[PATCH bpf-next] net, xdp: allow metadata > 32

[PATCH bpf-next] net, xdp: allow metadata > 32

[Larysa Zaremba]

32 bytes may be not enough for some custom metadata. Relax the restriction,
allow metadata larger than 32 bytes and make __skb_metadata_differs() work
with bigger lengths.

Now size of metadata is only limited by the fact it is stored as u8
in skb_shared_info, so the upper limit is now is 255. Other important
conditions, such as having enough space for xdp_frame building, are already
checked in bpf_xdp_adjust_meta().

The requirement of having its length aligned to 4 bytes is still
valid.

Signed-off-by: Aleksander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com>
---
This patch was previously a part of an old BTF-based hints RFC.
Then it was included into "XDP metadata via kfuncs for ice":
https://lore.kernel.org/bpf/20230811161509.19722-1-larysa.zaremba@intel.com/
It is not longer needed in the series, but presents a useful change on its own.

 include/linux/skbuff.h | 13 ++++++++-----
 include/net/xdp.h      |  7 ++++++-
 2 files changed, 14 insertions(+), 6 deletions(-)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 97bfef071255..a361a9b8767c 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -4232,10 +4232,13 @@ static inline bool __skb_metadata_differs(const struct sk_buff *skb_a,
 {
 	const void *a = skb_metadata_end(skb_a);
 	const void *b = skb_metadata_end(skb_b);
-	/* Using more efficient varaiant than plain call to memcmp(). */
-#if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
 	u64 diffs = 0;
 
+	if (!IS_ENABLED(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) ||
+	    BITS_PER_LONG != 64)
+		goto slow;
+
+	/* Using more efficient variant than plain call to memcmp(). */
 	switch (meta_len) {
 #define __it(x, op) (x -= sizeof(u##op))
 #define __it_diff(a, b, op) (*(u##op *)__it(a, op)) ^ (*(u##op *)__it(b, op))
@@ -4255,11 +4258,11 @@ static inline bool __skb_metadata_differs(const struct sk_buff *skb_a,
 		fallthrough;
 	case  4: diffs |= __it_diff(a, b, 32);
 		break;
+	default:
+slow:
+		return memcmp(a - meta_len, b - meta_len, meta_len);
 	}
 	return diffs;
-#else
-	return memcmp(a - meta_len, b - meta_len, meta_len);
-#endif
 }
 
 static inline bool skb_metadata_differs(const struct sk_buff *skb_a,
diff --git a/include/net/xdp.h b/include/net/xdp.h
index 349c36fb5fd8..84ba1bb50b8e 100644
--- a/include/net/xdp.h
+++ b/include/net/xdp.h
@@ -369,7 +369,12 @@ xdp_data_meta_unsupported(const struct xdp_buff *xdp)
 
 static inline bool xdp_metalen_invalid(unsigned long metalen)
 {
-	return (metalen & (sizeof(__u32) - 1)) || (metalen > 32);
+	typeof(metalen) meta_max;
+
+	meta_max = type_max(typeof_member(struct skb_shared_info, meta_len));
+	BUILD_BUG_ON(!__builtin_constant_p(meta_max));
+
+	return !IS_ALIGNED(metalen, sizeof(u32)) || metalen > meta_max;
 }
 
 struct xdp_attachment_info {
-- 
2.41.0

Re: [PATCH bpf-next] net, xdp: allow metadata > 32

[Jakub Kicinski]

On Thu, 26 Oct 2023 18:56:59 +0200 Larysa Zaremba wrote:
>  static inline bool xdp_metalen_invalid(unsigned long metalen)
>  {
> -	return (metalen & (sizeof(__u32) - 1)) || (metalen > 32);
> +	typeof(metalen) meta_max;

The use of typeof() looks a bit unnecessary..

Re: [PATCH bpf-next] net, xdp: allow metadata > 32

[Larysa Zaremba]

On Fri, Oct 27, 2023 at 01:09:30PM -0700, Jakub Kicinski wrote:
> On Thu, 26 Oct 2023 18:56:59 +0200 Larysa Zaremba wrote:
> >  static inline bool xdp_metalen_invalid(unsigned long metalen)
> >  {
> > -	return (metalen & (sizeof(__u32) - 1)) || (metalen > 32);
> > +	typeof(metalen) meta_max;
> 
> The use of typeof() looks a bit unnecessary..
>

You are probably right, will send v2 without it.