netdev - Re: [PATCH net-next 4/4] net: add dedicated kmem

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

Message-ID: <CANn89iL+rJVTaqUL40YoP-0YGb8u0XZy+u4jKbxRxXmrgE3qJg@mail.gmail.com>
Date:   Fri, 3 Feb 2023 08:00:10 +0100
From:   Eric Dumazet <edumazet@...gle.com>
To:     Jakub Kicinski <kuba@...nel.org>
Cc:     "David S . Miller" <davem@...emloft.net>,
        Paolo Abeni <pabeni@...hat.com>, netdev@...r.kernel.org,
        eric.dumazet@...il.com, Alexander Duyck <alexanderduyck@...com>,
        Soheil Hassas Yeganeh <soheil@...gle.com>
Subject: Re: [PATCH net-next 4/4] net: add dedicated kmem_cache for
 typical/small skb->head

On Fri, Feb 3, 2023 at 6:11 AM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Thu,  2 Feb 2023 18:58:01 +0000 Eric Dumazet wrote:
> > +/* We want SKB_SMALL_HEAD_CACHE_SIZE to not be a power of two. */
>
> Why is that?  Is it to prevent potential mixing up of objects from
> the cache with objects from general slabs (since we only do a
> end_offset == SKB_SMALL_HEAD_HEADROOM check)?

Good question.

Some alloc_skb() callers use GFP_DMA (or __GFP_ACCOUNT)
we can not use the dedicated kmem_cache for them.

They could get an object of size 512 or 1024

Since I chose not adding yet another
skb->head_has_been_allocated_from_small_head_cache,
we want to make sure we will  not have issues in the future, if
SKB_HEAD_ALIGN(MAX_TCP_HEADER)
becomes a power-of-two. (for example for some of us increasing MAX_SKB_FRAGS)

Alternative would be to add a check at boot time, making sure
no standard cache has the same object size.

This might have an issue with CONFIG_SLOB=y, I wish this was gone already...