On Wed, Jul 24, 2024 at 06:48:33PM +0200, Larysa Zaremba wrote:
> The main threat to data consistency in ice_xdp() is a possible asynchronous
> PF reset. It can be triggered by a user or by TX timeout handler.
>
> XDP setup and PF reset code access the same resources in the following
> sections:
> * ice_vsi_close() in ice_prepare_for_reset() - already rtnl-locked
> * ice_vsi_rebuild() for the PF VSI - not protected
> * ice_vsi_open() - already rtnl-locked
>
> With an unfortunate timing, such accesses can result in a crash such as the
> one below:
>
> [ +1.999878] ice 0000:b1:00.0: Registered XDP mem model
> MEM_TYPE_XSK_BUFF_POOL on Rx ring 14
> [ +2.002992] ice 0000:b1:00.0: Registered XDP mem model
> MEM_TYPE_XSK_BUFF_POOL on Rx ring 18
> [Mar15 18:17] ice 0000:b1:00.0 ens801f0np0: NETDEV WATCHDOG: CPU: 38:
> transmit queue 14 timed out 80692736 ms
> [ +0.000093] ice 0000:b1:00.0 ens801f0np0: tx_timeout: VSI_num: 6, Q 14, NTC:
> 0x0, HW_HEAD: 0x0, NTU: 0x0, INT: 0x4000001
> [ +0.000012] ice 0000:b1:00.0 ens801f0np0: tx_timeout recovery level 1,
> txqueue 14
> [ +0.394718] ice 0000:b1:00.0: PTP reset successful
> [ +0.006184] BUG: kernel NULL pointer dereference, address: 0000000000000098
> [ +0.000045] #PF: supervisor read access in kernel mode
> [ +0.000023] #PF: error_code(0x0000) - not-present page
> [ +0.000023] PGD 0 P4D 0
> [ +0.000018] Oops: 0000 [#1] PREEMPT SMP NOPTI
> [ +0.000023] CPU: 38 PID: 7540 Comm: kworker/38:1 Not tainted 6.8.0-rc7 #1
> [ +0.000031] Hardware name: Intel Corporation S2600WFT/S2600WFT, BIOS
> SE5C620.86B.02.01.0014.082620210524 08/26/2021
> [ +0.000036] Workqueue: ice ice_service_task [ice]
> [ +0.000183] RIP: 0010:ice_clean_tx_ring+0xa/0xd0 [ice]
> [...]
> [ +0.000013] Call Trace:
> [ +0.000016] <TASK>
> [ +0.000014] ? __die+0x1f/0x70
> [ +0.000029] ? page_fault_oops+0x171/0x4f0
> [ +0.000029] ? schedule+0x3b/0xd0
> [ +0.000027] ? exc_page_fault+0x7b/0x180
> [ +0.000022] ? asm_exc_page_fault+0x22/0x30
> [ +0.000031] ? ice_clean_tx_ring+0xa/0xd0 [ice]
> [ +0.000194] ice_free_tx_ring+0xe/0x60 [ice]
> [ +0.000186] ice_destroy_xdp_rings+0x157/0x310 [ice]
> [ +0.000151] ice_vsi_decfg+0x53/0xe0 [ice]
> [ +0.000180] ice_vsi_rebuild+0x239/0x540 [ice]
> [ +0.000186] ice_vsi_rebuild_by_type+0x76/0x180 [ice]
> [ +0.000145] ice_rebuild+0x18c/0x840 [ice]
> [ +0.000145] ? delay_tsc+0x4a/0xc0
> [ +0.000022] ? delay_tsc+0x92/0xc0
> [ +0.000020] ice_do_reset+0x140/0x180 [ice]
> [ +0.000886] ice_service_task+0x404/0x1030 [ice]
> [ +0.000824] process_one_work+0x171/0x340
> [ +0.000685] worker_thread+0x277/0x3a0
> [ +0.000675] ? preempt_count_add+0x6a/0xa0
> [ +0.000677] ? _raw_spin_lock_irqsave+0x23/0x50
> [ +0.000679] ? __pfx_worker_thread+0x10/0x10
> [ +0.000653] kthread+0xf0/0x120
> [ +0.000635] ? __pfx_kthread+0x10/0x10
> [ +0.000616] ret_from_fork+0x2d/0x50
> [ +0.000612] ? __pfx_kthread+0x10/0x10
> [ +0.000604] ret_from_fork_asm+0x1b/0x30
> [ +0.000604] </TASK>
>
> The previous way of handling this through returning -EBUSY is not viable,
> particularly when destroying AF_XDP socket, because the kernel proceeds
> with removal anyway.
>
> There is plenty of code between those calls and there is no need to create
> a large critical section that covers all of them, same as there is no need
> to protect ice_vsi_rebuild() with rtnl_lock().
>
> Add xdp_state_lock mutex to protect ice_vsi_rebuild() and ice_xdp().
