netdev - Re: [PULL] Networking for v6.2-rc1

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Wed, 21 Dec 2022 10:16:40 -0800
From:   Jakub Kicinski <kuba@...nel.org>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     davem@...emloft.net, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org, pabeni@...hat.com
Subject: Re: [PULL] Networking for v6.2-rc1

On Wed, 21 Dec 2022 09:16:35 -0800 Linus Torvalds wrote:
> On Tue, Dec 20, 2022 at 12:30 PM Jakub Kicinski <kuba@...nel.org> wrote:
> > Traffic is winding down significantly so let us pass our fixes to you
> > earlier than the usual Thu schedule.
> >
> > We have a fix for the BPF issue we were looking at before 6.1 final,
> > no surprises there. RxRPC fixes were merged relatively late so there's
> > an outpour of follow ups. Last one worth mentioning is the tree-wide
> > fix for network file systems / in-tree socket users, to prevent nested
> > networking calls from corrupting socket memory allocator.  
> 
> The  biggest changes seem to be to the intel 2.5Gb driver ("igc"), but
> they weren't mentioned...

Right, it was an odd is-it-a-fix-or-is-it-a-feature series,
if I'm completely honest I wasn't sure what to say about it.
It seemed safe to merge either way - it only changes the TSN parts
of igc, so the stuff very embedded users would care about.
It's somewhat interesting how Intel uses similar parts for embedded 
and client platforms. There are features which are supported on laptops
(due to being embedded-adjacent) which I'd really like to see in servers
but they are not there :S

> Also, maybe more people should look at this one:
> 
>    https://lore.kernel.org/all/6b971a4e-c7d8-411e-1f92-fda29b5b2fb9@kernel.org/
> 
> which seems to be a regression in 6.1 (and still present, afaik).

Ah, yes, the bhash2 conversion, the gift that keeps giving...
Kuniyuki seems to be on top of it:

https://lore.kernel.org/all/20221221151258.25748-1-kuniyu@amazon.com/

we should have a fix for rc2, rc3 worst case.

FWIW this is completely different to the previous python test problem.