On Tue, May 15, 2018 at 10:53:51AM +0100, Mark Rutland wrote: > On Mon, May 14, 2018 at 10:34:14PM +0200, Dominik Brodowski wrote: > > On Mon, May 14, 2018 at 06:14:28PM +0100, Mark Rutland wrote: > > > On Mon, May 14, 2018 at 10:46:35AM +0100, Mark Rutland wrote: > > > > +#ifdef CONFIG_COMPAT > > > > +int kcompat_sys_statfs64(const char __user * pathname, compat_size_t sz, > > > > + struct compat_statfs64 __user * buf); > > > > +int kcompat_sys_fstatfs64(unsigned int fd, compat_size_t sz, > > > > + struct compat_statfs64 __user * buf); > > > > +#endif > > > > > > I've moved these to <linux/compat.h>, so that they live with the rest of > > > the compat syscall stuff. That should avoid build failures the kbuild > > > test robot picked up where compat_size_t wasn't dfined. > > > > Please add a comment there, similar to what is in syscalls.h: > > > > /* > > * Kernel code should not call syscalls (i.e., sys_xyzyyz()) directly. > > * Instead, use one of the functions which work equivalently, such as > > * the ksys_xyzyyz() functions prototyped below. > > */ > > To make the kcompat_sys_* naming scheme clearer, I've added compat references to > the above, i.e. > > /* > * Kernel code should not call compat syscalls (i.e., compat_sys_xyzyyz()) > * directly. Instead, use one of the functions which work equivalently, such > * as the kcompat_sys_xyzyyz() functions prototyped below. > */ That's what I meant ;) > > Once you have done so, feel free to add my > > > > Reviewed-by: Dominik Brodowski <linux@xxxxxxxxxxxxxxxxxxxx> > > I hope that this still stands with the changes above? It does. Thanks! Dominik
