At Tue, 10 Dec 2013 11:46:43 -0800, Paul Walmsley wrote:
On Mon, 9 Dec 2013, Takashi Iwai wrote:
At Mon, 9 Dec 2013 18:40:48 -0800, Paul Walmsley wrote:
Treat both negative and zero return values from clk_round_rate() as errors. This is needed since subsequent patches will convert clk_round_rate()'s return value to be an unsigned type, rather than a signed type, since some clock sources can generate rates higher than (2^31)-1 Hz.
Is the behavior "returning zero upon error" already in 3.13? That is, should this (and another) patch be taken as a 3.13-fix patch, or it's for 3.14?
It depends on the platform. The Common Clock Framework code returns 0 upon error right now. But other clock framework implementations, such as the one used by the Atmel AT91 boards, return negative error codes. And looking at the mainline code, it looks like the at73c213 chip is most likely to be used on AT91 boards.
So if you want to send this patch for v3.13-rc, it's probably justified, but it's low-priority. v3.14 is also fine.
I'll be sending some followup patches to the platform maintainers to change the clock framework code to return 0 upon error. But those can't be applied until the drivers are fixed, if we want to avoid regressions in error path handling. So from that point of view, applying these driver patches in v3.13-rc would mean there is less delay to getting the platform clock framework fixes upstream :-)
OK, unless any known regressions are reported, I'm inclined to put this for 3.14. Now applied to for-next branch. Thanks!
Takashi