Messages in this thread | | | Date | Thu, 16 Jan 2020 07:03:53 +0100 | From | Martin Schiller <> | Subject | Re: [PATCH v2 1/2] wan/hdlc_x25: make lapb params configurable |
| |
On 2020-01-15 22:43, David Miller wrote: > From: Martin Schiller <ms@dev.tdt.de> > Date: Tue, 14 Jan 2020 15:02:22 +0100 > >> This enables you to configure mode (DTE/DCE), Modulo, Window, T1, T2, >> N2 via >> sethdlc (which needs to be patched as well). >> >> Signed-off-by: Martin Schiller <ms@dev.tdt.de> > > I don't know how wise it is to add new ioctls to this old driver.
As an user of this framework I can tell you that you need to be able to tune this parameters if it's used in an professional environment.
> > Also, none of these ioctls even have COMPAT handling so they will > never work from a 32-bit binary running on a 64-bit kernel for > example.
How often does this constellation occur in reality? I really have no idea. Our software is either 32-bit or 64-bit, not a mix.
> > Also: > >> +static struct x25_state* state(hdlc_device *hdlc) > > It is always "type *func" never "type* func"
Ok. I will change that. I've copied that from hdlc_fr.c to keep the same coding style. But you are right: "Don't look back, always look forward." :) > >> static int x25_open(struct net_device *dev) >> { >> int result; >> + hdlc_device *hdlc = dev_to_hdlc(dev); >> + struct lapb_parms_struct params; >> static const struct lapb_register_struct cb = { > > Please make this reverse christmas tree ordered.
OK, will do.
> >> @@ -186,6 +217,9 @@ static struct hdlc_proto proto = { >> >> static int x25_ioctl(struct net_device *dev, struct ifreq *ifr) >> { >> + x25_hdlc_proto __user *x25_s = ifr->ifr_settings.ifs_ifsu.x25; >> + const size_t size = sizeof(x25_hdlc_proto); >> + x25_hdlc_proto new_settings; >> hdlc_device *hdlc = dev_to_hdlc(dev); >> int result; > > Likewise.
ditto.
| |