On Thu, 20 Mar 2003, Jan Ciger wrote: > Actually, the supermount is still buggy, just a bit less than the old one in > 9.0 - yesterday my USB Zip drive (the new 750MB variety) started to act up, > showing the free capacity of the disk to be only cca 100MB, even when only > 300MB file was on the disk. Disabling supermount and rebooting solved the > problem. > > Not to mention the annoying scanning of the Zip drive accompanied by a > jet-like noise of the drive spinning up every time I enter the /mnt directory > or connect some USB device (e.g. my webcam or Sony Clie). Actually, this > scanning blocks the system for few seconds completely, until the drive spins > up. Pretty bad, IMHO. And I am using the multimedia kernel, with low latency > on, kernel preempting on, so you would expect this not to happen.
Andrey Borsenkov has done a tremendous job of improving supermount. Actually it is an almost complete rewrite which could not go into the main kernel yet. I did decide to already put it in kernel-multimedia because it was much stabler on my boxes (and it had some trivial conveniences with the other patches I added). However, there may ofcourse be bugs than I am not aware off. If the zipdrive problem was the kernel-mm 16mdk, it would be very nice if you can try to reproduce that case. About the jet noise: If you enter mnt with konqueror, yes, the drive may start spinning. This is a konqueror bug, and the patch (checking whether the fstab entry contains 'none' is actually a hack, and should be improved). Also, if you configured the zip with fstype=auto, it may try to mount it as UFS, which will take a long time. I'm testing a new version of supermount now which will let you specify multiple fstypes per device, so hopefully that will solve the issue (in addition, it will let you remove the CD even when files are still open (which includes being in a dir on the drive with konqueror). Since it make take a while before the kernel team decides to include this new supermount (I think Juan would like to very critically look at it, and I do not when he has time). But if people like it, I can make kernel rpms available with these patches available somewhere. Either in the kernel-mm (but Adam was right that this should not be the correct place for testing these things) or somewhere else (ideas). This way, we will be able to do a lot of testing, with more hardware. And hopefully get a very good supermount in 9.2. danny