>
> Leaving unprotected sections in between would result in two states that
> have to be considered:
> 1. when the VSI is closed, but not yet rebuild
> 2. when VSI is already rebuild, but not yet open
>
> The latter case is actually already handled through !netif_running() case,
> we just need to adjust flag checking a little. The former one is not as
> trivial, because between ice_vsi_close() and ice_vsi_rebuild(), a lot of
> hardware interaction happens, this can make adding/deleting rings exit
> with an error. Luckily, VSI rebuild is pending and can apply new
> configuration for us in a managed fashion.
>
> Therefore, add an additional VSI state flag ICE_VSI_REBUILD_PENDING to
> indicate that ice_xdp() can just hot-swap the program.
couldn't this be a separate patch?
>
> Fixes: 2d4238f55697 ("ice: Add support for AF_XDP")
> Fixes: efc2214b6047 ("ice: Add support for XDP")
> Reviewed-by: Wojciech Drewek <wojciech.dre...@intel.com>
> Signed-off-by: Larysa Zaremba <larysa.zare...@intel.com>
> ---
> drivers/net/ethernet/intel/ice/ice.h | 2 ++
> drivers/net/ethernet/intel/ice/ice_lib.c | 26 +++++++++++++++--------
> drivers/net/ethernet/intel/ice/ice_main.c | 19 ++++++++++++-----
> drivers/net/ethernet/intel/ice/ice_xsk.c | 3 ++-
> 4 files changed, 35 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/net/ethernet/intel/ice/ice.h
> b/drivers/net/ethernet/intel/ice/ice.h
> index 99a75a59078e..3d7a0abc13ab 100644
> --- a/drivers/net/ethernet/intel/ice/ice.h
> +++ b/drivers/net/ethernet/intel/ice/ice.h
> @@ -318,6 +318,7 @@ enum ice_vsi_state {
> ICE_VSI_UMAC_FLTR_CHANGED,
> ICE_VSI_MMAC_FLTR_CHANGED,
> ICE_VSI_PROMISC_CHANGED,
> + ICE_VSI_REBUILD_PENDING,
> ICE_VSI_STATE_NBITS /* must be last */
> };
>
> @@ -411,6 +412,7 @@ struct ice_vsi {
> struct ice_tx_ring **xdp_rings; /* XDP ring array */
> u16 num_xdp_txq; /* Used XDP queues */
> u8 xdp_mapping_mode; /* ICE_MAP_MODE_[CONTIG|SCATTER] */
> + struct mutex xdp_state_lock;
>
> struct net_device **target_netdevs;
>
> diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c
> b/drivers/net/ethernet/intel/ice/ice_lib.c
> index 5f2ddcaf7031..bbef5ec67cae 100644
> --- a/drivers/net/ethernet/intel/ice/ice_lib.c
> +++ b/drivers/net/ethernet/intel/ice/ice_lib.c
> @@ -447,6 +447,7 @@ static void ice_vsi_free(struct ice_vsi *vsi)
>
> ice_vsi_free_stats(vsi);
> ice_vsi_free_arrays(vsi);
> + mutex_destroy(&vsi->xdp_state_lock);
> mutex_unlock(&pf->sw_mutex);
> devm_kfree(dev, vsi);
> }
> @@ -626,6 +627,8 @@ static struct ice_vsi *ice_vsi_alloc(struct ice_pf *pf)
> pf->next_vsi = ice_get_free_slot(pf->vsi, pf->num_alloc_vsi,
> pf->next_vsi);
>
> + mutex_init(&vsi->xdp_state_lock);
> +
> unlock_pf:
> mutex_unlock(&pf->sw_mutex);
> return vsi;
> @@ -2973,19 +2976,24 @@ int ice_vsi_rebuild(struct ice_vsi *vsi, u32
> vsi_flags)
> if (WARN_ON(vsi->type == ICE_VSI_VF && !vsi->vf))
> return -EINVAL;
>
> + mutex_lock(&vsi->xdp_state_lock);
> + clear_bit(ICE_VSI_REBUILD_PENDING, vsi->state);
> +
> ret = ice_vsi_realloc_stat_arrays(vsi);
> if (ret)
> - goto err_vsi_cfg;
> + goto unlock;
>
> ice_vsi_decfg(vsi);
> ret = ice_vsi_cfg_def(vsi);
> if (ret)
> - goto err_vsi_cfg;
> + goto unlock;
>
> coalesce = kcalloc(vsi->num_q_vectors,
> sizeof(struct ice_coalesce_stored), GFP_KERNEL);
> - if (!coalesce)
> - return -ENOMEM;
> + if (!coalesce) {
> + ret = -ENOMEM;
Knee-jerk reaction would be to deconfig things that ice_vsi_cfg_def()
setup above.
So I think the order of kfree and ice_vsi_decfg should be swapped,
something like:
if (!coalesce) {
ret = -ENOMEM;
goto err_mem_alloc;
}
err_vsi_cfg_tc_lan:
kfree(coalesce);
err_mem_alloc:
ice_vsi_decfg(vsi);
unlock:
mutex_unlock(&vsi->xdp_state_lock);
return ret;
or am I missing something?
