| Message ID | 20240724134024.2182959-4-aleksander.lobakin@intel.com |
|---|---|
| State | Accepted |
| Delegated to: | Anthony Nguyen |
| Headers | show |
| Series | idpf: fix 3 bugs revealed by the Chapter I | expand |
On Wed, Jul 24, 2024 at 03:40:24PM +0200, Alexander Lobakin wrote: > The second tagged commit started sometimes (very rarely, but possible) > throwing WARNs from > net/core/page_pool.c:page_pool_disable_direct_recycling(). > Turned out idpf frees interrupt vectors with embedded NAPIs *before* > freeing the queues making page_pools' NAPI pointers lead to freed > memory before these pools are destroyed by libeth. > It's not clear whether there are other accesses to the freed vectors > when destroying the queues, but anyway, we usually free queue/interrupt > vectors only when the queues are destroyed and the NAPIs are guaranteed > to not be referenced anywhere. > > Invert the allocation and freeing logic making queue/interrupt vectors > be allocated first and freed last. Vectors don't require queues to be > present, so this is safe. Additionally, this change allows to remove > that useless queue->q_vector pointer cleanup, as vectors are still > valid when freeing the queues (+ both are freed within one function, > so it's not clear why nullify the pointers at all). > > Fixes: 1c325aac10a8 ("idpf: configure resources for TX queues") > Fixes: 90912f9f4f2d ("idpf: convert header split mode to libeth + napi_build_skb()") > Reported-by: Michal Kubiak <michal.kubiak@intel.com> > Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com> Reviewed-by: Simon Horman <horms@kernel.org>
> -----Original Message----- > From: Simon Horman <horms@kernel.org> > Sent: Friday, July 26, 2024 9:22 AM > To: Lobakin, Aleksander <aleksander.lobakin@intel.com> > Cc: intel-wired-lan@lists.osuosl.org; Nguyen, Anthony L > <anthony.l.nguyen@intel.com>; David S. Miller <davem@davemloft.net>; Eric > Dumazet <edumazet@google.com>; Jakub Kicinski <kuba@kernel.org>; Paolo > Abeni <pabeni@redhat.com>; NEX SW NCIS OSDT ITP Upstreaming > <nex.sw.ncis.osdt.itp.upstreaming@intel.com>; netdev@vger.kernel.org; linux- > kernel@vger.kernel.org; Kubiak, Michal <michal.kubiak@intel.com> > Subject: Re: [PATCH iwl-net 3/3] idpf: fix UAFs when destroying the queues > > On Wed, Jul 24, 2024 at 03:40:24PM +0200, Alexander Lobakin wrote: > > The second tagged commit started sometimes (very rarely, but possible) > > throwing WARNs from > > net/core/page_pool.c:page_pool_disable_direct_recycling(). > > Turned out idpf frees interrupt vectors with embedded NAPIs *before* > > freeing the queues making page_pools' NAPI pointers lead to freed > > memory before these pools are destroyed by libeth. > > It's not clear whether there are other accesses to the freed vectors > > when destroying the queues, but anyway, we usually free queue/interrupt > > vectors only when the queues are destroyed and the NAPIs are guaranteed > > to not be referenced anywhere. > > > > Invert the allocation and freeing logic making queue/interrupt vectors > > be allocated first and freed last. Vectors don't require queues to be > > present, so this is safe. Additionally, this change allows to remove > > that useless queue->q_vector pointer cleanup, as vectors are still > > valid when freeing the queues (+ both are freed within one function, > > so it's not clear why nullify the pointers at all). > > > > Fixes: 1c325aac10a8 ("idpf: configure resources for TX queues") > > Fixes: 90912f9f4f2d ("idpf: convert header split mode to libeth + > napi_build_skb()") > > Reported-by: Michal Kubiak <michal.kubiak@intel.com> > > Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com> > > Reviewed-by: Simon Horman <horms@kernel.org> > Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
