Re: [PATCH 5/9] fs: remove various compat readv/writev helpers

[Al Viro]

On Wed, Sep 23, 2020 at 08:45:51PM +0200, Arnd Bergmann wrote:
> On Wed, Sep 23, 2020 at 6:38 PM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> >
> > I wonder if we should do something like
> >
> > SYSCALL_DECLARE3(readv, unsigned long, fd, const struct iovec __user *, vec,
> >                  unsigned long, vlen);
> > in syscalls.h instead, and not under that ifdef.
> >
> > Let it expand to declaration of sys_...() in generic case and, on x86, into
> > __do_sys_...() and __ia32_sys_...()/__x64_sys_...(), with types matching
> > what SYSCALL_DEFINE ends up using.
> >
> > Similar macro would cover compat_sys_...() declarations.  That would
> > restore mismatch checking for x86 and friends.  AFAICS, the cost wouldn't
> > be terribly high - cpp would have more to chew through in syscalls.h,
> > but it shouldn't be all that costly.  Famous last words, of course...
> >
> > Does anybody see fundamental problems with that?
> 
> I've had some ideas along those lines in the past and I think it should work.
> 
> As a variation of this, the SYSCALL_DEFINEx() macros could go away
> entirely, leaving only the macro instantiations from the header to
> require that syntax. It would require first changing the remaining
> architectures to build the syscall table from C code instead of
> assembler though.
> 
> Regardless of that, another advantage of having the SYSCALL_DECLAREx()
> would be the ability to include that header file from elsewhere with a different
> macro definition to create a machine-readable version of the interface when
> combined with the syscall.tbl files. This could be used to create a user
> space stub for calling into the low-level syscall regardless of the
> libc interfaces,
> or for synchronizing the interfaces with strace, qemu-user, or anything that
> needs to deal with the low-level interface.

FWIW, after playing with that for a while...  Do we really want the
compat_sys_...() declarations to live in linux/compat.h?  Most of
the users of that file don't want those; why not move them to
linux/syscalls.h?

Reason: there's a lot more users of linux/compat.h than those of
linux/syscalls.h - it's pulled by everything in the networking stack,
for starters...