From b7f406bb883ba7ac3222298f6b44cebc4cfe2dde Mon Sep 17 00:00:00 2001 From: Chuhong Yuan Date: Wed, 17 Jul 2019 16:21:01 +0800 Subject: IB/mlx5: Replace kfree with kvfree Memory allocated by kvzalloc should not be freed by kfree(), use kvfree() instead. Fixes: 813e90b1aeaa ("IB/mlx5: Add advise_mr() support") Link: https://lore.kernel.org/r/20190717082101.14196-1-hslester96@gmail.com Signed-off-by: Chuhong Yuan Reviewed-by: Jason Gunthorpe Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/odp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/infiniband/hw/mlx5/odp.c') diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 5b642d81e617..36ba901cc9a5 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -1771,7 +1771,7 @@ static void mlx5_ib_prefetch_mr_work(struct work_struct *work) num_pending_prefetch_dec(to_mdev(w->pd->device), w->sg_list, w->num_sge, 0); - kfree(w); + kvfree(w); } int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, @@ -1813,7 +1813,7 @@ int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, if (valid_req) queue_work(system_unbound_wq, &work->work); else - kfree(work); + kvfree(work); srcu_read_unlock(&dev->mr_srcu, srcu_key); -- cgit v1.2.3 From 296e3a2aad09d328f22e54655c3d736033fe1ae8 Mon Sep 17 00:00:00 2001 From: Moni Shoua Date: Tue, 23 Jul 2019 09:57:30 +0300 Subject: IB/mlx5: Prevent concurrent MR updates during invalidation The device requires that memory registration work requests that update the address translation table of a MR will be fenced if posted together. This scenario can happen when address ranges are invalidated by the mmu in separate concurrent calls to the invalidation callback. We prefer to block concurrent address updates for a single MR over fencing since making the decision if a WQE needs fencing will be more expensive and fencing all WQEs is a too radical choice. Further, it isn't clear that this code can even run safely concurrently, so a lock is a safer choice. Fixes: b4cfe447d47b ("IB/mlx5: Implement on demand paging by adding support for MMU notifiers") Link: https://lore.kernel.org/r/20190723065733.4899-8-leon@kernel.org Signed-off-by: Moni Shoua Reviewed-by: Artemy Kovalyov Signed-off-by: Leon Romanovsky Reviewed-by: Jason Gunthorpe Signed-off-by: Jason Gunthorpe --- drivers/infiniband/hw/mlx5/odp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/infiniband/hw/mlx5/odp.c') diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 36ba901cc9a5..81da82050d05 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -246,7 +246,7 @@ void mlx5_ib_invalidate_range(struct ib_umem_odp *umem_odp, unsigned long start, * overwrite the same MTTs. Concurent invalidations might race us, * but they will write 0s as well, so no difference in the end result. */ - + mutex_lock(&umem_odp->umem_mutex); for (addr = start; addr < end; addr += BIT(umem_odp->page_shift)) { idx = (addr - ib_umem_start(umem_odp)) >> umem_odp->page_shift; /* @@ -278,6 +278,7 @@ void mlx5_ib_invalidate_range(struct ib_umem_odp *umem_odp, unsigned long start, idx - blk_start_idx + 1, 0, MLX5_IB_UPD_XLT_ZAP | MLX5_IB_UPD_XLT_ATOMIC); + mutex_unlock(&umem_odp->umem_mutex); /* * We are now sure that the device will not access the * memory. We can safely unmap it, and mark it as dirty if -- cgit v1.2.3 From f591822c3cf314442819486f45ff7dc1f690e0c0 Mon Sep 17 00:00:00 2001 From: Yishai Hadas Date: Mon, 5 Aug 2019 11:30:10 +0300 Subject: IB/mlx5: Fix implicit MR release flow Once implicit MR is being called to be released by ib_umem_notifier_release() its leaves were marked as "dying". However, when dereg_mr()->mlx5_ib_free_implicit_mr()->mr_leaf_free() is called, it skips running the mr_leaf_free_action (i.e. umem_odp->work) when those leaves were marked as "dying". As such ib_umem_release() for the leaves won't be called and their MRs will be leaked as well. When an application exits/killed without calling dereg_mr we might hit the above flow. This fatal scenario is reported by