lkml.org 
[lkml]   [2019]   [Feb]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH] xen-netback: fix occasional leak of grant ref mappings under memory pressure
On Thu, Feb 28, 2019 at 09:46:57AM +0000, Paul Durrant wrote:
> > -----Original Message-----
> > From: Igor Druzhinin [mailto:igor.druzhinin@citrix.com]
> > Sent: 28 February 2019 02:03
> > To: xen-devel@lists.xenproject.org; netdev@vger.kernel.org; linux-kernel@vger.kernel.org
> > Cc: Wei Liu <wei.liu2@citrix.com>; Paul Durrant <Paul.Durrant@citrix.com>; davem@davemloft.net; Igor
> > Druzhinin <igor.druzhinin@citrix.com>
> > Subject: [PATCH] xen-netback: fix occasional leak of grant ref mappings under memory pressure
> >
> > Zero-copy callback flag is not yet set on frag list skb at the moment
> > xenvif_handle_frag_list() returns -ENOMEM. This eventually results in
> > leaking grant ref mappings since xenvif_zerocopy_callback() is never
> > called for these fragments. Those eventually build up and cause Xen
> > to kill Dom0 as the slots get reused for new mappings.
> >
> > That behavior is observed under certain workloads where sudden spikes
> > of page cache usage for writes coexist with active atomic skb allocations.
> >
> > Signed-off-by: Igor Druzhinin <igor.druzhinin@citrix.com>
> > ---
> > drivers/net/xen-netback/netback.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c
> > index 80aae3a..2023317 100644
> > --- a/drivers/net/xen-netback/netback.c
> > +++ b/drivers/net/xen-netback/netback.c
> > @@ -1146,9 +1146,12 @@ static int xenvif_tx_submit(struct xenvif_queue *queue)
> >
> > if (unlikely(skb_has_frag_list(skb))) {
> > if (xenvif_handle_frag_list(queue, skb)) {
> > + struct sk_buff *nskb =
> > + skb_shinfo(skb)->frag_list;
> > if (net_ratelimit())
> > netdev_err(queue->vif->dev,
> > "Not enough memory to consolidate frag_list!\n");
> > + xenvif_skb_zerocopy_prepare(queue, nskb);
> > xenvif_skb_zerocopy_prepare(queue, skb);
> > kfree_skb(skb);
> > continue;
>
> Whilst this fix will do the job, I think it would be better to get rid of the kfree_skb() from inside xenvif_handle_frag_list() and always deal with it here rather than having it happen in two different places. Something like the following...

+1 for having only one place.

>
> ---8<---
> diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c
> index 80aae3a32c2a..093c7b860772 100644
> --- a/drivers/net/xen-netback/netback.c
> +++ b/drivers/net/xen-netback/netback.c
> @@ -1027,13 +1027,13 @@ static void xenvif_tx_build_gops(struct xenvif_queue *queue,
> /* Consolidate skb with a frag_list into a brand new one with local pages on
> * frags. Returns 0 or -ENOMEM if can't allocate new pages.
> */
> -static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *skb)
> +static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c
> index 80aae3a32c2a..093c7b860772 100644
> --- a/drivers/net/xen-netback/netback.c
> +++ b/drivers/net/xen-netback/netback.c
> @@ -1027,13 +1027,13 @@ static void xenvif_tx_build_gops(struct xenvif_queue *qu
> eue,
> /* Consolidate skb with a frag_list into a brand new one with local pages on
> * frags. Returns 0 or -ENOMEM if can't allocate new pages.
> */
> -static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *
> skb)
> +static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *
> skb,
> + struct sk_buff *nskb)
> {
> unsigned int offset = skb_headlen(skb);
> skb_frag_t frags[MAX_SKB_FRAGS];
> int i, f;
> struct ubuf_info *uarg;
> - struct sk_buff *nskb = skb_shinfo(skb)->frag_list;
>
> queue->stats.tx_zerocopy_sent += 2;
> queue->stats.tx_frag_overflow++;
> @@ -1072,11 +1072,6 @@ static int xenvif_handle_frag_list(struct xenvif_queue *q
> ueue, struct sk_buff *s
> skb_frag_size_set(&frags[i], len);
> }
>
> - /* Copied all the bits from the frag list -- free it. */
> - skb_frag_list_init(skb);
> - xenvif_skb_zerocopy_prepare(queue, nskb);
> - kfree_skb(nskb);
> -
> /* Release all the original (foreign) frags. */
> for (f = 0; f < skb_shinfo(skb)->nr_frags; f++)
> skb_frag_unref(skb, f);
> @@ -1145,7 +1140,11 @@ static int xenvif_tx_submit(struct xenvif_queue *queue)
> xenvif_fill_frags(queue, skb);
>
> if (unlikely(skb_has_frag_list(skb))) {
> - if (xenvif_handle_frag_list(queue, skb)) {
> + struct sk_buff *nskb = skb_shinfo(skb)->frag_list;
> +
> + xenvif_skb_zerocopy_prepare(queue, nskb);
> +
> + if (xenvif_handle_frag_list(queue, skb, nskb)) {
> if (net_ratelimit())
> netdev_err(queue->vif->dev,
> "Not enough memory to consolidate frag_list!\n");
> @@ -1153,6 +1152,10 @@ static int xenvif_tx_submit(struct xenvif_queue *queue)
> kfree_skb(skb);
> continue;
> }
> +
> + /* Copied all the bits from the frag list. */
> + skb_frag_list_init(skb);
> + kfree(nskb);

I think you want kfree_skb here?

Wei.

> }
>
> skb->dev = queue->vif->dev;
> ---8<---
>
> What do you think?
>
> Paul
>
> > --
> > 2.7.4
>

\
 
 \ /
  Last update: 2019-02-28 12:02    [W:0.287 / U:3.448 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site