[BusyBox] One more #*%(&#% umount corner case.

[Rob Landley]

On Friday 05 August 2005 17:29, Rob Landley wrote:

> Now, even in that case it shouldn't cause a problem.  If it's not there the
> syscall should fail even if our mtab checking logic still thinks it's there
> the second time.  But how about this one?
>
> mount -t ramfs /thingy
> mount /dev/ram0 /thingy
> umount /dev/ram0 /dev/ram0
>
> Because umounts of block devices are (where possible) converted into
> umounts of the corresponding directories (which is the only way to make the
> "umount all instances of this block device" logic work; iterate through all
> of them),

It's not the only way: if you feed the block device to the syscall the kernel 
will umount the most recent occurrence itself, so if block devices aren't 
translated even this corner case can be made to work.

Although I still have to add a test for this.  After dinner, I think.

My mount rewrite is once again compiling.  I still need to update the usage 
for the new command line arguments (umount -d is now the default, and -D has 
been added to _not_ do that.  And umount -l will now lazy umount.  And umount 
-f is now small enough it's not optional anymore, but I need to update the 
config to yank out that option.  And no, it shouldn't be optional because 
"old kernels don't support it", either.  I've isolated the header 
dependencies (hard wired in the constants with comments about what they came 
from) and the umount2 syscall itself was added during 2.1 so it should work 
on any kernel we remotely care about...)

Lots of testing and polishing still needed, but no more cans o' worms: it's 
finally feature complete.  (And I'm tired...)

Rob