Received-SPF: pass (google.com: domain of linux-kernel+bounces-205604-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Precedence: bulk
MIME-Version: 1.0
References: <20240607023116.1720640-1-zhaoyang.huang@unisoc.com>
In-Reply-To: <20240607023116.1720640-1-zhaoyang.huang@unisoc.com>
From: Zhaoyang Huang <huangzhaoyang@gmail.com>
Date: Fri, 7 Jun 2024 16:30:41 +0800
Message-ID: <CAGWkznEODMbDngM3toQFo-bgkezEpmXf_qE=SpuYcqsjEJk1DQ@mail.gmail.com>
Subject: Re: [Resend PATCHv4 1/1] mm: fix incorrect vbq reference in purge_fragmented_block
To: "zhaoyang.huang" <zhaoyang.huang@unisoc.com>
Cc: Andrew Morton <akpm@linux-foundation.org>, Uladzislau Rezki <urezki@gmail.com>, 
	Christoph Hellwig <hch@infradead.org>, Lorenzo Stoakes <lstoakes@gmail.com>, Baoquan He <bhe@redhat.com>, 
	Thomas Gleixner <tglx@linutronix.de>, hailong liu <hailong.liu@oppo.com>, linux-mm@kvack.org, 
	linux-kernel@vger.kernel.org, stable@vger.kernel.org, steve.kang@unisoc.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Patchv4 was updated based on Hailong and Uladzislau's comments, where
vbq is obtained from vb->cpu, thus avoiding disabling preemption.
Furthermore, Baoquan's suggestion was not adopted because it made vbq
accesses completely interleaved across all CPUs, which defeats the
goal of per_cpu.

On Fri, Jun 7, 2024 at 10:31=E2=80=AFAM zhaoyang.huang
<zhaoyang.huang@unisoc.com> wrote:
>
> From: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
>
> vmalloc area runs out in our ARM64 system during an erofs test as
> vm_map_ram failed[1]. By following the debug log, we find that
> vm_map_ram()->vb_alloc() will allocate new vb->va which corresponding
> to 4MB vmalloc area as list_for_each_entry_rcu returns immediately
> when vbq->free->next points to vbq->free. That is to say, 65536 times
> of page fault after the list's broken will run out of the whole
> vmalloc area. This should be introduced by one vbq->free->next point to
> vbq->free which makes list_for_each_entry_rcu can not iterate the list
> and find the BUG.
>
> [1]
> PID: 1        TASK: ffffff80802b4e00  CPU: 6    COMMAND: "init"
>  #0 [ffffffc08006afe0] __switch_to at ffffffc08111d5cc
>  #1 [ffffffc08006b040] __schedule at ffffffc08111dde0
>  #2 [ffffffc08006b0a0] schedule at ffffffc08111e294
>  #3 [ffffffc08006b0d0] schedule_preempt_disabled at ffffffc08111e3f0
>  #4 [ffffffc08006b140] __mutex_lock at ffffffc08112068c
>  #5 [ffffffc08006b180] __mutex_lock_slowpath at ffffffc08111f8f8
>  #6 [ffffffc08006b1a0] mutex_lock at ffffffc08111f834
>  #7 [ffffffc08006b1d0] reclaim_and_purge_vmap_areas at ffffffc0803ebc3c
>  #8 [ffffffc08006b290] alloc_vmap_area at ffffffc0803e83fc
>  #9 [ffffffc08006b300] vm_map_ram at ffffffc0803e78c0
>
> Fixes: fc1e0d980037 ("mm/vmalloc: prevent stale TLBs in fully utilized bl=
ocks")
>
> For detailed reason of broken list, please refer to below URL
> https://lore.kernel.org/all/20240531024820.5507-1-hailong.liu@oppo.com/
>
> Suggested-by: Hailong.Liu <hailong.liu@oppo.com>
> Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
> ---
> v2: introduce cpu in vmap_block to record the right CPU number
> v3: use get_cpu/put_cpu to prevent schedule between core
> v4: replace get_cpu/put_cpu by another API to avoid disabling preemption
> ---
> ---
>  mm/vmalloc.c | 21 +++++++++++++++------
>  1 file changed, 15 insertions(+), 6 deletions(-)
>
> diff --git a/mm/vmalloc.c b/mm/vmalloc.c
> index 22aa63f4ef63..89eb034f4ac6 100644
> --- a/mm/vmalloc.c
> +++ b/mm/vmalloc.c
> @@ -2458,6 +2458,7 @@ struct vmap_block {
>         struct list_head free_list;
>         struct rcu_head rcu_head;
>         struct list_head purge;
> +       unsigned int cpu;
>  };
>
>  /* Queue of free and dirty vmap blocks, for allocation and flushing purp=
oses */
> @@ -2585,8 +2586,15 @@ static void *new_vmap_block(unsigned int order, gf=
p_t gfp_mask)
>                 free_vmap_area(va);
>                 return ERR_PTR(err);
>         }
> -
> -       vbq =3D raw_cpu_ptr(&vmap_block_queue);
> +       /*
> +        * list_add_tail_rcu could happened in another core
> +        * rather than vb->cpu due to task migration, which
> +        * is safe as list_add_tail_rcu will ensure the list's
> +        * integrity together with list_for_each_rcu from read
> +        * side.
> +        */
> +       vb->cpu =3D raw_smp_processor_id();
> +       vbq =3D per_cpu_ptr(&vmap_block_queue, vb->cpu);
>         spin_lock(&vbq->lock);
>         list_add_tail_rcu(&vb->free_list, &vbq->free);
>         spin_unlock(&vbq->lock);
> @@ -2614,9 +2622,10 @@ static void free_vmap_block(struct vmap_block *vb)
>  }
>
>  static bool purge_fragmented_block(struct vmap_block *vb,
> -               struct vmap_block_queue *vbq, struct list_head *purge_lis=
t,
> -               bool force_purge)
> +               struct list_head *purge_list, bool force_purge)
>  {
> +       struct vmap_block_queue *vbq =3D &per_cpu(vmap_block_queue, vb->c=
pu);
> +
>         if (vb->free + vb->dirty !=3D VMAP_BBMAP_BITS ||
>             vb->dirty =3D=3D VMAP_BBMAP_BITS)
>                 return false;
> @@ -2664,7 +2673,7 @@ static void purge_fragmented_blocks(int cpu)
>                         continue;
>
>                 spin_lock(&vb->lock);
> -               purge_fragmented_block(vb, vbq, &purge, true);
> +               purge_fragmented_block(vb, &purge, true);
>                 spin_unlock(&vb->lock);
>         }
>         rcu_read_unlock();
> @@ -2801,7 +2810,7 @@ static void _vm_unmap_aliases(unsigned long start, =
unsigned long end, int flush)
>                          * not purgeable, check whether there is dirty
>                          * space to be flushed.
>                          */
> -                       if (!purge_fragmented_block(vb, vbq, &purge_list,=
 false) &&
> +                       if (!purge_fragmented_block(vb, &purge_list, fals=
e) &&
>                             vb->dirty_max && vb->dirty !=3D VMAP_BBMAP_BI=
TS) {
>                                 unsigned long va_start =3D vb->va->va_sta=
rt;
>                                 unsigned long s, e;
> --
> 2.25.1
>