On 01/02, Bryan O'Donoghue wrote: > On 02/01/18 19:01, Stephen Boyd wrote: > >On 12/31, Bryan O'Donoghue wrote: > >>On 30/12/17 16:36, Mikko Perttunen wrote: > >>>FWIW, we had this problem some years ago with the Tegra CPU clock > >>>- then it was determined that a simpler solution was to have the > >>>determine_rate callback support unsigned long rates - so clock > >>>drivers that need to return rates higher than 2^31 can instead > >>>implement the determine_rate callback. That is what's currently > >>>implemented. > >>> > >>>Mikko > >> > >>Granted we could work around it but, having both zero and less than > >>zero indicate error means you can't support larger than LONG_MAX > >>which is I think worth fixing. > >> > > > >Ok. But can you implement the determine_rate op instead of the > >round_rate op for your clk? > > Don't know . Please try. > > >It's not a work-around, it's the > >preferred solution. That would allow rates larger than 2^31 for > >the clk without pushing through a change to all the drivers to > >express zero as "error" and non-zero as the rounded rate. > > > >I'm not entirely opposed to this approach, because we probably > >don't care to pass the particular error value from a clk provider > >to a clk consumer about what the error is. > > Which was my thought. The return value of clk_ops->round_rate() > appears not to get pushed up the stack, which is what the last patch > in this series deals with. > > [PATCH 33/33] clk: change handling of round_rate() such that only > zero is an error Hmm? clk_core_determine_round_nolock() returns 'rate' if rate < 0 from the round_rate op. clk_core_round_rate_nolock() returns that value to clk_round_rate() which returns it to the consumer. > > >It's actually what we > >proposed as the solution for clk_round_rate() to return values > >larger than LONG_MAX to consumers. But doing that consumer API > >change or this provider side change is going to require us to > >evaluate all the consumers of these clks to make sure they don't > >check for some error value that's less than zero. This series > >does half the work, > > Do you mean users of clk_rounda_rate() ? I have a set of patches for > that but wanted to separate that from clk_ops->round_rate() so as > not to send ~70 patches out to LKML at once - even if they are in > two blocks. Ok. What have you done to the consumers of clk_round_rate()? Made them treat 0 as an error instead of less than zero? The documentation in clk.h needs to be updated. See this patch from Paul Wamsley[1] for one proposed patch that went nowhere. Also include Russell King please. It was also proposed to change the function signature of clk_round_rate() to return unsigned long, but that didn't go anywhere either. > > If so, I can publish that set too for reference. > > AFAICT on clk_ops->round_rate the last patch #33 ought to cover the > usage of the return value of clk_ops->round_rate(). > > Have I missed something ? Hopefully not! > > >by changing the provider side, while ignoring > >the consumer side and any potential fallout of the less than zero > >to zero return value change. > > > > Can you look at #33 ? I'm not sure if you saw that one. > Yeah I looked at it. From what I can tell it makes clk_round_rate() return 0 now instead of whatever negative value the clk_ops::round_rate function returns. [1] https://lkml.kernel.org/r/alpine.DEB.2.02.1311251603310.23090@tamien -- Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project