[PATCH] clk: qcom: Leave mmss noc on for 8998

[PATCH] clk: qcom: Leave mmss noc on for 8998

[Jeffrey Hugo]

Similar to other qcom targets, gcc_mmss_noc_cfg_ahb_clk should not be
disabled.  Any mmss access depends on this clock, and its been observed
that enabling mmssnoc_axi_rpm_clk with rpmcc results in an implicit
access to mmss and will crash the system if gcc_mmss_noc_cfg_ahb_clk is
disabled.

Signed-off-by: Jeffrey Hugo <jhugo@codeaurora.org>
---
 drivers/clk/qcom/gcc-msm8998.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/clk/qcom/gcc-msm8998.c b/drivers/clk/qcom/gcc-msm8998.c
index 4b75a09..b371308 100644
--- a/drivers/clk/qcom/gcc-msm8998.c
+++ b/drivers/clk/qcom/gcc-msm8998.c
@@ -2014,6 +2014,7 @@ enum {
 		.hw.init = &(struct clk_init_data){
 			.name = "gcc_mmss_noc_cfg_ahb_clk",
 			.ops = &clk_branch2_ops,
+			.flags = CLK_IGNORE_UNUSED,
 		},
 	},
 };
-- 
Qualcomm Datacenter Technologies as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

Re: [PATCH] clk: qcom: Leave mmss noc on for 8998

[Stephen Boyd]

Quoting Jeffrey Hugo (2018-12-13 08:59:30)
> Similar to other qcom targets, gcc_mmss_noc_cfg_ahb_clk should not be
> disabled.  Any mmss access depends on this clock, and its been observed
> that enabling mmssnoc_axi_rpm_clk with rpmcc results in an implicit
> access to mmss and will crash the system if gcc_mmss_noc_cfg_ahb_clk is
> disabled.
> 
> Signed-off-by: Jeffrey Hugo <jhugo@codeaurora.org>
> ---
>  drivers/clk/qcom/gcc-msm8998.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/clk/qcom/gcc-msm8998.c b/drivers/clk/qcom/gcc-msm8998.c
> index 4b75a09..b371308 100644
> --- a/drivers/clk/qcom/gcc-msm8998.c
> +++ b/drivers/clk/qcom/gcc-msm8998.c
> @@ -2014,6 +2014,7 @@ enum {
>                 .hw.init = &(struct clk_init_data){
>                         .name = "gcc_mmss_noc_cfg_ahb_clk",
>                         .ops = &clk_branch2_ops,
> +                       .flags = CLK_IGNORE_UNUSED,

It should be marked as CLK_IS_CRITICAL and a comment should be added to
the code indicating why it's critical.

Re: [PATCH] clk: qcom: Leave mmss noc on for 8998

[Stephen Boyd]

Quoting Jeffrey Hugo (2018-12-13 08:59:30)
> Similar to other qcom targets, gcc_mmss_noc_cfg_ahb_clk should not be
> disabled.  Any mmss access depends on this clock, and its been observed
> that enabling mmssnoc_axi_rpm_clk with rpmcc results in an implicit
> access to mmss and will crash the system if gcc_mmss_noc_cfg_ahb_clk is
> disabled.
> 
> Signed-off-by: Jeffrey Hugo <jhugo@codeaurora.org>
> ---
>  drivers/clk/qcom/gcc-msm8998.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/clk/qcom/gcc-msm8998.c b/drivers/clk/qcom/gcc-msm8998.c
> index 4b75a09..b371308 100644
> --- a/drivers/clk/qcom/gcc-msm8998.c
> +++ b/drivers/clk/qcom/gcc-msm8998.c
> @@ -2014,6 +2014,7 @@ enum {
>                 .hw.init = &(struct clk_init_data){
>                         .name = "gcc_mmss_noc_cfg_ahb_clk",
>                         .ops = &clk_branch2_ops,
> +                       .flags = CLK_IGNORE_UNUSED,

It should be marked as CLK_IS_CRITICAL and a comment should be added to
the code indicating why it's critical.