2008-06-04 07:48:44

by Pavel Emelyanov

Subject: [PATCH 2/2] devscgroup: relax task to dev_cgroup conversion

Two functions, that need to get a device_cgroup from a task (they
are devcgroup_inode_permission and devcgroup_inode_mknod) make it
in a strange way:

They get a css_set from task, then a subsys_state from css_set,
then a cgroup from the state and then a subsys_state again from
the cgroup. Besides, the devices_subsys_id is read from memory,
whilst there's a enum-ed constant for it.

Optimize this part a bit:
1. Get the subsys_stats form the task and be done - no 2 extra
dereferences,
2. Use the device_subsys_id constant, not the value from memory
(i.e. one less dereference).

Found while preparing 2.6.26 OpenVZ port.

Signed-off-by: Pavel Emelyanov <[email protected]>

---
security/device_cgroup.c | 10 ++++------
1 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/security/device_cgroup.c b/security/device_cgroup.c
index 15f2f80..f9941a7 100644
--- a/security/device_cgroup.c
+++ b/security/device_cgroup.c
@@ -506,7 +506,6 @@ struct cgroup_subsys devices_subsys = {

int devcgroup_inode_permission(struct inode *inode, int mask)
{
- struct cgroup *cgroup;
struct dev_cgroup *dev_cgroup;
struct dev_whitelist_item *wh;

@@ -515,8 +514,8 @@ int devcgroup_inode_permission(struct inode *inode, int mask)
return 0;
if (!S_ISBLK(inode->i_mode) && !S_ISCHR(inode->i_mode))
return 0;
- cgroup = task_cgroup(current, devices_subsys.subsys_id);
- dev_cgroup = cgroup_to_devcgroup(cgroup);
+ dev_cgroup = css_to_devcgroup(task_subsys_state(current,
+ devices_subsys_id));
if (!dev_cgroup)
return 0;

@@ -547,12 +546,11 @@ acc_check:

int devcgroup_inode_mknod(int mode, dev_t dev)
{
- struct cgroup *cgroup;
struct dev_cgroup *dev_cgroup;
struct dev_whitelist_item *wh;

- cgroup = task_cgroup(current, devices_subsys.subsys_id);
- dev_cgroup = cgroup_to_devcgroup(cgroup);
+ dev_cgroup = css_to_devcgroup(task_subsys_state(current,
+ devices_subsys_id));
if (!dev_cgroup)
return 0;

--
1.5.3.4

2008-06-04 09:20:40

by Paul Menage

Subject: Re: [PATCH 2/2] devscgroup: relax task to dev_cgroup conversion

On Wed, Jun 4, 2008 at 12:39 AM, Pavel Emelyanov <[email protected]> wrote:
> Two functions, that need to get a device_cgroup from a task (they
> are devcgroup_inode_permission and devcgroup_inode_mknod) make it
> in a strange way:
>
> They get a css_set from task, then a subsys_state from css_set,
> then a cgroup from the state and then a subsys_state again from
> the cgroup. Besides, the devices_subsys_id is read from memory,
> whilst there's a enum-ed constant for it.
>
> Optimize this part a bit:
> 1. Get the subsys_stats form the task and be done - no 2 extra
> dereferences,
> 2. Use the device_subsys_id constant, not the value from memory
> (i.e. one less dereference).
>
> Found while preparing 2.6.26 OpenVZ port.
>
> Signed-off-by: Pavel Emelyanov <[email protected]>

Acked-by: Paul Menage <[email protected]>

Yes, that's more how the css_set pointers are meant to be used. Maybe
improve it further by extracting a task_to_devcgroup() inline
function?

