Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200818115712.36497-1-linmiaohe@huawei.com>
In-Reply-To: <20200818115712.36497-1-linmiaohe@huawei.com>
From:   Eric Dumazet <edumazet@google.com>
Date:   Tue, 18 Aug 2020 07:15:35 -0700
Message-ID: <CANn89iLuh-3OHDmRKKm3N59oHzC15K+k84KD42Ej+3qP_GC57Q@mail.gmail.com>
Subject: Re: [PATCH] net: Relax the npages test against MAX_SKB_FRAGS
To:     Miaohe Lin <linmiaohe@huawei.com>
Cc:     David Miller <davem@davemloft.net>,
        Jakub Kicinski <kuba@kernel.org>, martin.varghese@nokia.com,
        Florian Westphal <fw@strlen.de>,
        Pravin B Shelar <pshelar@ovn.org>,
        Davide Caratti <dcaratti@redhat.com>,
        Steffen Klassert <steffen.klassert@secunet.com>,
        Paolo Abeni <pabeni@redhat.com>, shmulik@metanetworks.com,
        kyk.segfault@gmail.com, netdev <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, Aug 18, 2020 at 4:58 AM Miaohe Lin <linmiaohe@huawei.com> wrote:
>
> The npages test against MAX_SKB_FRAGS can be relaxed if we succeed to
> allocate high order pages as the note in comment said.
>
> Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
> ---
>  net/core/skbuff.c | 11 ++++-------
>  1 file changed, 4 insertions(+), 7 deletions(-)
>
> diff --git a/net/core/skbuff.c b/net/core/skbuff.c
> index 2f7dd689bccc..ca432bbfd90b 100644
> --- a/net/core/skbuff.c
> +++ b/net/core/skbuff.c
> @@ -5758,13 +5758,6 @@ struct sk_buff *alloc_skb_with_frags(unsigned long header_len,
>         struct page *page;
>         int i;
>
> -       *errcode = -EMSGSIZE;
> -       /* Note this test could be relaxed, if we succeed to allocate
> -        * high order pages...
> -        */
> -       if (npages > MAX_SKB_FRAGS)
> -               return NULL;
> -
>         *errcode = -ENOBUFS;
>         skb = alloc_skb(header_len, gfp_mask);
>         if (!skb)
> @@ -5775,6 +5768,10 @@ struct sk_buff *alloc_skb_with_frags(unsigned long header_len,
>         for (i = 0; npages > 0; i++) {
>                 int order = max_page_order;
>
> +               if (unlikely(i >= MAX_SKB_FRAGS)) {
> +                       *errcode = -EMSGSIZE;
> +                       goto failure;
> +               }
>                 while (order) {
>                         if (npages >= 1 << order) {
>                                 page = alloc_pages((gfp_mask & ~__GFP_DIRECT_RECLAIM) |
> --
> 2.19.1


We do not want this change.

This interface is used by datagram providers, we do not want to claim
they can safely use skb allocations over 64KB.

Returning -EMSGSIZE should not depend on availability of high-order pages.

The comment was a hint, but we need first a valid user before
considering expanding the interface.