On Thu, Feb 28, 2019 at 02:47:57PM +0300, Kirill Smelkov wrote: > On Thu, Feb 28, 2019 at 09:10:15AM +0100, Miklos Szeredi wrote: > > On Wed, Feb 27, 2019 at 9:39 PM Kirill Smelkov <kirr@xxxxxxxxxx> wrote: > > > > > I more or less agree with this statement. However can we please make the > > > breakage to be explicitly visible with an error instead of exhibiting it > > > via harder to debug stucks/deadlocks? For example sys_read < max_write > > > -> error instead of getting stuck. And if notify_retrieve requests > > > buffer larger than max_write -> error or cut to max_write, but don't > > > return OK when we know we will never send what was requested to > > > filesystem even if it uses max_write sized reads. What is the point of > > > breaking in hard to diagnose way when we can make the breakage showing > > > itself explicitly? Would a patch for such behaviour accepted? > > > > Sure, if it's only adds a couple of lines. Adding more than say ten > > lines for such a non-bug fix is definitely excessive. > > Ok, thanks. Please consider applying the following patch. (It's a bit > pity to hear the problem is not considered to be a bug, but anyway). > > I will also send the second patch as another mail, since I could not > made `git am --scissors` to apply several patched extracted from one > mail successfully. Ping. Miklos, is there anything wrong with this patch and its second counterpart? Thank beforehand for feedback, Kirill > ---- 8< ---- > From: Kirill Smelkov <kirr@xxxxxxxxxx> > Date: Thu, 28 Feb 2019 13:06:18 +0300 > Subject: [PATCH 1/2] fuse: retrieve: cap requested size to negotiated > max_write > MIME-Version: 1.0 > Content-Type: text/plain; charset=utf-8 > Content-Transfer-Encoding: 8bit > > FUSE filesystem server and kernel client negotiate during initialization > phase, what should be the maximum write size the client will ever issue. > Correspondingly the filesystem server then queues sys_read calls to read > requests with buffer capacity large enough to carry request header > + that max_write bytes. A filesystem server is free to set its max_write > in anywhere in the range between [1·page, fc->max_pages·page]. In > particular go-fuse[2] sets max_write by default as 64K, wheres default > fc->max_pages corresponds to 128K. Libfuse also allows users to > configure max_write, but by default presets it to possible maximum. > > If max_write is < fc->max_pages·page, and in NOTIFY_RETRIEVE handler we > allow to retrieve more than max_write bytes, corresponding prepared > NOTIFY_REPLY will be thrown away by fuse_dev_do_read, because the > filesystem server, in full correspondence with server/client contract, > will be only queuing sys_read with ~max_write buffer capacity, and > fuse_dev_do_read throws away requests that cannot fit into server > request buffer. In turn the filesystem server could get stuck waiting > indefinitely for NOTIFY_REPLY since NOTIFY_RETRIEVE handler returned OK > which is understood by clients as that NOTIFY_REPLY was queued and will > be sent back. > > -> Cap requested size to negotiate max_write to avoid the problem. > This aligns with the way NOTIFY_RETRIEVE handler works, which already > unconditionally caps requested retrieve size to fuse_conn->max_pages. > This way it should not hurt NOTIFY_RETRIEVE semantic if we return less > data than was originally requested. > > Please see [1] for context where the problem of stuck filesystem was hit > for real, how the situation was traced and for more involving patch that > did not make it into the tree. > > [1] https://marc.info/?l=linux-fsdevel&m=155057023600853&w=2 > [2] https://github.com/hanwen/go-fuse > > Signed-off-by: Kirill Smelkov <kirr@xxxxxxxxxx> > Cc: Han-Wen Nienhuys <hanwen@xxxxxxxxxx> > Cc: Jakob Unterwurzacher <jakobunt@xxxxxxxxx> > Cc: <stable@xxxxxxxxxxxxxxx> # v2.6.36+ > --- > fs/fuse/dev.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c > index 8a63e52785e9..38e94bc43053 100644 > --- a/fs/fuse/dev.c > +++ b/fs/fuse/dev.c > @@ -1749,7 +1749,7 @@ static int fuse_retrieve(struct fuse_conn *fc, struct inode *inode, > offset = outarg->offset & ~PAGE_MASK; > file_size = i_size_read(inode); > > - num = outarg->size; > + num = min(outarg->size, fc->max_write); > if (outarg->offset > file_size) > num = 0; > else if (outarg->offset + num > file_size) > -- > 2.21.0.352.gf09ad66450
