On Sun, 2020-07-05 at 00:50 +0200, Michal Kubecek wrote: > On Mon, Jun 08, 2020 at 10:52:54AM -0700, Govindarajulu Varadarajan wrote: > > Add support for ETHTOOL_GTUNABLE and ETHTOOL_STUNABLE options. > > > > Tested rx-copybreak on enic driver. Tested ETHTOOL_TUNNABLE_STRING > TUNABLE > > options with test/debug changes in kernel. > > This makes me wonder how are string tunables supposed to work. > Unfortunately there is neither documentation nor code one could look at. > I tried to understand it from this patch but it didn't help much either: > do_stunable() will pass a string of arbitrary size to kernel but > do_gtunable() allocates a buffer of fixed size (for a given tunable). > Is this supposed to be the maximum value length? Or is kernel going to > return an error if the buffer is insufficient and userspace repeats the > request?
I do not know. string tunable isn't implemented/supported in kernel. I assumed driver's get/set_tunable() returns error in case insufficient size. Addressing the rest of the comments and sending v2. -- Govind
