| Message ID | 20221228012902.1510901-1-sj@kernel.org (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v2] mm/page_reporting: replace rcu_access_pointer() with rcu_dereference_protected() | expand |
On 2022/12/28 9:29, SeongJae Park wrote: > Page reporting fetches pr_dev_info using rcu_access_pointer(), which is > for safely fetching a pointer that will not be dereferenced but could > concurrently updated. The code indeed does not dereference pr_dev_info > after fetcing it using rcu_access_pointer(), but it fetches the pointer Thanks for your work. Might something to improve. s/fetcing/fetching/ > while concurrent updtes to the pointer is avoided by holding the update s/updtes/updates/ > side lock, page_reporting_mutex. > > In the case, rcu_dereference_protected() is recommended because it > provides better readability and performance on some cases, as > rcu_dereference_protected() avoids use of READ_ONCE(). Replace the > rcu_access_pointer() calls with rcu_dereference_protected(). > > Signed-off-by: SeongJae Park <sj@kernel.org> > --- > Changes from v1 > (https://lore.kernel.org/linux-mm/20221227192158.2553-1-sj@kernel.org/) > - Explicitly set the protection condition (Matthew Wilcox) > > mm/page_reporting.c | 6 ++++-- > 1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/mm/page_reporting.c b/mm/page_reporting.c > index 79a8554f024c..5c557a3e1423 100644 > --- a/mm/page_reporting.c > +++ b/mm/page_reporting.c > @@ -356,7 +356,8 @@ int page_reporting_register(struct page_reporting_dev_info *prdev) > mutex_lock(&page_reporting_mutex); > > /* nothing to do if already in use */ > - if (rcu_access_pointer(pr_dev_info)) { > + if (rcu_dereference_protected(pr_dev_info, > + lockdep_is_held(&page_reporting_order))) { I think it should be lockdep_is_held(&page_reporting_mutex) instead of lockdep_is_held(&page_reporting_order) here? Thanks, Miaohe Lin
On Wed, 28 Dec 2022 09:45:00 +0800 Miaohe Lin <linmiaohe@huawei.com> wrote: > On 2022/12/28 9:29, SeongJae Park wrote: > > Page reporting fetches pr_dev_info using rcu_access_pointer(), which is > > for safely fetching a pointer that will not be dereferenced but could > > concurrently updated. The code indeed does not dereference pr_dev_info > > after fetcing it using rcu_access_pointer(), but it fetches the pointer > > Thanks for your work. Might something to improve. > > s/fetcing/fetching/ > > > while concurrent updtes to the pointer is avoided by holding the update > > s/updtes/updates/ Thank you! I shall add these to scripts/spelling.txt. > > > side lock, page_reporting_mutex. > > > > In the case, rcu_dereference_protected() is recommended because it > > provides better readability and performance on some cases, as > > rcu_dereference_protected() avoids use of READ_ONCE(). Replace the > > rcu_access_pointer() calls with rcu_dereference_protected(). > > > > Signed-off-by: SeongJae Park <sj@kernel.org> > > --- > > Changes from v1 > > (https://lore.kernel.org/linux-mm/20221227192158.2553-1-sj@kernel.org/) > > - Explicitly set the protection condition (Matthew Wilcox) > > > > mm/page_reporting.c | 6 ++++-- > > 1 file changed, 4 insertions(+), 2 deletions(-) > > > > diff --git a/mm/page_reporting.c b/mm/page_reporting.c > > index 79a8554f024c..5c557a3e1423 100644 > > --- a/mm/page_reporting.c > > +++ b/mm/page_reporting.c > > @@ -356,7 +356,8 @@ int page_reporting_register(struct page_reporting_dev_info *prdev) > > mutex_lock(&page_reporting_mutex); > > > > /* nothing to do if already in use */ > > - if (rcu_access_pointer(pr_dev_info)) { > > + if (rcu_dereference_protected(pr_dev_info, > > + lockdep_is_held(&page_reporting_order))) { > > I think it should be lockdep_is_held(&page_reporting_mutex) instead of > lockdep_is_held(&page_reporting_order) here? You're right, thank you for finding this. I will fix these in the next version. Thanks, SJ > > Thanks, > Miaohe Lin
diff --git a/mm/page_reporting.c b/mm/page_reporting.c index 79a8554f024c..5c557a3e1423 100644 --- a/mm/page_reporting.c +++ b/mm/page_reporting.c @@ -356,7 +356,8 @@ int page_reporting_register(struct page_reporting_dev_info *prdev) mutex_lock(&page_reporting_mutex); /* nothing to do if already in use */ - if (rcu_access_pointer(pr_dev_info)) { + if (rcu_dereference_protected(pr_dev_info, + lockdep_is_held(&page_reporting_order))) { err = -EBUSY; goto err_out; } @@ -401,7 +402,8 @@ void page_reporting_unregister(struct page_reporting_dev_info *prdev) { mutex_lock(&page_reporting_mutex); - if (rcu_access_pointer(pr_dev_info) == prdev) { + if (prdev == rcu_dereference_protected(pr_dev_info, + lockdep_is_held(&page_reporting_mutex))) { /* Disable page reporting notification */ RCU_INIT_POINTER(pr_dev_info, NULL); synchronize_rcu();
Page reporting fetches pr_dev_info using rcu_access_pointer(), which is for safely fetching a pointer that will not be dereferenced but could concurrently updated. The code indeed does not dereference pr_dev_info after fetcing it using rcu_access_pointer(), but it fetches the pointer while concurrent updtes to the pointer is avoided by holding the update side lock, page_reporting_mutex. In the case, rcu_dereference_protected() is recommended because it provides better readability and performance on some cases, as rcu_dereference_protected() avoids use of READ_ONCE(). Replace the rcu_access_pointer() calls with rcu_dereference_protected(). Signed-off-by: SeongJae Park <sj@kernel.org> --- Changes from v1 (https://lore.kernel.org/linux-mm/20221227192158.2553-1-sj@kernel.org/) - Explicitly set the protection condition (Matthew Wilcox) mm/page_reporting.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)