> + goto unlock;
> + }
>
> prev_num_q_vectors = ice_vsi_rebuild_get_coalesce(vsi, coalesce);
>
> @@ -2996,19 +3004,19 @@ int ice_vsi_rebuild(struct ice_vsi *vsi, u32
> vsi_flags)
> goto err_vsi_cfg_tc_lan;
> }
>
> - kfree(coalesce);
> - return ice_schedule_reset(pf, ICE_RESET_PFR);
> + ret = ice_schedule_reset(pf, ICE_RESET_PFR);
> + goto err_vsi_cfg_tc_lan;
> }
>
> ice_vsi_rebuild_set_coalesce(vsi, coalesce, prev_num_q_vectors);
> kfree(coalesce);
> -
> - return 0;
> + goto unlock;
>
> err_vsi_cfg_tc_lan:
> ice_vsi_decfg(vsi);
> kfree(coalesce);
> -err_vsi_cfg:
> +unlock:
> + mutex_unlock(&vsi->xdp_state_lock);
> return ret;
> }
>
> diff --git a/drivers/net/ethernet/intel/ice/ice_main.c
> b/drivers/net/ethernet/intel/ice/ice_main.c
> index 8ed1798bb06e..e50526b491fc 100644
> --- a/drivers/net/ethernet/intel/ice/ice_main.c
> +++ b/drivers/net/ethernet/intel/ice/ice_main.c
> @@ -611,6 +611,7 @@ ice_prepare_for_reset(struct ice_pf *pf, enum
> ice_reset_req reset_type)
> /* clear SW filtering DB */
> ice_clear_hw_tbls(hw);
> /* disable the VSIs and their queues that are not already DOWN */
> + set_bit(ICE_VSI_REBUILD_PENDING, ice_get_main_vsi(pf)->state);
> ice_pf_dis_all_vsi(pf, false);
>
> if (test_bit(ICE_FLAG_PTP_SUPPORTED, pf->flags))
> @@ -3011,7 +3012,8 @@ ice_xdp_setup_prog(struct ice_vsi *vsi, struct bpf_prog
> *prog,
> }
>
> /* hot swap progs and avoid toggling link */
> - if (ice_is_xdp_ena_vsi(vsi) == !!prog) {
> + if (ice_is_xdp_ena_vsi(vsi) == !!prog ||
> + test_bit(ICE_VSI_REBUILD_PENDING, vsi->state)) {
> ice_vsi_assign_bpf_prog(vsi, prog);
> return 0;
> }
> @@ -3083,21 +3085,28 @@ static int ice_xdp(struct net_device *dev, struct
> netdev_bpf *xdp)
> {
> struct ice_netdev_priv *np = netdev_priv(dev);
> struct ice_vsi *vsi = np->vsi;
> + int ret;
>
> if (vsi->type != ICE_VSI_PF) {
> NL_SET_ERR_MSG_MOD(xdp->extack, "XDP can be loaded only on PF
> VSI");
> return -EINVAL;
> }
>
> + mutex_lock(&vsi->xdp_state_lock);
> +
> switch (xdp->command) {
> case XDP_SETUP_PROG:
> - return ice_xdp_setup_prog(vsi, xdp->prog, xdp->extack);
> + ret = ice_xdp_setup_prog(vsi, xdp->prog, xdp->extack);
> + break;
> case XDP_SETUP_XSK_POOL:
> - return ice_xsk_pool_setup(vsi, xdp->xsk.pool,
> - xdp->xsk.queue_id);
> + ret = ice_xsk_pool_setup(vsi, xdp->xsk.pool, xdp->xsk.queue_id);
> + break;
> default:
> - return -EINVAL;
> + ret = -EINVAL;
> }
> +
> + mutex_unlock(&vsi->xdp_state_lock);
> + return ret;
> }
>
> /**
> diff --git a/drivers/net/ethernet/intel/ice/ice_xsk.c
> b/drivers/net/ethernet/intel/ice/ice_xsk.c
> index a65955eb23c0..2c1a843ba200 100644
> --- a/drivers/net/ethernet/intel/ice/ice_xsk.c
> +++ b/drivers/net/ethernet/intel/ice/ice_xsk.c
> @@ -379,7 +379,8 @@ int ice_xsk_pool_setup(struct ice_vsi *vsi, struct
> xsk_buff_pool *pool, u16 qid)
> goto failure;
> }
>
> - if_running = netif_running(vsi->netdev) && ice_is_xdp_ena_vsi(vsi);
> + if_running = !test_bit(ICE_VSI_DOWN, vsi->state) &&
> + ice_is_xdp_ena_vsi(vsi);
>
> if (if_running) {
> struct ice_rx_ring *rx_ring = vsi->rx_rings[qid];
> --
> 2.43.0
>
