From mboxrd@z Thu Jan 1 00:00:00 1970 From: Petr Vorel Date: Wed, 17 Mar 2021 12:47:21 +0100 Subject: [LTP] [PATCH 1/2] pec: Convert to the new API In-Reply-To: <87117f6f-58e7-8b68-38c3-be5b080e86b6@jv-coder.de> References: <20210315092844.991073-1-lkml@jv-coder.de> <87117f6f-58e7-8b68-38c3-be5b080e86b6@jv-coder.de> Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ltp@lists.linux.it Hi J?rg, > Hi Petr, > On 3/17/2021 10:34 AM, Petr Vorel wrote: > > BTW: would it make sense to rename runtest/connectors to runtest/netlink > > and move code from testcases/kernel/connectors/ to testcases/kernel/netlink/? > Maybe kernel/netlink/connector/. Connector is an abstraction on top of +1 > netlink with a slightly > different interface than pure netlink (eg. cn_msg instead of nlmsghdr). > There are only very few other modules, that use this interface at the moment > (MS HyperV, MD,? uvesafb? and dallas' 1-wire). I'd still put it into netlink and try to put there more general netlink subsystem tests. My objection is that runtest/connectors has only single test and connectors is very generic name. > The question is: Is the interface the correct think to categorize by? If the > interface (i.e. netlink) is used for categorization, then some of the crypto > tests should also be in this netlink category. Thus maybe add some of relevant tests also into runtest/netlink? We shave some kind of duplicity (see runtest/cve, it contains some tests which are also in runtest/syscalls). That would also justify creating runtest/netlink (because having just single test for legacy connector interface in runtest/netlink does not sound good to me). Kind regards, Petr > J?rg