Re: [fuse-devel] [PATCH v2] fuse: Fix IOC_[GS]ET{FLAGS, VERSION} argument size brokenness.

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Mon, Jan 06, 2014 at 06:50:20PM +0100, Miklos Szeredi wrote:
> On Fri, Dec 20, 2013 at 03:35:34PM -0800, Darrick J. Wong wrote:
> > The IOC_[GS]ETFLAGS and IOC_[GS]ETVERSION ioctls, despite being
> > defined to take a "long" parameter, actually take "int" parameters.
> > FUSE unfortunately assumed that the ioctl definitions never lie, and
> 
> FUSE doesn't really assume anything.  It just says, that ioctl's that *properly*
> use the _IO* macros will work.  Everything else will not.
> 
> It's unfortunate that IOC_SET/GETFLAGS is something that may actually make sense
> on a fuse filesystem...
> 
> Two ways out of this:
> 
>  A) Add quirks to fuse
>  B) Add new, fixed ioctls and use those from chattr if available
> 
> I'd very much prefer B.

Ok.  I'll work on a new interface, then.  It was suggested that I explore an
interface that isn't limited to bitmasks.  The xattr interface seems like it
could be hooked for that, so I'll play with that for a while.

Something along the lines of:

attr -s system.iflags -V e,i,c,weird_fs_specific_flag /path/to/file ?

(Pretend that attr will actually set system.* attributes, which it won't.)

--D
> 
> Thanks,
> Miklos
> 
> > transfers a long's worth of data in and out of userspace, which causes
> > stack smashing in chattr, and other bugs elsewhere.
> > 
> > So, special-case this in FUSE so that we don't crash userland.
> > 
> > v2: Do the same for the IOC_[GS]ETVERSION ioctls, as Richard Hansen
> >     points out.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> >  fs/fuse/file.c |   16 ++++++++++++++++
> >  1 file changed, 16 insertions(+)
> > 
> > diff --git a/fs/fuse/file.c b/fs/fuse/file.c
> > index 7e70506..f8766ab 100644
> > --- a/fs/fuse/file.c
> > +++ b/fs/fuse/file.c
> > @@ -2385,6 +2385,22 @@ long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
> >  		iov->iov_base = (void __user *)arg;
> >  		iov->iov_len = _IOC_SIZE(cmd);
> >  
> > +		/*
> > +		 * The IOC_[GS]ETFLAGS and IOC_[GS]ETVERSION ioctls take int
> > +		 * parameters even though the ioctl definition specifies long.
> > +		 * Userland has been expecting int for ages (and chattr
> > +		 * segfaults on FUSE filesystems), so special case that here.
> > +		 * The IOC32 variants were declared with int, so they don't
> > +		 * need this correction.
> > +		 */
> > +		switch (cmd) {
> > +		case FS_IOC_GETFLAGS:
> > +		case FS_IOC_SETFLAGS:
> > +		case FS_IOC_GETVERSION:
> > +		case FS_IOC_SETVERSION:
> > +			iov->iov_len = sizeof(int);
> > +		}
> > +
> >  		if (_IOC_DIR(cmd) & _IOC_WRITE) {
> >  			in_iov = iov;
> >  			in_iovs = 1;
> > 
> > ------------------------------------------------------------------------------
> > Rapidly troubleshoot problems before they affect your business. Most IT 
> > organizations don't have a clear picture of how application performance 
> > affects their revenue. With AppDynamics, you get 100% visibility into your 
> > Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics Pro!
> > http://pubads.g.doubleclick.net/gampad/clk?id=84349831&iu=/4140/ostg.clktrk
> > _______________________________________________
> > fuse-devel mailing list
> > fuse-devel@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/fuse-devel