[PATCH lnf-queue] src: prepare for new release

[PATCH lnf-queue] src: prepare for new release

[Florian Westphal]

Its been 5 years, things like GSO and NFQA_CT have been in the master
branch for years but no released versions support them.

Needs bump of libversion, we have new interfaces but are compatible with
still compatible with the earlier 1.0.2 release.

Signed-off-by: Florian Westphal <fw@strlen.de>
---
 Hi, some people are rightfully complaining about new features
 not being in any release, so I suggest we do one soon.

diff --git a/src/Makefile.am b/src/Makefile.am
index 6e5d50325525..9fdccfbd9ea5 100644
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -18,7 +18,7 @@
 # set age to 0.
 # </snippet>
 #
-LIBVERSION=4:0:3
+LIBVERSION=5:0:4
 
 include ${top_srcdir}/Make_global.am
 
diff --git a/src/libnetfilter_queue.c b/src/libnetfilter_queue.c
index 211a8ba75d79..69db0e02fc5b 100644
--- a/src/libnetfilter_queue.c
+++ b/src/libnetfilter_queue.c
@@ -658,6 +658,7 @@ EXPORT_SYMBOL(nfq_set_mode);
  * - NFQA_CFG_F_GSO (requires Linux kernel >= 3.10): the kernel will
  *   not normalize offload packets, i.e. your application will need to
  *   be able to handle packets larger than the mtu (up to 64k).
+ *   Normalization is expensive, so this flag should always be set.
  *
  *   If your application validates checksums (e.g., tcp checksum),
  *   then you must also check if the NFQA_SKB_INFO attribute is present.
-- 
2.13.0

Re: [PATCH lnf-queue] src: prepare for new release

[Pablo Neira Ayuso]

On Wed, Jun 21, 2017 at 12:28:52AM +0200, Florian Westphal wrote:
> diff --git a/src/libnetfilter_queue.c b/src/libnetfilter_queue.c
> index 211a8ba75d79..69db0e02fc5b 100644
> --- a/src/libnetfilter_queue.c
> +++ b/src/libnetfilter_queue.c
> @@ -658,6 +658,7 @@ EXPORT_SYMBOL(nfq_set_mode);
>   * - NFQA_CFG_F_GSO (requires Linux kernel >= 3.10): the kernel will
>   *   not normalize offload packets, i.e. your application will need to
>   *   be able to handle packets larger than the mtu (up to 64k).
> + *   Normalization is expensive, so this flag should always be set.

With this on, I see more chances to hit the netlink attribute
limitation, so packets on the 64k will be truncated?

Probably good to document this too.