On 04/15/2014 06:01 AM, Dan Carpenter wrote: > On Mon, Apr 14, 2014 at 03:17:19PM -0400, Mark Hounschell wrote: >> On 04/14/2014 11:49 AM, Greg Kroah-Hartman wrote: >>> On Tue, Mar 25, 2014 at 04:38:14PM -0400, Mark Hounschell wrote: >>>> The config file is currently read for each board found. >>>> It only needs to be read one time. The buffer it is read >>>> into can now be freed immediately after it is parsed >>>> instead of at driver unload time. >>>> >>>> Signed-off-by: Mark Hounschell <markh@xxxxxxxxxx> >>>> --- >>>> drivers/staging/dgap/dgap.c | 18 +++++++++--------- >>>> 1 file changed, 9 insertions(+), 9 deletions(-) >>> >>> Kernel drivers shouldn't be reading config files in the first place :) >>> >>> But I'll take this for now, as it does fix the code up. >>> >> >> Ya, I know. I'm just not sure how to work around that yet. >> > > You're probably going to have different solutions for different bits. > For example, can we auto detect the board type? If so then remove that > chunk from the config file. We'll have to have a struct somewhere with > the number of ports each type has. Some things may have to be module > init options and somethings maybe we can configure with the tty ioctls > and some things through sysfs. > > Creating your own driver specific api is bad so always prefer shared > infrastructure over, for example, doing your own thing in sysfs. > Understood. This driver supports many boards that I do not have access to. Some of the other boards support external port expanders that allow external control of the port configuration. These boards also require different "concentrator" firmware depending on what external port configuration you have connected. In any case, I'm not quite ready to dive into all that just yet. There are still lots of little things to do. Probably a lot more than I know of. Regards Mark _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