Paul

>
> ---
> security/device_cgroup.c | 10 ++++------
> 1 files changed, 4 insertions(+), 6 deletions(-)
>
> diff --git a/security/device_cgroup.c b/security/device_cgroup.c
> index 15f2f80..f9941a7 100644
> --- a/security/device_cgroup.c
> +++ b/security/device_cgroup.c
> @@ -506,7 +506,6 @@ struct cgroup_subsys devices_subsys = {
>
> int devcgroup_inode_permission(struct inode *inode, int mask)
> {
> - struct cgroup *cgroup;
> struct dev_cgroup *dev_cgroup;
> struct dev_whitelist_item *wh;
>
> @@ -515,8 +514,8 @@ int devcgroup_inode_permission(struct inode *inode, int mask)
> return 0;
> if (!S_ISBLK(inode->i_mode) && !S_ISCHR(inode->i_mode))
> return 0;
> - cgroup = task_cgroup(current, devices_subsys.subsys_id);
> - dev_cgroup = cgroup_to_devcgroup(cgroup);
> + dev_cgroup = css_to_devcgroup(task_subsys_state(current,
> + devices_subsys_id));
> if (!dev_cgroup)
> return 0;
>
> @@ -547,12 +546,11 @@ acc_check:
>
> int devcgroup_inode_mknod(int mode, dev_t dev)
> {
> - struct cgroup *cgroup;
> struct dev_cgroup *dev_cgroup;
> struct dev_whitelist_item *wh;
>
> - cgroup = task_cgroup(current, devices_subsys.subsys_id);
> - dev_cgroup = cgroup_to_devcgroup(cgroup);
> + dev_cgroup = css_to_devcgroup(task_subsys_state(current,
> + devices_subsys_id));
> if (!dev_cgroup)
> return 0;
>
> --
> 1.5.3.4
>
>

2008-06-04 09:27:59

by Pavel Emelyanov

Subject: Re: [PATCH 2/2] devscgroup: relax task to dev_cgroup conversion

Paul Menage wrote:
> On Wed, Jun 4, 2008 at 12:39 AM, Pavel Emelyanov <[email protected]> wrote:
>> Two functions, that need to get a device_cgroup from a task (they
>> are devcgroup_inode_permission and devcgroup_inode_mknod) make it
>> in a strange way:
>>
>> They get a css_set from task, then a subsys_state from css_set,
>> then a cgroup from the state and then a subsys_state again from
>> the cgroup. Besides, the devices_subsys_id is read from memory,
>> whilst there's a enum-ed constant for it.
>>
>> Optimize this part a bit:
>> 1. Get the subsys_stats form the task and be done - no 2 extra
>> dereferences,
>> 2. Use the device_subsys_id constant, not the value from memory
>> (i.e. one less dereference).
>>
>> Found while preparing 2.6.26 OpenVZ port.
>>
>> Signed-off-by: Pavel Emelyanov <[email protected]>
>
> Acked-by: Paul Menage <[email protected]>
>
> Yes, that's more how the css_set pointers are meant to be used. Maybe
> improve it further by extracting a task_to_devcgroup() inline
> function?

Well... Maybe. I'll make a 3rd patch a bit later.
I'm now testing RCU-protected whitelists.

> Paul
>
>> ---
>> security/device_cgroup.c | 10 ++++------
>> 1 files changed, 4 insertions(+), 6 deletions(-)
>>
>> diff --git a/security/device_cgroup.c b/security/device_cgroup.c
>> index 15f2f80..f9941a7 100644
>> --- a/security/device_cgroup.c
>> +++ b/security/device_cgroup.c
>> @@ -506,7 +506,6 @@ struct cgroup_subsys devices_subsys = {
>>
>> int devcgroup_inode_permission(struct inode *inode, int mask)
>> {
>> - struct cgroup *cgroup;
>> struct dev_cgroup *dev_cgroup;
>> struct dev_whitelist_item *wh;
>>
>> @@ -515,8 +514,8 @@ int devcgroup_inode_permission(struct inode *inode, int mask)
>> return 0;
>> if (!S_ISBLK(inode->i_mode) && !S_ISCHR(inode->i_mode))
>> return 0;
>> - cgroup = task_cgroup(current, devices_subsys.subsys_id);
>> - dev_cgroup = cgroup_to_devcgroup(cgroup);
>> + dev_cgroup = css_to_devcgroup(task_subsys_state(current,
>> + devices_subsys_id));
>> if (!dev_cgroup)
>> return 0;
>>
>> @@ -547,12 +546,11 @@ acc_check:
>>
>> int devcgroup_inode_mknod(int mode, dev_t dev)
>> {
>> - struct cgroup *cgroup;
>> struct dev_cgroup *dev_cgroup;
>> struct dev_whitelist_item *wh;
>>
>> - cgroup = task_cgroup(current, devices_subsys.subsys_id);
>> - dev_cgroup = cgroup_to_devcgroup(cgroup);
>> + dev_cgroup = css_to_devcgroup(task_subsys_state(current,
>> + devices_subsys_id));
>> if (!dev_cgroup)
>> return 0;
>>
>> --
>> 1.5.3.4
>>
>>
>