Quoting Aisheng Dong (2018-12-20 17:44:28) > > -----Original Message----- > > From: Stephen Boyd [mailto:sboyd@xxxxxxxxxx] > > Quoting Aisheng Dong (2018-12-19 22:26:50) > > > > -----Original Message----- > > > > From: Fabio Estevam [mailto:festevam@xxxxxxxxx] Hi Aisheng, > > > > > > > > On Wed, Dec 19, 2018 at 1:29 PM Aisheng Dong > > <aisheng.dong@xxxxxxx> > > > > wrote: > > > > > > > > > > SCU clock can be used in a similar way by IMX8QXP and IMX8QM SoCs. > > > > > Let's make the name of clock ID generic to allow other SoCs to > > > > > reuse the common part. > > > > > > > > > > This patch only changes the clock id name and file name, so no > > > > > functional change. > > > > > > > > > > Cc: Shawn Guo <shawnguo@xxxxxxxxxx> > > > > > Cc: Sascha Hauer <kernel@xxxxxxxxxxxxxx> > > > > > Cc: Fabio Estevam <fabio.estevam@xxxxxxx> > > > > > Cc: Stephen Boyd <sboyd@xxxxxxxxxx> > > > > > Cc: Michael Turquette <mturquette@xxxxxxxxxxxx> > > > > > Signed-off-by: Dong Aisheng <aisheng.dong@xxxxxxx> > > > > > --- > > > > > drivers/clk/imx/clk-imx8qxp-lpcg.c | 146 +++++++-------- > > > > > drivers/clk/imx/clk-imx8qxp.c | 152 ++++++++-------- > > > > > include/dt-bindings/clock/imx8-clock.h | 289 > > > > ++++++++++++++++++++++++++++++ > > > > > include/dt-bindings/clock/imx8qxp-clock.h | 289 > > > > > ------------------------------ > > > > > 4 files changed, 438 insertions(+), 438 deletions(-) > > > > > > > > This seems like unneeded churn. Just keep the existing naming and > > > > other SoCs can reuse it without problem. > > > > > > There's a reason. > > > We still need add some clock IDs which exist on MX8QM only (probably > > > more on future SoCs) Having them prefixed with 'IMX8QXP' may look > > strange. > > > > > > Besides that, as those clock IDs will also be used in devicetree, make > > > the name generic for different SoCs seems still better and less confusing. > > > > > > > This looks like an integration problem. Is anyone using the defines that are > > merged through clk tree now? A grep of linux-next says no, but how can I be > > sure? > > I'm sure no one is using it now. > The only user DTS (imx8qxp mek) is waiting for merge after 4.21-rc1 is out. > So it's a chance to update it. > > > > > So can you leave around the header file with the deprecated define names and > > introduce another one that's a copy of the header file but with the non-SoC > > specific names? Then you can wait for a release cycle and delete the old > > header file that nobody should be using in DT anymore. > > Unless you plan to make sure that nobody uses the deprecated defines > > somehow? > > > > Yes, the plan is to make sure nobody uses that. > Otherwise, it may be hard to send DTS patches using new CLK ID name through > different tree later. > I've already tagged and prepared to send out clk tree for the next merge window. It's getting difficult to change this before -rc1 too because I'm traveling for the holidays in the US. I'm fine with changing the world if it doesn't break anything, but I'm not interested in making the pull request wait another few days for this particular topic, so I'd rather send this as part of a follow-on fixes PR during the merge window next week. Can you resend with just the things that need to be changed for the next release (v4.21 or v5.0)? And Cc the relevant people/lists and try to collect acks and reviewed-by tags before next Wednesday? I can send a fixes PR on Friday next week then.
