[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240919071801.GB1044577@kernel.org>
Date: Thu, 19 Sep 2024 08:19:37 +0100
From: Simon Horman <horms@...nel.org>
To: Pablo Neira Ayuso <pablo@...filter.org>
Cc: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
Jozsef Kadlecsik <kadlec@...filter.org>,
"David S. Miller" <davem@...emloft.net>,
Eric Dumazet <edumazet@...gle.com>,
Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
Nathan Chancellor <nathan@...nel.org>,
Nick Desaulniers <ndesaulniers@...gle.com>,
Bill Wendling <morbo@...gle.com>,
Justin Stitt <justinstitt@...gle.com>,
netfilter-devel@...r.kernel.org, coreteam@...filter.org,
netdev@...r.kernel.org, llvm@...ts.linux.dev
Subject: Re: [PATCH nf-next 0/2] netfilter: conntrack: label helpers
conditional compilation updates
On Wed, Sep 18, 2024 at 11:56:24PM +0200, Pablo Neira Ayuso wrote:
> On Wed, Sep 18, 2024 at 08:01:21PM +0300, Andy Shevchenko wrote:
> > On Wed, Sep 18, 2024 at 04:29:14PM +0200, Pablo Neira Ayuso wrote:
> > > On Wed, Sep 18, 2024 at 04:55:15PM +0300, Andy Shevchenko wrote:
> > > > On Wed, Sep 18, 2024 at 01:52:14PM +0200, Pablo Neira Ayuso wrote:
> > > > > On Mon, Sep 16, 2024 at 04:14:40PM +0100, Simon Horman wrote:
> > > > > > Hi,
> > > > > >
> > > > > > This short series updates conditional compilation of label helpers to:
> > > > > >
> > > > > > 1) Compile them regardless of if CONFIG_NF_CONNTRACK_LABELS is enabled
> > > > > > or not. It is safe to do so as the functions will always return 0 if
> > > > > > CONFIG_NF_CONNTRACK_LABELS is not enabled. And the compiler should
> > > > > > optimise waway the code. Which is the desired behaviour.
> > > > > >
> > > > > > 2) Only compile ctnetlink_label_size if CONFIG_NF_CONNTRACK_EVENTS is
> > > > > > enabled. This addresses a warning about this function being unused
> > > > > > in this case.
> > > > >
> > > > > Patch 1)
> > > > >
> > > > > -#ifdef CONFIG_NF_CONNTRACK_LABELS
> > > > > static inline int ctnetlink_label_size(const struct nf_conn *ct)
> > > > >
> > > > > Patch 2)
> > > > >
> > > > > +#ifdef CONFIG_NF_CONNTRACK_EVENTS
> > > > > static inline int ctnetlink_label_size(const struct nf_conn *ct)
> > > > >
> > > > > They both refer to ctnetlink_label_size(), #ifdef check is not
> > > > > correct.
> > > >
> > > > But the first one touches more, no?
> > >
> > > Yes, it also remove a #define ctnetlink_label_size() macro in patch #1.
> > > I am fine with this series as is.
> >
> > What I meant is that the original patch 1 takes care about definitions of
> > two functions. Not just a single one.
>
> My understanding is that #ifdef CONFIG_NF_CONNTRACK_LABELS that wraps
> ctnetlink_label_size() is not correct (patch 1), instead
> CONFIG_NF_CONNTRACK_EVENTS should be used (patch 2).
>
> Then, as a side effect this goes away (patch 1):
>
> -#else
> -#define ctnetlink_dump_labels(a, b) (0)
> -#define ctnetlink_label_size(a) (0)
> -#endif
>
> that is why I am proposing to coaleasce these two patches in one.
Thanks,
Just to clarify. I did think there is value in separating the two changes.
But that was a subjective judgement on my part.
Your understanding of the overall change is correct.
And if it is preferred to have a single patch - as seems to be the case -
then that is fine by me.
Going forward, I'll try to remember not to split-up patches for netfilter
so much.
Kind regards,
Simon
Powered by blists - more mailing lists