mbox series

[net,0/6,pull,request] ice: fix synchronization between .ndo_bpf() and reset

Message ID 20240903183034.3530411-1-anthony.l.nguyen@intel.com (mailing list archive)
Headers show
Series ice: fix synchronization between .ndo_bpf() and reset | expand

Message

Tony Nguyen Sept. 3, 2024, 6:30 p.m. UTC 
Larysa Zaremba says:

PF reset can be triggered asynchronously, by tx_timeout or by a user. With some
unfortunate timings both ice_vsi_rebuild() and .ndo_bpf will try to access and
modify XDP rings at the same time, causing system crash.

The first patch factors out rtnl-locked code from VSI rebuild code to avoid
deadlock. The following changes lock rebuild and .ndo_bpf() critical sections
with an internal mutex as well and provide complementary fixes.
---
IWL: https://lore.kernel.org/intel-wired-lan/20240823095933.17922-1-larysa.zaremba@intel.com/

The following are changes since commit cfd433cecef929b4d92685f570f1a480762ec260:
  Merge branch 'ptp-ocp-fix-serial-port-information-export'
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue 100GbE

Larysa Zaremba (6):
  ice: move netif_queue_set_napi to rtnl-protected sections
  ice: protect XDP configuration with a mutex
  ice: check for XDP rings instead of bpf program when unconfiguring
  ice: check ICE_VSI_DOWN under rtnl_lock when preparing for reset
  ice: remove ICE_CFG_BUSY locking from AF_XDP code
  ice: do not bring the VSI up, if it was down before the XDP setup

 drivers/net/ethernet/intel/ice/ice.h      |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c |  11 +-
 drivers/net/ethernet/intel/ice/ice_lib.c  | 179 ++++++++--------------
 drivers/net/ethernet/intel/ice/ice_lib.h  |  10 +-
 drivers/net/ethernet/intel/ice/ice_main.c |  47 ++++--
 drivers/net/ethernet/intel/ice/ice_xsk.c  |  18 +--
 6 files changed, 106 insertions(+), 161 deletions(-)

Comments

patchwork-bot+netdevbpf@kernel.org Sept. 5, 2024, 2:10 a.m. UTC | #1 
Hello:

This series was applied to netdev/net.git (main)
by Tony Nguyen <anthony.l.nguyen@intel.com>:

On Tue,  3 Sep 2024 11:30:26 -0700 you wrote:
> Larysa Zaremba says:
> 
> PF reset can be triggered asynchronously, by tx_timeout or by a user. With some
> unfortunate timings both ice_vsi_rebuild() and .ndo_bpf will try to access and
> modify XDP rings at the same time, causing system crash.
> 
> The first patch factors out rtnl-locked code from VSI rebuild code to avoid
> deadlock. The following changes lock rebuild and .ndo_bpf() critical sections
> with an internal mutex as well and provide complementary fixes.
> 
> [...]

Here is the summary with links:
  - [net,1/6] ice: move netif_queue_set_napi to rtnl-protected sections
    https://git.kernel.org/netdev/net/c/2a5dc090b92c
  - [net,2/6] ice: protect XDP configuration with a mutex
    https://git.kernel.org/netdev/net/c/2504b8405768
  - [net,3/6] ice: check for XDP rings instead of bpf program when unconfiguring
    https://git.kernel.org/netdev/net/c/f50c68763436
  - [net,4/6] ice: check ICE_VSI_DOWN under rtnl_lock when preparing for reset
    https://git.kernel.org/netdev/net/c/d8c40b9d3a6c
  - [net,5/6] ice: remove ICE_CFG_BUSY locking from AF_XDP code
    https://git.kernel.org/netdev/net/c/7e3b407ccbea
  - [net,6/6] ice: do not bring the VSI up, if it was down before the XDP setup
    https://git.kernel.org/netdev/net/c/04c7e14e5b0b

You are awesome, thank you!