Hm, what a mess. I tried to properly mark up the __user pointers,
but for some of these structures, we use them both in the kernel,
and across the user/kernel boundry, which isn't ok. So we end
up generating a few new sparse warnings in places we were not before,
but the large majority of things are now properly tagged in the fops
file.
The whole ioctl interface needs to be carefully looked at in the future.
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Frank Mori Hess <fmhess@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The whole file should be converted to use seqfile, if it's even
still needed. Or move to debugfs.
Anyway, I fixed up the minor issues here.
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Frank Mori Hess <fmhess@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
No one outside of the comedi core calls this function, so create
an internal.h file to put the prototype in, and don't export
it to the world.
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Frank Mori Hess <fmhess@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
If we really are passing in a struct comedi_device, then say we are,
don't mess around with void pointers for no reason.
This also fixes up the comedi_bond.c driver, which is the only
user of the kcomedilib code.
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Frank Mori Hess <fmhess@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Now that we know we are only making 2 different types of instructions,
only handle those two types.
Also make the call a bit more typesafe by passing the correct pointer
type.
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Frank Mori Hess <fmhess@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
No one else calls this function, so mark it static.
Now we can strip out the unneeded functionality in here as well.
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Frank Mori Hess <fmhess@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Merge these two files into kcomedilib_main.c as they are tiny.
This will also let us get rid of another global symbol in the future.
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Frank Mori Hess <fmhess@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
On Mon, 2010-05-03 at 10:09 -0700, Randy Dunlap wrote:
> Lots of cxt1e1 source code uses THIS_MODULE->name, which won't build
> when CONFIG_MODULES is not enabled, so use KBUILD_MODNAME instead.
Perhaps a conversion to pr_<level> is better?
Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch fixes some long line lengths in drivers.c that checkpatch.pl was
complaining about
Signed-off-by: Mark Rankilor <reodge@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is a patch to range.c that rearranges some EXPORT_SYMBOL() macros to please
checkpatch.pl
Signed-off-by: Mark Rankilor <reodge@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch solves a compilation error in today linux-next tree.
The adis16255 staging driver Makefile seems to be wrong. I think this patch solves the issue.
Signed-off-by: Javier Martinez Canillas <martinez.javier@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Greg Kroah-Hartman
Joe Perches
Iain Churcher
Mark Rankilor
Charles Clément
Javier Martinez Canillas
Lars Lindley
Andres More