Hi again. On Tue, 2008-10-28 at 21:25 +0100, Miklos Szeredi wrote: > On Tue, 28 Oct 2008, Nigel Cunningham wrote: > > The answer is to freeze the fuse filesystems first, stopping new > > requests (freezing the processes making them) before they are passed on > > to userspace and allowing existing requests to complete or freeze. Once > > that is done, the userspace fuse processes will be idle, at least as far > > as fuse activity is concerned. After fuse activity is stopped, userspace > > can be frozen without worrying about what processes are fuse and what > > aren't. This is what my patch implements so far. > > OK. > > > To deal with requests that are already in progress, I'd suggest three > > possibilities, in the order I think they should be preferred. > > > > 1) Use wait_event_freezeable[_timeout] in fuse code. Probably preferable > > to #2, but I thought of it later :) > > > > 2) Use freezer_do_not_count as part of FUSE_MIGHT_FREEZE (resetting > > before exiting the callers, of course). If the request doesn't complete > > during the freezing period, it must be because the userspace thread was > > already frozen. If the request does complete, we're counted again during > > the normal freezing of userspace that follows the freezing of > > filesystems. > > > > 3) Adding a means to check whether processes being frozen are fuse > > requests. The code could then wait for fc->num_waiting - (say) > > fc->num_frozen == 0. > > Yup, these fix the freezing of tasks which have outstanding fuse > requests. > > However it does not fix the freezing of tasks which are waiting for > VFS locks (i.e. inode->i_mutex) held by the outstanding fuse requests. > This is the tricky part... Okay. Looking back on our conversation brought me back to this message, which I think needs another reply. If we take the strategy of holding new requests and allowing existing ones to complete, then am I right in thinking that we only need to worry about where inode->i_mutex is taken in fs/fuse/file.c (I don't see it taken in other fs/fuse/*.c files). If that's correct, dealing with that issue looks relatively straight forward: we need some more FUSE_MIGHT_FREEZE calls for those functions, and something done to the vfs_check_frozen call - I'm a bit confused by that - inode->i_sb will refer to the fuse filesystem, won't it? Regards, Nigel _______________________________________________ linux-pm mailing list linux-pm@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/linux-pm
