Messages in this thread | | | Date | Sun, 25 Aug 2013 20:38:41 -0700 | Subject | Re: [PATCH v6 2/3] mmc: dw_mmc: Honor requests to set the clock to 0 (turn off clock) | From | Doug Anderson <> |
| |
Jaehoon,
On Sun, Aug 25, 2013 at 6:31 PM, Jaehoon Chung <jh80.chung@samsung.com> wrote: > Hi Doug, > > On 08/24/2013 05:40 AM, Doug Anderson wrote: >> Jaehoon, >> >> On Fri, Aug 23, 2013 at 6:21 AM, Jaehoon Chung <jh80.chung@samsung.com> wrote: >>> Hi Doug, >>> >>> If the clock-gating is enabled, then maybe it's continuously printed the kernel message for Bus_speed. >> >> Can you explain? I don't think dw_mmc has support for clock gating >> right now. ...or are there some patches that I'm not aware of? I >> could believe that if you've got some non-upstream clock gating >> patches that these would need to be modified to handle it... ...but >> unless those are slated to land upstream it seems like I can't take >> them into account, can I? > If i enabled the CONFIG_MMC_CLK_GATE, the i have found the below message whenever some operation is run. > I will test more with your patch.
Ah, sorry! I wasn't aware of that config option. I was thinking of automatic clock gating based on something like the common clock framework. When there are no more users of a gate clock it will get turned off. To have that work dw_mmc would need to release its biu / ciu clocks at some point.
If I had to guess, I'd speculate that perhaps we should just change the printout to a dev_debug(), though I do find that printout incredibly useful. If I had to guess I'd say that the mmc core is switching between a clock of 0 and a full speed clock constantly. If that's true then it means that dw_mmc used to treat that like a no-op. Now it actually gates the clock. If you comment out the printout, do things still work? Does your power consumption go down?
Let me know if you find anything. Otherwise I can try to reproduce this week.
-Doug
| |