[PATH] cgroup: add cgroup_favordynmods= command-line option

[PATH] cgroup: add cgroup_favordynmods= command-line option

[Luiz Capitulino]

We have a need of using favordynmods with cgroup v1, which doesn't support
changing mount flags during remount. Enabling CONFIG_FAVOR_DYNMODS at
build-time is not an option because we want to be able to selectively
enable it for certain systems.

This commit addresses this by introducing the cgroup_favordynmods=
command-line option. This option works for both cgroup v1 and v2 and
also allows for disabling favorynmods when the kernel built with
CONFIG_FAVOR_DYNMODS=y.

Signed-off-by: Luiz Capitulino <luizcap-vV1OtcyAfmbQT0dZR+AlfA@public.gmane.org>
---
 Documentation/admin-guide/kernel-parameters.txt |  4 ++++
 kernel/cgroup/cgroup.c                          | 14 +++++++++++---
 2 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 0c38a8af95ce..672f76a3c002 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -580,6 +580,10 @@
 			named mounts. Specifying both "all" and "named" disables
 			all v1 hierarchies.
 
+	cgroup_favordynmods= [KNL] Enable or Disable favordynmods.
+			Format: { "true" | "false" }
+			Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS.
+
 	cgroup.memory=	[KNL] Pass options to the cgroup memory controller.
 			Format: <string>
 			nosocket -- Disable socket memory accounting.
diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index 5fa95f86cb4d..b625825e270b 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -207,6 +207,8 @@ static u16 have_exit_callback __read_mostly;
 static u16 have_release_callback __read_mostly;
 static u16 have_canfork_callback __read_mostly;
 
+static bool have_favordynmods __read_mostly = IS_ENABLED(CONFIG_CGROUP_FAVOR_DYNMODS);
+
 /* cgroup namespace for init task */
 struct cgroup_namespace init_cgroup_ns = {
 	.ns.count	= REFCOUNT_INIT(2),
@@ -2265,9 +2267,9 @@ static int cgroup_init_fs_context(struct fs_context *fc)
 	fc->user_ns = get_user_ns(ctx->ns->user_ns);
 	fc->global = true;
 
-#ifdef CONFIG_CGROUP_FAVOR_DYNMODS
-	ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS;
-#endif
+	if (have_favordynmods)
+		ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS;
+
 	return 0;
 }
 
@@ -6767,6 +6769,12 @@ static int __init enable_cgroup_debug(char *str)
 }
 __setup("cgroup_debug", enable_cgroup_debug);
 
+static int __init cgroup_favordynmods_setup(char *str)
+{
+	return (kstrtobool(str, &have_favordynmods) == 0);
+}
+__setup("cgroup_favordynmods=", cgroup_favordynmods_setup);
+
 /**
  * css_tryget_online_from_dir - get corresponding css from a cgroup dentry
  * @dentry: directory dentry of interest
-- 
2.40.1

Re: [PATH] cgroup: add cgroup_favordynmods= command-line option

[Luiz Capitulino]

On Thu, Aug 31, 2023 at 10:20:46AM -0400, Luiz Capitulino wrote:
> We have a need of using favordynmods with cgroup v1, which doesn't support
> changing mount flags during remount. Enabling CONFIG_FAVOR_DYNMODS at
> build-time is not an option because we want to be able to selectively
> enable it for certain systems.
> 
> This commit addresses this by introducing the cgroup_favordynmods=
> command-line option. This option works for both cgroup v1 and v2 and
> also allows for disabling favorynmods when the kernel built with
> CONFIG_FAVOR_DYNMODS=y.
> 
> Signed-off-by: Luiz Capitulino <luizcap-vV1OtcyAfmbQT0dZR+AlfA@public.gmane.org>

Hi,

Would somebody take a look at this patch please?

Thanks,

- Luiz

> ---
>  Documentation/admin-guide/kernel-parameters.txt |  4 ++++
>  kernel/cgroup/cgroup.c                          | 14 +++++++++++---
>  2 files changed, 15 insertions(+), 3 deletions(-)
> 
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index 0c38a8af95ce..672f76a3c002 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -580,6 +580,10 @@
>  			named mounts. Specifying both "all" and "named" disables
>  			all v1 hierarchies.
>  
> +	cgroup_favordynmods= [KNL] Enable or Disable favordynmods.
> +			Format: { "true" | "false" }
> +			Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS.
> +
>  	cgroup.memory=	[KNL] Pass options to the cgroup memory controller.
>  			Format: <string>
>  			nosocket -- Disable socket memory accounting.
> diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
> index 5fa95f86cb4d..b625825e270b 100644
> --- a/kernel/cgroup/cgroup.c
> +++ b/kernel/cgroup/cgroup.c
> @@ -207,6 +207,8 @@ static u16 have_exit_callback __read_mostly;
>  static u16 have_release_callback __read_mostly;
>  static u16 have_canfork_callback __read_mostly;
>  
> +static bool have_favordynmods __read_mostly = IS_ENABLED(CONFIG_CGROUP_FAVOR_DYNMODS);
> +
>  /* cgroup namespace for init task */
>  struct cgroup_namespace init_cgroup_ns = {
>  	.ns.count	= REFCOUNT_INIT(2),
> @@ -2265,9 +2267,9 @@ static int cgroup_init_fs_context(struct fs_context *fc)
>  	fc->user_ns = get_user_ns(ctx->ns->user_ns);
>  	fc->global = true;
>  
> -#ifdef CONFIG_CGROUP_FAVOR_DYNMODS
> -	ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS;
> -#endif
> +	if (have_favordynmods)
> +		ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS;
> +
>  	return 0;
>  }
>  
> @@ -6767,6 +6769,12 @@ static int __init enable_cgroup_debug(char *str)
>  }
>  __setup("cgroup_debug", enable_cgroup_debug);
>  
> +static int __init cgroup_favordynmods_setup(char *str)
> +{
> +	return (kstrtobool(str, &have_favordynmods) == 0);
> +}
> +__setup("cgroup_favordynmods=", cgroup_favordynmods_setup);
> +
>  /**
>   * css_tryget_online_from_dir - get corresponding css from a cgroup dentry
>   * @dentry: directory dentry of interest
> -- 
> 2.40.1
> 