WARN_ON() upon mlx5_ib_dealloc_ucontext() as ibcontext->per_mm_list is not empty, the call trace can be seen below. Originally the "dying" mark as part of ib_umem_notifier_release() was introduced to prevent pagefault_mr() from returning a success response once this happened. However, we already have today the completion mechanism so no need for that in those flows any more. Even in case a success response will be returned the firmware will not find the pages and an error will be returned in the following call as a released mm will cause ib_umem_odp_map_dma_pages() to permanently fail mmget_not_zero(). Fix the above issue by dropping the "dying" from the above flows. The other flows that are using "dying" are still needed it for their synchronization purposes. WARNING: CPU: 1 PID: 7218 at drivers/infiniband/hw/mlx5/main.c:2004 mlx5_ib_dealloc_ucontext+0x84/0x90 [mlx5_ib] CPU: 1 PID: 7218 Comm: ibv_rc_pingpong Tainted: G E 5.2.0-rc6+ #13 Call Trace: uverbs_destroy_ufile_hw+0xb5/0x120 [ib_uverbs] ib_uverbs_close+0x1f/0x80 [ib_uverbs] __fput+0xbe/0x250 task_work_run+0x88/0xa0 do_exit+0x2cb/0xc30 ? __fput+0x14b/0x250 do_group_exit+0x39/0xb0 get_signal+0x191/0x920 ? _raw_spin_unlock_bh+0xa/0x20 ? inet_csk_accept+0x229/0x2f0 do_signal+0x36/0x5e0 ? put_unused_fd+0x5b/0x70 ? __sys_accept4+0x1a6/0x1e0 ? inet_hash+0x35/0x40 ? release_sock+0x43/0x90 ? _raw_spin_unlock_bh+0xa/0x20 ? inet_listen+0x9f/0x120 exit_to_usermode_loop+0x5c/0xc6 do_syscall_64+0x182/0x1b0 entry_SYSCALL_64_after_hwframe+0x44/0xa9 Fixes: 81713d3788d2 ("IB/mlx5: Add implicit MR support") Link: https://lore.kernel.org/r/20190805083010.21777-1-leon@kernel.org Signed-off-by: Yishai Hadas Reviewed-by: Artemy Kovalyov Signed-off-by: Leon Romanovsky Reviewed-by: Jason Gunthorpe Signed-off-by: Jason Gunthorpe --- drivers/infiniband/core/umem_odp.c | 4 ---- drivers/infiniband/hw/mlx5/odp.c | 24 +++++++++--------------- 2 files changed, 9 insertions(+), 19 deletions(-) (limited to 'drivers/infiniband/hw/mlx5/odp.c') diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c index 2a75c6f8d827..c0e15db34680 100644 --- a/drivers/infiniband/core/umem_odp.c +++ b/drivers/infiniband/core/umem_odp.c @@ -112,10 +112,6 @@ static int ib_umem_notifier_release_trampoline(struct ib_umem_odp *umem_odp, * prevent any further fault handling on this MR. */ ib_umem_notifier_start_account(umem_odp); - umem_odp->dying = 1; - /* Make sure that the fact the umem is dying is out before we release - * all pending page faults. */ - smp_wmb(); complete_all(&umem_odp->notifier_completion); umem_odp->umem.context->invalidate_range( umem_odp, ib_umem_start(umem_odp), ib_umem_end(umem_odp)); diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 81da82050d05..1d257d1b3b0d 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -579,7 +579,6 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr, u32 flags) { int npages = 0, current_seq, page_shift, ret, np; - bool implicit = false; struct ib_umem_odp *odp_mr = to_ib_umem_odp(mr->umem); bool downgrade = flags & MLX5_PF_FLAGS_DOWNGRADE; bool prefetch = flags & MLX5_PF_FLAGS_PREFETCH; @@ -594,7 +593,6 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr, if (IS_ERR(odp)) return PTR_ERR(odp); mr = odp->private; - implicit = true; } else { odp = odp_mr; } @@ -682,19 +680,15 @@ next_mr: out: if (ret == -EAGAIN) { - if (implicit || !odp->dying) { - unsigned long timeout = - msecs_to_jiffies(MMU_NOTIFIER_TIMEOUT); - - if (!wait_for_completion_timeout( - &odp->notifier_completion, - timeout)) { - mlx5_ib_warn(dev, "timeout waiting for mmu notifier. seq %d against %d. notifiers_count=%d\n", - current_seq, odp->notifiers_seq, odp->notifiers_count); - } - } else { - /* The MR is being killed, kill the QP as well. */ - ret = -EFAULT; + unsigned long timeout = msecs_to_jiffies(MMU_NOTIFIER_TIMEOUT); + + if (!wait_for_completion_timeout(&odp->notifier_completion, + timeout)) { + mlx5_ib_warn( + dev, + "timeout waiting for mmu notifier. seq %d