to make at least some
TT> consessions to pre-programmed behaviour.

TT> By the way, ext2 has a serial number, so it *would* be possible to make
TT> ext2 check the superblock parameter when it was informed that a media
TT> change may have happened. So it's not just the FAT filesystem that can
TT> do this.

This is all doable as I said from the very beginning but it'll require DEEP
rework of VFS, not just small changes. It IS usefull (for example for discussed
here situation where you have big disk with rarely used stuff and want to not
re-fsck it after crash if in fact disk was not used) but it's also HARD work
if you want to do it right.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:56    [W:3.327 / U:45.532 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site