Re: [PATCH 2/2] dlmfs: convert dlmfs_file_read() to copy_to_user()

[Al Viro]

On Thu, May 28, 2020 at 08:42:25PM -0700, Linus Torvalds wrote:

> >         struct sigset_argpack argpack = { NULL, 0 };
> >
> >         if (get_sigset_argpack(sig, &argpack))
> >                 return -EFAULT;
> 
> and now you can use "argpack.sigset" and "argpack.sigset_size".
> 
> No?
> 
> Same exact deal for the compat case, where you'd just need that compat
> struct (using "compat_uptr_t" and "compat_size_t"), and then
> 
> >         struct compat_sigset_argpack argpack = { 0, 0 };
> >
> > +       if (get_compat_sigset_argpack(sig, &argpack))
> > +               return -EFAULT;
> 
> and then you use the result with "compat_ptr(argpack.sigset)" and
> "argpack.sigset_size".
> 
> Or did I mis-read anything and get confused by that code in your patch?

Umm...  I'd been concerned about code generation, but it actually gets
split into a pair of scalars just fine...

Al, trying to resist the temptation to call those struct bad_idea and
struct bad_idea_32...

All jokes aside, when had we (or anybody else, really) _not_ gotten
into trouble when passing structs across the kernel boundary?  Sure,
sometimes you have to (stat, for example), but just look at the amount
of PITA stat() has spawned...