Re: [PATCH v4 1/6] fs: add write-stream management ioctls

On Wed, Aug 19, 2026 at 09:04:59AM +0200, Christoph Hellwig wrote:
> On Thu, Jul 30, 2026 at 10:22:11AM -0700, Darrick J. Wong wrote:
> > "The third argument is an untyped pointer to memory.  It's traditionally
> > char  *argp (from the days before void * was valid C), and will be so
> > named for this discussion."
> 
> That documentation is wrong or at least very incomplete.  A pointer
> is the typical use, but there are plenty of ioctls passing scalar
> values <= sizeof(long), going all the way back to early UNIX days.
> 
> > You can also see this reflected in the discussion of the R/W/WR variants
> > of _IO:
> > https://docs.kernel.org/driver-api/ioctl.html
> 
> _IO* are optional helpers, and plenty especially older ioctls are
> defined without using them.

Well, yeah, we don't force anyone to use the _IOR/_IOW/_IOWR helpers.
But since they were used here, I thought it worth pointing out.

> > Granted your kernel code can treat that argp as an integer and not a
> > __user pointer if it wants to, but that goes against most ioctl
> > implementations.  Further, subsystems that marshal ioctl information for
> > passing through to another layer (e.g. fuse) assume that the third
> > argument is a pointer and that it can copy sizeof(data_type) bytes at
> > that address in and out of the kernel as part of marshalling.
> 
> That is not a good assumption and things will break.

I know; fuse is a total mess w.r.t. ioctl handling because it assumes
that it can just go ahead and do the userspace accesses.

--D