On 2022/9/17 16:21, Andy Shevchenko wrote:
On Sat, Sep 17, 2022 at 5:12 AM Shang XiaoJing <shangxiaojing@xxxxxxxxxx> wrote:
Instead of invoking a synchronize_rcu() to free a pointer after a grace
period, we can directly make use of new API that does the same but in
a new
more efficient way.
...
sorry, the previous same patch have the wrong v3 flag in subject.
has
First of all, this brings even more confusion. Since the below needs
some work, we want to have a v4 with a proper changelog, explaining
also v1 and v3.
Ok, I 'll figure out a more clear changelog.
Besides, v3 is my miss, please ignore it, next patch will be v2.
...
- synchronize_rcu();
- kfree(peer);
+ kvfree_rcu(peer);
This is not equivalent. The commit message doesn't explain why the "v"
variant is needed. So, shouldn't it be the plain kfree_rcu()?
kfree_rcu() is defined as kvfree_rcu(), which will free the pointer with the
corresponding way(use is_vmalloc_addr() to determine vfree or kfree). For
clearlity, kfree_rcu() is a better way, will be fixed in patch v2.
Thanks :-),
Shang XiaoJing