Re: [PATH] cgroup: add cgroup_favordynmods= command-line option

[Waiman Long]

On 9/5/23 13:03, Luiz Capitulino wrote:
> On Thu, Aug 31, 2023 at 10:20:46AM -0400, Luiz Capitulino wrote:
>> We have a need of using favordynmods with cgroup v1, which doesn't support
>> changing mount flags during remount. Enabling CONFIG_FAVOR_DYNMODS at
>> build-time is not an option because we want to be able to selectively
>> enable it for certain systems.
>>
>> This commit addresses this by introducing the cgroup_favordynmods=
>> command-line option. This option works for both cgroup v1 and v2 and
>> also allows for disabling favorynmods when the kernel built with
>> CONFIG_FAVOR_DYNMODS=y.
>>
>> Signed-off-by: Luiz Capitulino <luizcap-vV1OtcyAfmbQT0dZR+AlfA@public.gmane.org>
> Hi,
>
> Would somebody take a look at this patch please?
>
> Thanks,
>
> - Luiz

The patch looks good to me. I do have a minor comment that it may be 
better to use the relatively new __ro_after_init qualifier instead of 
__read_mostly.

Cheers,
Longman

>> ---
>>   Documentation/admin-guide/kernel-parameters.txt |  4 ++++
>>   kernel/cgroup/cgroup.c                          | 14 +++++++++++---
>>   2 files changed, 15 insertions(+), 3 deletions(-)
>>
>> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
>> index 0c38a8af95ce..672f76a3c002 100644
>> --- a/Documentation/admin-guide/kernel-parameters.txt
>> +++ b/Documentation/admin-guide/kernel-parameters.txt
>> @@ -580,6 +580,10 @@
>>   			named mounts. Specifying both "all" and "named" disables
>>   			all v1 hierarchies.
>>   
>> +	cgroup_favordynmods= [KNL] Enable or Disable favordynmods.
>> +			Format: { "true" | "false" }
>> +			Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS.
>> +
>>   	cgroup.memory=	[KNL] Pass options to the cgroup memory controller.
>>   			Format: <string>
>>   			nosocket -- Disable socket memory accounting.
>> diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
>> index 5fa95f86cb4d..b625825e270b 100644
>> --- a/kernel/cgroup/cgroup.c
>> +++ b/kernel/cgroup/cgroup.c
>> @@ -207,6 +207,8 @@ static u16 have_exit_callback __read_mostly;
>>   static u16 have_release_callback __read_mostly;
>>   static u16 have_canfork_callback __read_mostly;
>>   
>> +static bool have_favordynmods __read_mostly = IS_ENABLED(CONFIG_CGROUP_FAVOR_DYNMODS);
>> +
>>   /* cgroup namespace for init task */
>>   struct cgroup_namespace init_cgroup_ns = {
>>   	.ns.count	= REFCOUNT_INIT(2),
>> @@ -2265,9 +2267,9 @@ static int cgroup_init_fs_context(struct fs_context *fc)
>>   	fc->user_ns = get_user_ns(ctx->ns->user_ns);
>>   	fc->global = true;
>>   
>> -#ifdef CONFIG_CGROUP_FAVOR_DYNMODS
>> -	ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS;
>> -#endif
>> +	if (have_favordynmods)
>> +		ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS;
>> +
>>   	return 0;
>>   }
>>   
>> @@ -6767,6 +6769,12 @@ static int __init enable_cgroup_debug(char *str)
>>   }
>>   __setup("cgroup_debug", enable_cgroup_debug);
>>   
>> +static int __init cgroup_favordynmods_setup(char *str)
>> +{
>> +	return (kstrtobool(str, &have_favordynmods) == 0);
>> +}
>> +__setup("cgroup_favordynmods=", cgroup_favordynmods_setup);
>> +
>>   /**
>>    * css_tryget_online_from_dir - get corresponding css from a cgroup dentry
>>    * @dentry: directory dentry of interest
>> -- 
>> 2.40.1
>>

