Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1400736 > unrolled thread

[PATCH net-next v2 0/6] hv_netvsc: avoid races on mtu change/set channels

Started byVitaly Kuznetsov <vkuznets@redhat.com>
First post2016-05-13 14:00 +0200
Last post2016-05-16 19:30 +0200
Articles 2 — 2 participants

Back to article view | Back to linux.kernel


Contents

  [PATCH net-next v2 0/6] hv_netvsc: avoid races on mtu change/set channels Vitaly Kuznetsov <vkuznets@redhat.com> - 2016-05-13 14:00 +0200
    Re: [PATCH net-next v2 0/6] hv_netvsc: avoid races on mtu  change/set channels David Miller <davem@davemloft.net> - 2016-05-16 19:30 +0200

#1400736 — [PATCH net-next v2 0/6] hv_netvsc: avoid races on mtu change/set channels

FromVitaly Kuznetsov <vkuznets@redhat.com>
Date2016-05-13 14:00 +0200
Subject[PATCH net-next v2 0/6] hv_netvsc: avoid races on mtu change/set channels
Message-ID<ryjV7-2yY-9@gated-at.bofh.it>
Changes since v1:
- Rebased to net-next [Haiyang Zhang]

Original description:

MTU change and set channels operations are implemented as netvsc device
re-creation destroying internal structures (struct net_device stays). This
is really unfortunate but there is no support from Hyper-V host to do it
in a different way. Such re-creation is unsurprisingly racy, Haiyang
reported a crash when netvsc_change_mtu() is racing with
netvsc_link_change() but I was able to identify additional races upon
investigation. Both netvsc_set_channels() and netvsc_change_mtu() race
against:
1) netvsc_link_change()
2) netvsc_remove()
3) netvsc_send()

To solve these issues without introducing new locks some refactoring is
required. We need to get rid of very complex link graph in all the
internal structures and avoid traveling through structures which are being
removed.

Vitaly Kuznetsov (6):
  hv_netvsc: move start_remove flag to net_device_context
  hv_netvsc: use start_remove flag to protect netvsc_link_change()
  hv_netvsc: untangle the pointer mess
  hv_netvsc: get rid of struct net_device pointer in struct
    netvsc_device
  hv_netvsc: synchronize netvsc_change_mtu()/netvsc_set_channels() with
    netvsc_remove()
  hv_netvsc: set nvdev link after populating chn_table

 drivers/net/hyperv/hyperv_net.h   |  17 ++--
 drivers/net/hyperv/netvsc.c       | 136 +++++++++++++----------------
 drivers/net/hyperv/netvsc_drv.c   | 178 ++++++++++++++++++++++----------------
 drivers/net/hyperv/rndis_filter.c |  82 +++++++++---------
 4 files changed, 215 insertions(+), 198 deletions(-)

-- 
2.5.5

[toc] | [next] | [standalone]


#1401657 — Re: [PATCH net-next v2 0/6] hv_netvsc: avoid races on mtu change/set channels

FromDavid Miller <davem@davemloft.net>
Date2016-05-16 19:30 +0200
SubjectRe: [PATCH net-next v2 0/6] hv_netvsc: avoid races on mtu change/set channels
Message-ID<rzuve-1tB-37@gated-at.bofh.it>
In reply to#1400736
From: Vitaly Kuznetsov <vkuznets@redhat.com>
Date: Fri, 13 May 2016 13:55:19 +0200

> Changes since v1:
> - Rebased to net-next [Haiyang Zhang]
> 
> Original description:
> 
> MTU change and set channels operations are implemented as netvsc device
> re-creation destroying internal structures (struct net_device stays). This
> is really unfortunate but there is no support from Hyper-V host to do it
> in a different way. Such re-creation is unsurprisingly racy, Haiyang
> reported a crash when netvsc_change_mtu() is racing with
> netvsc_link_change() but I was able to identify additional races upon
> investigation. Both netvsc_set_channels() and netvsc_change_mtu() race
> against:
> 1) netvsc_link_change()
> 2) netvsc_remove()
> 3) netvsc_send()
> 
> To solve these issues without introducing new locks some refactoring is
> required. We need to get rid of very complex link graph in all the
> internal structures and avoid traveling through structures which are being
> removed.

Series applied, thanks.

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web