On Wed, Sep 25, 2002 at 01:18:46PM -0400, Johannes Erdfelt wrote: > > That's good. I'll cleanup the patch and send it on the Greg. Thanks for > the testing.
Good, as I think I'm seeing the same problems on some of my machines :) > > I'm incrementing usage count by side effect :(. I've implemented a > > ram-based driver fs similar to what Greg put in PCI Hotplug. The > > problem is that I've called kern_mount in my connection sequence. > > That is incrementing the usage count. I don't know how to get around > > this. The disconnect routine is the only place that I'm unmounting > > the driver fs. Is there some routine that I can know will get called > > at module unload time to insure my driver fs is unmounted? > > Do you really need to mount the filesystem from the module? Why not just > let userspace do that? kern_mount is different from userspace mounting the fs. It's needed to be done so that the driver can create files within the fs and have that work if the fs is mounted at a real mount point or not. See the gyrations I had to do in the usbfs code in 2.5 for an example of what I'm talking about. thanks, greg k-h ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