Re: [PATH] cgroup: add cgroup_favordynmods= command-line option

[Waiman Long]

On 9/5/23 14:59, Waiman Long wrote:
>
> On 9/5/23 13:03, Luiz Capitulino wrote:
>> On Thu, Aug 31, 2023 at 10:20:46AM -0400, Luiz Capitulino wrote:
>>> We have a need of using favordynmods with cgroup v1, which doesn't 
>>> support
>>> changing mount flags during remount. Enabling CONFIG_FAVOR_DYNMODS at
>>> build-time is not an option because we want to be able to selectively
>>> enable it for certain systems.
>>>
>>> This commit addresses this by introducing the cgroup_favordynmods=
>>> command-line option. This option works for both cgroup v1 and v2 and
>>> also allows for disabling favorynmods when the kernel built with
>>> CONFIG_FAVOR_DYNMODS=y.
>>>
>>> Signed-off-by: Luiz Capitulino <luizcap-vV1OtcyAfmbQT0dZR+AlfA@public.gmane.org>
>> Hi,
>>
>> Would somebody take a look at this patch please?
>>
>> Thanks,
>>
>> - Luiz
>
> The patch looks good to me. I do have a minor comment that it may be 
> better to use the relatively new __ro_after_init qualifier instead of 
> __read_mostly.
>
With this patch, CGROUP_FAVOR_DYNMODS kconfig option will have no user. 
So this config option should be removed as well.

Cheers,
Longman

Re: [PATH] cgroup: add cgroup_favordynmods= command-line option

[Waiman Long]

On 9/5/23 15:03, Waiman Long wrote:
>
> On 9/5/23 14:59, Waiman Long wrote:
>>
>> On 9/5/23 13:03, Luiz Capitulino wrote:
>>> On Thu, Aug 31, 2023 at 10:20:46AM -0400, Luiz Capitulino wrote:
>>>> We have a need of using favordynmods with cgroup v1, which doesn't 
>>>> support
>>>> changing mount flags during remount. Enabling CONFIG_FAVOR_DYNMODS at
>>>> build-time is not an option because we want to be able to selectively
>>>> enable it for certain systems.
>>>>
>>>> This commit addresses this by introducing the cgroup_favordynmods=
>>>> command-line option. This option works for both cgroup v1 and v2 and
>>>> also allows for disabling favorynmods when the kernel built with
>>>> CONFIG_FAVOR_DYNMODS=y.
>>>>
>>>> Signed-off-by: Luiz Capitulino <luizcap@amazon.com>
>>> Hi,
>>>
>>> Would somebody take a look at this patch please?
>>>
>>> Thanks,
>>>
>>> - Luiz
>>
>> The patch looks good to me. I do have a minor comment that it may be 
>> better to use the relatively new __ro_after_init qualifier instead of 
>> __read_mostly.
>>
> With this patch, CGROUP_FAVOR_DYNMODS kconfig option will have no 
> user. So this config option should be removed as well.

My bad. That is not true. Please ignore this comment.

Regards,
Longman

Re: [PATH] cgroup: add cgroup_favordynmods= command-line option

[Luiz Capitulino]

On 2023-09-05 15:06, Waiman Long wrote:

> 
> 
> 
> On 9/5/23 15:03, Waiman Long wrote:
>>
>> On 9/5/23 14:59, Waiman Long wrote:
>>>
>>> On 9/5/23 13:03, Luiz Capitulino wrote:
>>>> On Thu, Aug 31, 2023 at 10:20:46AM -0400, Luiz Capitulino wrote:
>>>>> We have a need of using favordynmods with cgroup v1, which doesn't
>>>>> support
>>>>> changing mount flags during remount. Enabling CONFIG_FAVOR_DYNMODS at
>>>>> build-time is not an option because we want to be able to selectively
>>>>> enable it for certain systems.
>>>>>
>>>>> This commit addresses this by introducing the cgroup_favordynmods=
>>>>> command-line option. This option works for both cgroup v1 and v2 and
>>>>> also allows for disabling favorynmods when the kernel built with
>>>>> CONFIG_FAVOR_DYNMODS=y.
>>>>>
>>>>> Signed-off-by: Luiz Capitulino <luizcap-vV1OtcyAfmbQT0dZR+AlfA@public.gmane.org>
>>>> Hi,
>>>>
>>>> Would somebody take a look at this patch please?
>>>>
>>>> Thanks,
>>>>
>>>> - Luiz
>>>
>>> The patch looks good to me. I do have a minor comment that it may be
>>> better to use the relatively new __ro_after_init qualifier instead of
>>> __read_mostly.

Will do the change and post v2 shortly.

>>>
>> With this patch, CGROUP_FAVOR_DYNMODS kconfig option will have no
>> user. So this config option should be removed as well.
> 
> My bad. That is not true. Please ignore this comment.

OK, big thanks for taking a look so quickly!

- Luiz

> 
> Regards,
> Longman
>