[bpf] xsk: Add truesize to skb_add_rx_frag().

Commit Message

Sebastian Andrzej Siewior Feb. 2, 2024, 4:32 p.m. UTC

xsk_build_skb() allocates a page and adds it to the skb via
skb_add_rx_frag() and specifies 0 for truesize. This leads to a warning
in skb_add_rx_frag() with CONFIG_DEBUG_NET enabled because size is
larger than truesize.

Increasing truesize requires to add the same amount to socket's
sk_wmem_alloc counter in order not to underflow the counter during
release in the destructor (sock_wfree()).

Pass the size of the allocated page as truesize to skb_add_rx_frag().
Add this mount to socket's sk_wmem_alloc counter.

Fixes: cf24f5a5feea ("xsk: add support for AF_XDP multi-buffer on Tx path")
Signed-off-by: Sebastian Andrzej Siewior <[e-mail]>
---
Noticed by running test_xsk.sh

 net/xdp/xsk.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Comments

Fijalkowski, Maciej Feb. 5, 2024, 11:13 p.m. UTC | #1

On Fri, Feb 02, 2024 at 05:32:20PM +0100, Sebastian Andrzej Siewior wrote:
> xsk_build_skb() allocates a page and adds it to the skb via
> skb_add_rx_frag() and specifies 0 for truesize. This leads to a warning
> in skb_add_rx_frag() with CONFIG_DEBUG_NET enabled because size is
> larger than truesize.
> 
> Increasing truesize requires to add the same amount to socket's
> sk_wmem_alloc counter in order not to underflow the counter during
> release in the destructor (sock_wfree()).
> 
> Pass the size of the allocated page as truesize to skb_add_rx_frag().
> Add this mount to socket's sk_wmem_alloc counter.
> 
> Fixes: cf24f5a5feea ("xsk: add support for AF_XDP multi-buffer on Tx path")
> Signed-off-by: Sebastian Andrzej Siewior <[e-mail]>

Thanks a lot! Of course fix is totally correct.

Acked-by: Maciej Fijalkowski <[e-mail]>

> ---
> Noticed by running test_xsk.sh
> 
>  net/xdp/xsk.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
> index 0348e4bde23b2..3050739cfe1e0 100644
> --- a/net/xdp/xsk.c
> +++ b/net/xdp/xsk.c
> @@ -744,7 +744,8 @@ static struct sk_buff *xsk_build_skb(struct xdp_sock *xs,
>  			memcpy(vaddr, buffer, len);
>  			kunmap_local(vaddr);
>  
> -			skb_add_rx_frag(skb, nr_frags, page, 0, len, 0);
> +			skb_add_rx_frag(skb, nr_frags, page, 0, len, PAGE_SIZE);
> +			refcount_add(PAGE_SIZE, &xs->sk.sk_wmem_alloc);
>  		}
>  
>  		if (first_frag && desc->options & XDP_TX_METADATA) {
> -- 
> 2.43.0
>

Sebastian Andrzej Siewior Feb. 13, 2024, 3:37 p.m. UTC | #2

On 2024-02-02 17:32:20 [+0100], To [e-mail] wrote:
> xsk_build_skb() allocates a page and adds it to the skb via
> skb_add_rx_frag() and specifies 0 for truesize. This leads to a warning
> in skb_add_rx_frag() with CONFIG_DEBUG_NET enabled because size is
> larger than truesize.
> 
> Increasing truesize requires to add the same amount to socket's
> sk_wmem_alloc counter in order not to underflow the counter during
> release in the destructor (sock_wfree()).
> 
> Pass the size of the allocated page as truesize to skb_add_rx_frag().
> Add this mount to socket's sk_wmem_alloc counter.
> 
> Fixes: cf24f5a5feea ("xsk: add support for AF_XDP multi-buffer on Tx path")
> Signed-off-by: Sebastian Andrzej Siewior <[e-mail]>

Is this stuck somewhere or just waiting for review/ etc?
Patchwork says new and the ci failure on s390x seems unrelated.

Sebastian

Fijalkowski, Maciej Feb. 13, 2024, 3:59 p.m. UTC | #3

On Tue, Feb 13, 2024 at 04:37:37PM +0100, Sebastian Andrzej Siewior wrote:
> On 2024-02-02 17:32:20 [+0100], To [e-mail] wrote:
> > xsk_build_skb() allocates a page and adds it to the skb via
> > skb_add_rx_frag() and specifies 0 for truesize. This leads to a warning
> > in skb_add_rx_frag() with CONFIG_DEBUG_NET enabled because size is
> > larger than truesize.
> > 
> > Increasing truesize requires to add the same amount to socket's
> > sk_wmem_alloc counter in order not to underflow the counter during
> > release in the destructor (sock_wfree()).
> > 
> > Pass the size of the allocated page as truesize to skb_add_rx_frag().
> > Add this mount to socket's sk_wmem_alloc counter.
> > 
> > Fixes: cf24f5a5feea ("xsk: add support for AF_XDP multi-buffer on Tx path")
> > Signed-off-by: Sebastian Andrzej Siewior <[e-mail]>
> 
> Is this stuck somewhere or just waiting for review/ etc?
> Patchwork says new and the ci failure on s390x seems unrelated.

I acked it week ago or so, maybe bpf maintainers missed it as I don't see
them being CCed? Adding in now.

> 
> Sebastian

Sebastian Andrzej Siewior Feb. 13, 2024, 4:02 p.m. UTC | #4

On 2024-02-13 16:59:13 [+0100], Maciej Fijalkowski wrote:
> I acked it week ago or so, maybe bpf maintainers missed it as I don't see
> them being CCed? Adding in now.

Thank you. They did not pop up on get_maintainer so I did not add them. 

Sebastian

[e-mail] Feb. 13, 2024, 10:20 p.m. UTC | #5

Hello:

This patch was applied to bpf/bpf.git (master)
by Daniel Borkmann <[e-mail]>:

On Fri,  2 Feb 2024 17:32:20 +0100 you wrote:
> xsk_build_skb() allocates a page and adds it to the skb via
> skb_add_rx_frag() and specifies 0 for truesize. This leads to a warning
> in skb_add_rx_frag() with CONFIG_DEBUG_NET enabled because size is
> larger than truesize.
> 
> Increasing truesize requires to add the same amount to socket's
> sk_wmem_alloc counter in order not to underflow the counter during
> release in the destructor (sock_wfree()).
> 
> [...]

Here is the summary with links:
  - [bpf] xsk: Add truesize to skb_add_rx_frag().
    https://git.kernel.org/bpf/bpf/c/2127c6043836

You are awesome, thank you!

Patch

diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
index 0348e4bde23b2..3050739cfe1e0 100644
--- a/net/xdp/xsk.c
+++ b/net/xdp/xsk.c
@@ -744,7 +744,8 @@  static struct sk_buff *xsk_build_skb(struct xdp_sock *xs,
 			memcpy(vaddr, buffer, len);
 			kunmap_local(vaddr);
 
-			skb_add_rx_frag(skb, nr_frags, page, 0, len, 0);
+			skb_add_rx_frag(skb, nr_frags, page, 0, len, PAGE_SIZE);
+			refcount_add(PAGE_SIZE, &xs->sk.sk_wmem_alloc);
 		}
 
 		if (first_frag && desc->options & XDP_TX_METADATA) {