>Now, that's just rude. I mean, why can't a reference count be maintained >(by hsfs_mount(), hsfs_unmount(), and maybe one or two others) so that >modunload would be permitted whenver thereference count was zero? >Granted it might actually take a bit more than that, but I'd like to think it >still wouldn't be too hard.
>The more modules are unloadable, the more patches one could be able to >install (with some extra work in either associated scripts or manually) >without having to reboot. Some modules that would always be truly >busy (not just too lazy to provide a way to unload them) would still >require a reboot to pick up the new version; still, anything that helps >reduce the number of patches for which one can't work around a reboot >while getting the benefit of the patch is IMO an improvement. Absolutely; go write code :-) There's still, of course, the issue of having to remove all CDs and stuff when patching so there should be a mechanism which attempts unloading and if it does not succeed suggests unloading or alternate steps. Certainly, in many cases reboots aren't strictly necessary other than the get the patch to be in effect. Casper _______________________________________________ opensolaris-discuss mailing list [email protected]
