Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230817-free_pcppages_bulk-v1-0-c14574a9f80c@kernel.org>
 <20230821103225.qntnsotdzuthxn2y@techsingularity.net> <6a18b295-92ac-1633-0b79-6aaf18a8496d@huaweicloud.com>
In-Reply-To: <6a18b295-92ac-1633-0b79-6aaf18a8496d@huaweicloud.com>
From:   Chris Li <chrisl@kernel.org>
Date:   Tue, 22 Aug 2023 14:14:56 -0700
Message-ID: <CAF8kJuNq2Y5AAAQn66bTEfjkQ8kfPokmGgBLDPwEWUC=MO9NcQ@mail.gmail.com>
Subject: Re: [PATCH RFC 0/2] mm/page_alloc: free_pcppages_bulk safeguard
To:     Kemeng Shi <shikemeng@huaweicloud.com>
Cc:     Mel Gorman <mgorman@techsingularity.net>,
        Andrew Morton <akpm@linux-foundation.org>,
        baolin.wang@linux.alibaba.com, Michal Hocko <mhocko@suse.com>,
        david@redhat.com, Matthew Wilcox <willy@infradead.org>,
        linux-mm@kvack.org, Namhyung Kim <namhyung@google.com>,
        Greg Thelen <gthelen@google.com>,
        LKML <linux-kernel@vger.kernel.org>,
        John Sperbeck <jsperbeck@google.com>,
        Huang Ying <ying.huang@intel.com>,
        Alexei Starovoitov <ast@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hi Kemeng,

On Mon, Aug 21, 2023 at 6:27=E2=80=AFPM Kemeng Shi <shikemeng@huaweicloud.c=
om> wrote:
> >
> Agreed. We assume pcp->count is protected by pcp->lock. Instead of make c=
ode
> work in case pcp->count could be changed without lock held, it's more rea=
sonble
> to modify pcp->count with pcp->lock held in BPF program.

The lock is holded when pcp->count is modified. It is going through
the kernel page
allocation API. The issue is nest memory allocation inside spin_lock()
introduced by BPF.

The execution sequence is like this:

       count =3D min(pcp->count, count);

        /* Ensure requested pindex is drained first. */
        pindex =3D pindex - 1;
        bpf_injected_spin_lock_irqsave {
                 alloc_page();
                 original spin_lock_irqsave(&zone->lock, flags) ;
        }

Chris


Chris