>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/clk/rockchip/clk-pll.c?id=1c0cc5f1ae5ee5a6913704c0d75a6e99604ee30a#n913 >> https://elixir.bootlin.com/linux/v5.4-rc2/source/drivers/clk/rockchip/clk-pll.c#L913 >> >> * Do you find the usage of the format string “%s: could not allocate >> rate table for %s\n” still appropriate at this place? > > If there is an internal "no-memory" output from inside kmemdup now, > I guess the one in the clock driver would be a duplicate and could go away. How do you think about to recheck information sources around the Linux allocation failure report? https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/coding-style.rst?id=da94001239cceb93c132a31928d6ddc4214862d5#n878 >> * Is there a need to adjust the error handling here? > > There is no need for additional error handling. If you would like to omit the macro call “WARN”, I would expect also to express a corresponding null pointer check. > Like if the rate-table could not be duplicated, > the clock will still report the correct clockrate > you can just not set a new rate. How much will a different system configuration matter finally? (Do you really want to treat this setting as “optional”?) > And for a system it's always better to have the clock driver present > than for all device-drivers to fail probing. Especially as this start as > core clock driver, so there is no deferring possible. I imagine that such a view can be clarified further. Regards, Markus
