net/mlx5: Make ASO poll CQ usable in atomic context

Poll CQ functions shouldn't sleep as they are called in atomic context.
The following splat appears once the mlx5_aso_poll_cq() is used in such
flow.

 BUG: scheduling while atomic: swapper/17/0/0x00000100
 Modules linked in: sch_ingress openvswitch nsh mlx5_vdpa vringh vhost_iotlb vdpa mlx5_ib mlx5_core xt_conntrack xt_MASQUERADE nf_conntrack_netlink nfnetlink xt_addrtype iptable_nat nf_nat br_netfilter overlay rpcrdma rdma_ucm ib_iser libiscsi scsi_transport_iscsi ib_umad rdma_cm ib_ipoib iw_cm ib_cm ib_uverbs ib_core fuse [last unloaded: mlx5_core]
 CPU: 17 PID: 0 Comm: swapper/17 Tainted: G        W          6.0.0-rc2+ thesofproject#13
 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
 Call Trace:
  <IRQ>
  dump_stack_lvl+0x34/0x44
  __schedule_bug.cold+0x47/0x53
  __schedule+0x4b6/0x670
  ? hrtimer_start_range_ns+0x28d/0x360
  schedule+0x50/0x90
  schedule_hrtimeout_range_clock+0x98/0x120
  ? __hrtimer_init+0xb0/0xb0
  usleep_range_state+0x60/0x90
  mlx5_aso_poll_cq+0xad/0x190 [mlx5_core]
  mlx5e_ipsec_aso_update_curlft+0x81/0xb0 [mlx5_core]
  xfrm_timer_handler+0x6b/0x360
  ? xfrm_find_acq_byseq+0x50/0x50
  __hrtimer_run_queues+0x139/0x290
  hrtimer_run_softirq+0x7d/0xe0
  __do_softirq+0xc7/0x272
  irq_exit_rcu+0x87/0xb0
  sysvec_apic_timer_interrupt+0x72/0x90
  </IRQ>
  <TASK>
  asm_sysvec_apic_timer_interrupt+0x16/0x20
 RIP: 0010:default_idle+0x18/0x20
 Code: ae 7d ff ff cc cc cc cc cc cc cc cc cc cc cc cc cc cc 0f 1f 44 00 00 8b 05 b5 30 0d 01 85 c0 7e 07 0f 00 2d 0a e3 53 00 fb f4 <c3> 0f 1f 80 00 00 00 00 0f 1f 44 00 00 65 48 8b 04 25 80 ad 01 00
 RSP: 0018:ffff888100883ee0 EFLAGS: 00000242
 RAX: 0000000000000001 RBX: ffff888100849580 RCX: 4000000000000000
 RDX: 0000000000000001 RSI: 0000000000000083 RDI: 000000000008863c
 RBP: 0000000000000011 R08: 00000064e6977fa9 R09: 0000000000000001
 R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
 R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
  default_idle_call+0x37/0xb0
  do_idle+0x1cd/0x1e0
  cpu_startup_entry+0x19/0x20
  start_secondary+0xfe/0x120
  secondary_startup_64_no_verify+0xcd/0xdb
  </TASK>
 softirq: huh, entered softirq 8 HRTIMER 00000000a97c08cb with preempt_count 00000100, exited with 00000000?

Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Author: rleon

drivers/net/ethernet/mellanox/mlx5/core/en/tc/meter.c

@@ -115,6 +115,7 @@ mlx5e_tc_meter_modify(struct mlx5_core_dev *mdev,
 	struct mlx5e_flow_meters *flow_meters;
 	u8 cir_man, cir_exp, cbs_man, cbs_exp;
 	struct mlx5_aso_wqe *aso_wqe;
+	unsigned long expires;
 	struct mlx5_aso *aso;
 	u64 rate, burst;
 	u8 ds_cnt;
@@ -187,7 +188,12 @@ mlx5e_tc_meter_modify(struct mlx5_core_dev *mdev,
 	mlx5_aso_post_wqe(aso, true, &aso_wqe->ctrl);
 
 	/* With newer FW, the wait for the first ASO WQE is more than 2us, put the wait 10ms. */
-	err = mlx5_aso_poll_cq(aso, true, 10);
+	expires = jiffies + msecs_to_jiffies(10);
+	do {
+		err = mlx5_aso_poll_cq(aso, true);
+		if (err)
+			usleep_range(2, 10);
+	} while (err && time_is_after_jiffies(expires));
 	mutex_unlock(&flow_meters->aso_lock);
 
 	return err;

drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c

@@ -1405,7 +1405,7 @@ static int macsec_aso_set_arm_event(struct mlx5_core_dev *mdev, struct mlx5e_mac
 			   MLX5_ACCESS_ASO_OPC_MOD_MACSEC);
 	macsec_aso_build_ctrl(aso, &aso_wqe->aso_ctrl, in);
 	mlx5_aso_post_wqe(maso, false, &aso_wqe->ctrl);
-	err = mlx5_aso_poll_cq(maso, false, 10);
+	err = mlx5_aso_poll_cq(maso, false);
 	mutex_unlock(&aso->aso_lock);
 
 	return err;
@@ -1430,7 +1430,7 @@ static int macsec_aso_query(struct mlx5_core_dev *mdev, struct mlx5e_macsec *mac
 	macsec_aso_build_wqe_ctrl_seg(aso, &aso_wqe->aso_ctrl, NULL);
 
 	mlx5_aso_post_wqe(maso, false, &aso_wqe->ctrl);
-	err = mlx5_aso_poll_cq(maso, false, 10);
+	err = mlx5_aso_poll_cq(maso, false);
 	if (err)
 		goto err_out;
 

drivers/net/ethernet/mellanox/mlx5/core/lib/aso.c

@@ -381,20 +381,12 @@ void mlx5_aso_post_wqe(struct mlx5_aso *aso, bool with_data,
 	WRITE_ONCE(doorbell_cseg, NULL);
 }
 
-int mlx5_aso_poll_cq(struct mlx5_aso *aso, bool with_data, u32 interval_ms)
+int mlx5_aso_poll_cq(struct mlx5_aso *aso, bool with_data)
 {
 	struct mlx5_aso_cq *cq = &aso->cq;
 	struct mlx5_cqe64 *cqe;
-	unsigned long expires;
 
 	cqe = mlx5_cqwq_get_cqe(&cq->wq);
-
-	expires = jiffies + msecs_to_jiffies(interval_ms);
-	while (!cqe && time_is_after_jiffies(expires)) {
-		usleep_range(2, 10);
-		cqe = mlx5_cqwq_get_cqe(&cq->wq);
-	}
-
 	if (!cqe)
 		return -ETIMEDOUT;
 

drivers/net/ethernet/mellanox/mlx5/core/lib/aso.h

@@ -83,7 +83,7 @@ void mlx5_aso_build_wqe(struct mlx5_aso *aso, u8 ds_cnt,
 			u32 obj_id, u32 opc_mode);
 void mlx5_aso_post_wqe(struct mlx5_aso *aso, bool with_data,
 		       struct mlx5_wqe_ctrl_seg *doorbell_cseg);
-int mlx5_aso_poll_cq(struct mlx5_aso *aso, bool with_data, u32 interval_ms);
+int mlx5_aso_poll_cq(struct mlx5_aso *aso, bool with_data);
 
 struct mlx5_aso *mlx5_aso_create(struct mlx5_core_dev *mdev, u32 pdn);
 void mlx5_aso_destroy(struct mlx5_aso *aso);