[PATCH] fs/namei: Don't update atime when some errors occur in get_link

[PATCH] fs/namei: Don't update atime when some errors occur in get_link

[Hao Ge]

Perhaps we have some errors occur(like security),then we don't update
atime,because we didn't actually access it

Signed-off-by: Hao Ge <gehao@kylinos.cn>
---
 fs/namei.c | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/fs/namei.c b/fs/namei.c
index 71c13b2990b4..033d36d5c1c5 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -1779,15 +1779,6 @@ static const char *pick_link(struct nameidata *nd, struct path *link,
 			unlikely(link->mnt->mnt_flags & MNT_NOSYMFOLLOW))
 		return ERR_PTR(-ELOOP);
 
-	if (!(nd->flags & LOOKUP_RCU)) {
-		touch_atime(&last->link);
-		cond_resched();
-	} else if (atime_needs_update(&last->link, inode)) {
-		if (!try_to_unlazy(nd))
-			return ERR_PTR(-ECHILD);
-		touch_atime(&last->link);
-	}
-
 	error = security_inode_follow_link(link->dentry, inode,
 					   nd->flags & LOOKUP_RCU);
 	if (unlikely(error))
@@ -1810,6 +1801,16 @@ static const char *pick_link(struct nameidata *nd, struct path *link,
 		if (IS_ERR(res))
 			return res;
 	}
+
+	if (!(nd->flags & LOOKUP_RCU)) {
+		touch_atime(&last->link);
+		cond_resched();
+	} else if (atime_needs_update(&last->link, inode)) {
+		if (!try_to_unlazy(nd))
+			return ERR_PTR(-ECHILD);
+		touch_atime(&last->link);
+	}
+
 	if (*res == '/') {
 		error = nd_jump_root(nd);
 		if (unlikely(error))
-- 
2.25.1

Re: [PATCH] fs/namei: Don't update atime when some errors occur in get_link

[Christian Brauner]

On Tue, Dec 05, 2023 at 03:17:33PM +0800, Hao Ge wrote:
> Perhaps we have some errors occur(like security),then we don't update
> atime,because we didn't actually access it
> 
> Signed-off-by: Hao Ge <gehao@kylinos.cn>
> ---

We didn't follow the link but we accessed it. I guess it's not completey
clear what's correct here so I'd just leave it as is.

Re: [PATCH] fs/namei: Don't update atime when some errors occur in get_link

[Hao Ge]

> On Dec 5, 2023, at 19:07, Christian Brauner <brauner@kernel.org> wrote:
> 
> On Tue, Dec 05, 2023 at 03:17:33PM +0800, Hao Ge wrote:
>> Perhaps we have some errors occur(like security),then we don't update
>> atime,because we didn't actually access it
>> 
>> Signed-off-by: Hao Ge <gehao@kylinos.cn>
>> ---
> 
> We didn't follow the link but we accessed it. I guess it's not completey
> clear what's correct here so I'd just leave it as is.
Hi brauner
Thank your for your reply.
I just thought of a situation that user access a link failed due to some error(like permission issue),maybe report some error to user, actually user don’t get anything,but atime still update.
Maybe your are right,after all,we still tried to visit.
Thanks
Best Regards 
Hao