against %d. notifiers_count=%d\n", + current_seq, odp->notifiers_seq, + odp->notifiers_count); } } -- cgit v1.2.3 From 008157528ac5658502c0f87e872778c56c41109c Mon Sep 17 00:00:00 2001 From: Moni Shoua Date: Thu, 15 Aug 2019 11:38:32 +0300 Subject: IB/mlx5: Report and handle ODP support properly ODP depends on the several device capabilities, among them is the ability to send UMR WQEs with that modify atomic and entity size of the MR. Therefore, only if all conditions to send such a UMR WQE are met then driver can report that ODP is supported. Use this check of conditions in all places where driver needs to know about ODP support. Also, implicit ODP support depends on ability of driver to send UMR WQEs for an indirect mkey. Therefore, verify that all conditions to do so are met when reporting support. Fixes: c8d75a980fab ("IB/mlx5: Respect new UMR capabilities") Signed-off-by: Moni Shoua Reviewed-by: Guy Levi Signed-off-by: Leon Romanovsky Link: https://lore.kernel.org/r/20190815083834.9245-7-leon@kernel.org Signed-off-by: Doug Ledford --- drivers/infiniband/hw/mlx5/main.c | 6 +++--- drivers/infiniband/hw/mlx5/odp.c | 17 +++++++++-------- 2 files changed, 12 insertions(+), 11 deletions(-) (limited to 'drivers/infiniband/hw/mlx5/odp.c') diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index e12a4404096b..0569bcab02d4 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -1023,7 +1023,7 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, props->timestamp_mask = 0x7FFFFFFFFFFFFFFFULL; if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) { - if (MLX5_CAP_GEN(mdev, pg)) + if (dev->odp_caps.general_caps & IB_ODP_SUPPORT) props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING; props->odp_caps = dev->odp_caps; } @@ -6139,6 +6139,8 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev) dev->port[i].roce.last_port_state = IB_PORT_DOWN; } + mlx5_ib_internal_fill_odp_caps(dev); + err = mlx5_ib_init_multiport_master(dev); if (err) return err; @@ -6563,8 +6565,6 @@ static void mlx5_ib_stage_dev_res_cleanup(struct mlx5_ib_dev *dev) static int mlx5_ib_stage_odp_init(struct mlx5_ib_dev *dev) { - mlx5_ib_internal_fill_odp_caps(dev); - return mlx5_ib_odp_init_one(dev); } diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 1d257d1b3b0d..0a59912a4cef 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -301,7 +301,8 @@ void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev) memset(caps, 0, sizeof(*caps)); - if (!MLX5_CAP_GEN(dev->mdev, pg)) + if (!MLX5_CAP_GEN(dev->mdev, pg) || + !mlx5_ib_can_use_umr(dev, true)) return; caps->general_caps = IB_ODP_SUPPORT; @@ -355,7 +356,8 @@ void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev) if (MLX5_CAP_GEN(dev->mdev, fixed_buffer_size) && MLX5_CAP_GEN(dev->mdev, null_mkey) && - MLX5_CAP_GEN(dev->mdev, umr_extended_translation_offset)) + MLX5_CAP_GEN(dev->mdev, umr_extended_translation_offset) && + !MLX5_CAP_GEN(dev->mdev, umr_indirect_mkey_disabled)) caps->general_caps |= IB_ODP_SUPPORT_IMPLICIT; return; @@ -1622,8 +1624,10 @@ int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev) { int ret = 0; - if (dev->odp_caps.general_caps & IB_ODP_SUPPORT) - ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_odp_ops); + if (!(dev->odp_caps.general_caps & IB_ODP_SUPPORT)) + return ret; + + ib_set_device_ops(&dev->ib_dev, &mlx5_ib_dev_odp_ops); if (dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT) { ret = mlx5_cmd_null_mkey(dev->mdev, &dev->null_mkey); @@ -1633,9 +1637,6 @@ int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev) } } - if (!MLX5_CAP_GEN(dev->mdev, pg)) - return ret; - ret = mlx5_ib_create_pf_eq(dev, &dev->odp_pf_eq); return ret; @@ -1643,7 +1644,7 @@ int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev) void mlx5_ib_odp_cleanup_one(struct mlx5_ib_dev *dev) { - if (!MLX5_CAP_GEN(dev->mdev, pg)) + if (!(dev->odp_caps.general_caps & IB_ODP_SUPPORT)) return; mlx5_ib_destroy_pf_eq(dev, &dev->odp_pf_eq); -- cgit v1.2.3