clk: invoke clk_disable_unused() after resume
PATCH 1/1 clk: invoke clk_disable_unused() after resume
In a discussion[1], mturquette made the suggestion that
"It might be a good idea to call clk_disable_unused()
from the clk core after resuming from suspend."
This commit does exactly that. For Broadcom STB chips, clocks may come up after resume in an "on" state, so calling clk_disable_unused() again will turn off clocks that should be off.
[1] https://lkml.org/lkml/2014/11/17/666
--- drivers/clk/clk.c | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-)
drivers/clk/clk.c +13 -3
@@ -21,6 +21,7 @@#include <linux/device.h> #include <linux/init.h> #include <linux/sched.h> +#include <linux/syscore_ops.h> #include "clk.h"
@@ -500,13 +501,13 @@ static int __init clk_ignore_unused_setup(char *__unused) }__setup ("clk_ignore_unused" , clk_ignore_unused_setup); -static int clk_disable_unused (void ) +static void clk_disable_unused (void ) {struct clk *clk;if (clk_ignore_unused) {pr_warn ("clk: Not disabling unused clocks\n" ); -return 0 ; +return ; }clk_prepare_lock ();
@@ -524,10 +525,19 @@ static int clk_disable_unused(void)clk_unprepare_unused_subtree (clk);clk_prepare_unlock (); +} +static struct syscore_ops disable_unused_ops = { + .resume = clk_disable_unused, +}; + +static int clk_init_disable_unused (void ) +{ +clk_disable_unused (); +register_syscore_ops (&disable_unused_ops);return 0 ; } -late_initcall_sync (clk_disable_unused); +late_initcall_sync (clk_init_disable_unused);/*** helper functions ***/
Signature · 1 line
1.9.0.138.g2de3478
PATCH 1/1 clk: invoke clk_disable_unused() after resume
Quoted 5 lines · Jim Quinlan
In a discussion[1], mturquette made the suggestion that
"It might be a good idea to call
clk_disable_unused()
from the clk core after resuming from suspend."
Quoted 9 lines
This commit does exactly that. For Broadcom STB chips, clocks may come up after resume in an "on" state, so calling
clk_disable_unused()again will turn off clocks that should be off.[1] https://lkml.org/lkml/2014/11/17/666
Signed-off-by Jim Quinlan <[email protected]>
It would be better to use the driver core or syscore suspend/resume hooks to restore the state of the clocks as they were when the system was suspended. Plus it would make more sense to do this on a case-by-case basis instead of assuming that all platforms have clocks that are left on out of resume. It could still be generic by exporting the clk_disable_unused() function to drivers so they can call it if they need to but I don't see why we want to force this on every platform for every suspend/resume cycle.
Signature · 2 lines
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project