From mboxrd@z Thu Jan 1 00:00:00 1970 From: Simon Glass Date: Mon, 1 Apr 2019 10:18:47 -0600 Subject: [U-Boot] [PATCH] dm: Add a No-op uclass In-Reply-To: <94e5e0c9-1e8c-8074-45d2-ee6ed5c4b793@ti.com> References: <20190322164429.28637-1-jjhiblot@ti.com> <94e5e0c9-1e8c-8074-45d2-ee6ed5c4b793@ti.com> Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de Hi Jean-Jacques, On Mon, 1 Apr 2019 at 05:32, Jean-Jacques Hiblot wrote: > > Hi Simon, > > On 30/03/2019 22:18, Simon Glass wrote: > > Hi Jean-Jacques, > > > > On Fri, 22 Mar 2019 at 10:44, Jean-Jacques Hiblot wrote: > >> This uclass is intended for devices that do not need any features from the > >> uclass, including binding children. > >> This will typically be used by devices that are used to bind child devices > >> but do not use dm_scan_fdt_dev() to do it. > >> > > Can you expand this motivation a little? I am not sure why calling > > dm_scan_fdt_dev() would be problematic. > > In the case of the USB wrappers, there are 2 children nodes: one for > peripheral and one for host. The wrapper binds only one of them. > dm_scan_fdt_dev() would bind both. OK, please add this to the commit message and your uclass header file (or wherever else you can add docs for this uclass). > > > > > Also if you do add a new uclass it should have a sandbox driver and test. > > There isn't much to test, except to check that the uclass is present. > The UCLASS itself does not provide any feature. Well, one feature is that it does not scan its children. So you could add a child node in test.dts and test that it does not get bound. Regards, Simon