diff --git a/drivers/net/ethernet/intel/idpf/idpf_lib.c b/drivers/net/ethernet/intel/idpf/idpf_lib.c index 10b884dd3475..0b6c8fd5bc90 100644 --- a/drivers/net/ethernet/intel/idpf/idpf_lib.c +++ b/drivers/net/ethernet/intel/idpf/idpf_lib.c @@ -900,8 +900,8 @@ static void idpf_vport_stop(struct idpf_vport *vport) vport->link_up = false; idpf_vport_intr_deinit(vport); - idpf_vport_intr_rel(vport); idpf_vport_queues_rel(vport); + idpf_vport_intr_rel(vport); np->state = __IDPF_VPORT_DOWN; } @@ -1349,43 +1349,43 @@ static int idpf_vport_open(struct idpf_vport *vport) /* we do not allow interface up just yet */ netif_carrier_off(vport->netdev); - err = idpf_vport_queues_alloc(vport); - if (err) - return err; - err = idpf_vport_intr_alloc(vport); if (err) { dev_err(&adapter->pdev->dev, "Failed to allocate interrupts for vport %u: %d\n", vport->vport_id, err); - goto queues_rel; + return err; } + err = idpf_vport_queues_alloc(vport); + if (err) + goto intr_rel; + err = idpf_vport_queue_ids_init(vport); if (err) { dev_err(&adapter->pdev->dev, "Failed to initialize queue ids for vport %u: %d\n", vport->vport_id, err); - goto intr_rel; + goto queues_rel; } err = idpf_vport_intr_init(vport); if (err) { dev_err(&adapter->pdev->dev, "Failed to initialize interrupts for vport %u: %d\n", vport->vport_id, err); - goto intr_rel; + goto queues_rel; } err = idpf_rx_bufs_init_all(vport); if (err) { dev_err(&adapter->pdev->dev, "Failed to initialize RX buffers for vport %u: %d\n", vport->vport_id, err); - goto intr_rel; + goto queues_rel; } err = idpf_queue_reg_init(vport); if (err) { dev_err(&adapter->pdev->dev, "Failed to initialize queue registers for vport %u: %d\n", vport->vport_id, err); - goto intr_rel; + goto queues_rel; } idpf_rx_init_buf_tail(vport); @@ -1452,10 +1452,10 @@ static int idpf_vport_open(struct idpf_vport *vport) idpf_send_map_unmap_queue_vector_msg(vport, false); intr_deinit: idpf_vport_intr_deinit(vport); -intr_rel: - idpf_vport_intr_rel(vport); queues_rel: idpf_vport_queues_rel(vport); +intr_rel: + idpf_vport_intr_rel(vport); return err; } diff --git a/drivers/net/ethernet/intel/idpf/idpf_txrx.c b/drivers/net/ethernet/intel/idpf/idpf_txrx.c index a2f9f252694a..585c3dadd9bf 100644 --- a/drivers/net/ethernet/intel/idpf/idpf_txrx.c +++ b/drivers/net/ethernet/intel/idpf/idpf_txrx.c @@ -3576,9 +3576,7 @@ static void idpf_vport_intr_napi_dis_all(struct idpf_vport *vport) */ void idpf_vport_intr_rel(struct idpf_vport *vport) { - int i, j, v_idx; - - for (v_idx = 0; v_idx < vport->num_q_vectors; v_idx++) { + for (u32 v_idx = 0; v_idx < vport->num_q_vectors; v_idx++) { struct idpf_q_vector *q_vector = &vport->q_vectors[v_idx]; kfree(q_vector->complq); @@ -3593,26 +3591,6 @@ void idpf_vport_intr_rel(struct idpf_vport *vport) free_cpumask_var(q_vector->affinity_mask); } - /* Clean up the mapping of queues to vectors */ - for (i = 0; i < vport->num_rxq_grp; i++) { - struct idpf_rxq_group *rx_qgrp = &vport->rxq_grps[i]; - - if (idpf_is_queue_model_split(vport->rxq_model)) - for (j = 0; j < rx_qgrp->splitq.num_rxq_sets; j++) - rx_qgrp->splitq.rxq_sets[j]->rxq.q_vector = NULL; - else - for (j = 0; j < rx_qgrp->singleq.num_rxq; j++) - rx_qgrp->singleq.rxqs[j]->q_vector = NULL; - } - - if (idpf_is_queue_model_split(vport->txq_model)) - for (i = 0; i < vport->num_txq_grp; i++) - vport->txq_grps[i].complq->q_vector = NULL; - else - for (i = 0; i < vport->num_txq_grp; i++) - for (j = 0; j < vport->txq_grps[i].num_txq; j++) - vport->txq_grps[i].txqs[j]->q_vector = NULL; - kfree(vport->q_vectors); vport->q_vectors = NULL; }
The second tagged commit started sometimes (very rarely, but possible) throwing WARNs from net/core/page_pool.c:page_pool_disable_direct_recycling(). Turned out idpf frees interrupt vectors with embedded NAPIs *before* freeing the queues making page_pools' NAPI pointers lead to freed memory before these pools are destroyed by libeth. It's not clear whether there are other accesses to the freed vectors when destroying the queues, but anyway, we usually free queue/interrupt vectors only when the queues are destroyed and the NAPIs are guaranteed to not be referenced anywhere. Invert the allocation and freeing logic making queue/interrupt vectors be allocated first and freed last. Vectors don't require queues to be present, so this is safe. Additionally, this change allows to remove that useless queue->q_vector pointer cleanup, as vectors are still valid when freeing the queues (+ both are freed within one function, so it's not clear why nullify the pointers at all). Fixes: 1c325aac10a8 ("idpf: configure resources for TX queues") Fixes: 90912f9f4f2d ("idpf: convert header split mode to libeth + napi_build_skb()") Reported-by: Michal Kubiak <michal.kubiak@intel.com> Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com> --- drivers/net/ethernet/intel/idpf/idpf_lib.c | 24 ++++++++++----------- drivers/net/ethernet/intel/idpf/idpf_txrx.c | 24 +-------------------- 2 files changed, 13 insertions(+), 35 deletions(-)