Re: 2.6.2-rc2-mm2

[Andrew Morton]

Tim Hockin <thockin@xxxxxxx> wrote:
>
> On Fri, Jan 30, 2004 at 11:47:01AM -0800, Andrew Morton wrote:
> > > directly calling sys_ANYTHING sounds really wrong to me...
> 
> It sounded wrong to me, but it gets done ALL OVER.
> 
> > Tim, I do think it would be neater to add another entry point in sys.c for
> > nfsd and just do a memcpy.
> 
> Do you prefer:
> 
> a) make a function
> 	sys.c: ksetgroups(int gidsetsize, gid_t *grouplist)
>    which does the same as sys_setgroups, but without the copy_from_user()
>    stuff?  The only user (for now, maybe ever) is nfsd.
> 
> b) make a function
> 	sys.c: nfsd_setgroups(int gidsetsize, gid_t *grouplist)
>    which does the same as sys_setgroups, but without the copy_from_user()
> 
> c) make the nfsd code build a struct group_info and call
>    set_current_groups()
> 

Can we do d)?

static long do_setgroups(int gidsetsize, gid_t __user *user_grouplist,
			gid_t *kern_grouplist)
{
	gid_t groups[NGROUPS];
	int retval;

	if (!capable(CAP_SETGID))
		return -EPERM;
	if ((unsigned) gidsetsize > NGROUPS)
		return -EINVAL;
	if (user_grouplist) {
		if (copy_from_user(groups, user_grouplist,
				gidsetsize * sizeof(gid_t)))
			return -EFAULT;
	} else {
		memcpy(groups, kern_grouplist, gidsetsize * sizeof(gid_t));
	}
	retval = security_task_setgroups(gidsetsize, groups);
	if (retval)
		return retval;
	memcpy(current->groups, groups, gidsetsize * sizeof(gid_t));
	current->ngroups = gidsetsize;
	return 0;
}

asmlinkage long sys_setgroups(int gidsetsize, gid_t __user *grouplist)
{
	return do_setgroups(gidsetsize, grouplist, NULL);
}

long kern_setgroups(int gidsetsize, gid_t *grouplist)
{
	return do_setgroups(gidsetsize, NULL, grouplist);
}

It's a bit grubby, but the grubbiness is localised.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/