[net-next] net: msg_zerocopy: elide page accounting if RLIM_INFINITY

Commit Message

Willem de Bruijn Feb. 14, 2023, 3:57 p.m. UTC

From: Willem de Bruijn <[e-mail]>

MSG_ZEROCOPY ensures that pinned user pages do not exceed the limit.
If no limit is set, skip this accounting as otherwise expensive
atomic_long operations are called for no reason.

This accounting is already skipped for privileged users. Rely on the
same mechanism: if no mmp->user is set, mm_unaccount_pinned_pages does
not decrement either.

Tested by running tools/testing/selftests/net/msg_zerocopy.sh with
an unprivileged user for the TXMODE binary:

    ip netns exec "${NS1}" sudo -u "{$USER}" "${BIN}" "-${IP}" ...

Signed-off-by: Willem de Bruijn <[e-mail]>
---
 net/core/skbuff.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

Comments

Eric Dumazet Feb. 15, 2023, 9:08 a.m. UTC | #1

On Tue, Feb 14, 2023 at 4:57 PM Willem de Bruijn
<[e-mail]> wrote:
>
> From: Willem de Bruijn <[e-mail]>
>
> MSG_ZEROCOPY ensures that pinned user pages do not exceed the limit.
> If no limit is set, skip this accounting as otherwise expensive
> atomic_long operations are called for no reason.
>
> This accounting is already skipped for privileged users.

 (privileged as in CAP_IPC_LOCK)

Rely on the
> same mechanism: if no mmp->user is set, mm_unaccount_pinned_pages does
> not decrement either.
>
> Tested by running tools/testing/selftests/net/msg_zerocopy.sh with
> an unprivileged user for the TXMODE binary:
>
>     ip netns exec "${NS1}" sudo -u "{$USER}" "${BIN}" "-${IP}" ...
>
> Signed-off-by: Willem de Bruijn <[e-mail]>
> ---

Reviewed-by: Eric Dumazet <[e-mail]>

[e-mail] Feb. 16, 2023, 5:30 a.m. UTC | #2

Hello:

This patch was applied to netdev/net-next.git (master)
by Jakub Kicinski <[e-mail]>:

On Tue, 14 Feb 2023 10:57:40 -0500 you wrote:
> From: Willem de Bruijn <[e-mail]>
> 
> MSG_ZEROCOPY ensures that pinned user pages do not exceed the limit.
> If no limit is set, skip this accounting as otherwise expensive
> atomic_long operations are called for no reason.
> 
> This accounting is already skipped for privileged users. Rely on the
> same mechanism: if no mmp->user is set, mm_unaccount_pinned_pages does
> not decrement either.
> 
> [...]

Here is the summary with links:
  - [net-next] net: msg_zerocopy: elide page accounting if RLIM_INFINITY
    https://git.kernel.org/netdev/net-next/c/14ade6ba4120

You are awesome, thank you!

Patch

diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 13ea10cf8544..98ebce9f6a51 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -1406,14 +1406,18 @@  EXPORT_SYMBOL_GPL(skb_morph);
 
 int mm_account_pinned_pages(struct mmpin *mmp, size_t size)
 {
-	unsigned long max_pg, num_pg, new_pg, old_pg;
+	unsigned long max_pg, num_pg, new_pg, old_pg, rlim;
 	struct user_struct *user;
 
 	if (capable(CAP_IPC_LOCK) || !size)
 		return 0;
 
+	rlim = rlimit(RLIMIT_MEMLOCK);
+	if (rlim == RLIM_INFINITY)
+		return 0;
+
 	num_pg = (size >> PAGE_SHIFT) + 2;	/* worst case */
-	max_pg = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
+	max_pg = rlim >> PAGE_SHIFT;
 	user = mmp->user ? : current_user();
 
 	old_pg = atomic_long_read(&user->locked_vm);