Re: [PATCH net-next] net: core: Namespace-ify sysctl_wmem_default and sysctl_rmem_default

[Jakub Kicinski]

On Mon, 18 Jan 2021 12:15:18 +0100 Christian Brauner wrote:
> On Sun, Jan 17, 2021 at 06:23:19PM +0800, menglong8.dong@xxxxxxxxx wrote:
> > From: Menglong Dong <dong.menglong@xxxxxxxxxx>
> > 
> > For now, sysctl_wmem_default and sysctl_rmem_default are globally
> > unified. It's not convenient in some case. For example, when we
> > use docker and try to control the default udp socket receive buffer
> > for each container.
> > 
> > For that reason, make sysctl_wmem_default and sysctl_rmem_default
> > per-namespace.
> > 
> > Signed-off-by: Menglong Dong <dong.menglong@xxxxxxxxxx>
> > ---  
> 
> Hey Menglong,
> 
> I was about to review the two patches you sent:
> 
> 1. [PATCH net-next] net: core: Namespace-ify sysctl_rmem_max and sysctl_wmem_max
>    https://lore.kernel.org/lkml/20210117104743.217194-1-dong.menglong@xxxxxxxxxx
> 2. [PATCH net-next] net: core: Namespace-ify sysctl_wmem_default and sysctl_rmem_default
>    https://lore.kernel.org/lkml/20210117102319.193756-1-dong.menglong@xxxxxxxxxx

And perhaps

  0. [PATCH net-next] net: core: init every ctl_table in netns_core_table

? 

I'm dropping these three from patchwork please follow Christian
suggestions on how to repost properly, thanks!

> and I had to spend some time figuring out that 2. is dependent on 1. I
> first thought I got the base wrong.
> 
> I'd suggest you resend both patches as a part of a single series with a
> cover letter mentioning the goal and use-case for these changes and also
> pass --base=<base-commit>
> when creating the patch series which makes it way easier to figure out
> what to apply it to when wanting to review a series in the larger
> context of a tree.