[PATCH net 0/3][pull request] Intel Wired LAN Driver Updates 2022-04-05

[PATCH net 0/3][pull request] Intel Wired LAN Driver Updates 2022-04-05

[Tony Nguyen]

Maciej Fijalkowski says:

We were solving issues around AF_XDP busy poll's not-so-usual scenarios,
such as very big busy poll budgets applied to very small HW rings. This
set carries the things that were found during that work that apply to
net tree.

One thing that was fixed for all in-tree ZC drivers was missing on ice
side all the time - it's about syncing RCU before destroying XDP
resources. Next one fixes the bit that is checked in ice_xsk_wakeup and
third one avoids false setting of DD bits on Tx descriptors.

The following are changes since commit 1158f79f82d437093aeed87d57df0548bdd68146:
  ipv6: Fix stats accounting in ip6_pkt_drop
and are available in the git repository at:
  git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue 100GbE

Maciej Fijalkowski (3):
  ice: synchronize_rcu() when terminating rings
  ice: xsk: fix VSI state check in ice_xsk_wakeup()
  ice: clear cmd_type_offset_bsz for TX rings

 drivers/net/ethernet/intel/ice/ice.h      | 2 +-
 drivers/net/ethernet/intel/ice/ice_main.c | 6 ++++--
 drivers/net/ethernet/intel/ice/ice_xsk.c  | 6 ++++--
 3 files changed, 9 insertions(+), 5 deletions(-)

-- 
2.31.1

[PATCH net 1/3] ice: synchronize_rcu() when terminating rings

[Tony Nguyen]

From: Maciej Fijalkowski <maciej.fijalkowski@intel.com>

Unfortunately, the ice driver doesn't respect the RCU critical section that
XSK wakeup is surrounded with. To fix this, add synchronize_rcu() calls to
paths that destroy resources that might be in use.

This was addressed in other AF_XDP ZC enabled drivers, for reference see
for example commit b3873a5be757 ("net/i40e: Fix concurrency issues
between config flow and XSK")

Fixes: efc2214b6047 ("ice: Add support for XDP")
Fixes: 2d4238f55697 ("ice: Add support for AF_XDP")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Shwetha Nagaraju <shwetha.nagaraju@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice.h      | 2 +-
 drivers/net/ethernet/intel/ice/ice_main.c | 4 +++-
 drivers/net/ethernet/intel/ice/ice_xsk.c  | 4 +++-
 3 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice.h b/drivers/net/ethernet/intel/ice/ice.h
index 26eaee0b6503..8ed3c9ab7ff7 100644
--- a/drivers/net/ethernet/intel/ice/ice.h
+++ b/drivers/net/ethernet/intel/ice/ice.h
@@ -671,7 +671,7 @@ static inline struct ice_pf *ice_netdev_to_pf(struct net_device *netdev)
 
 static inline bool ice_is_xdp_ena_vsi(struct ice_vsi *vsi)
 {
-	return !!vsi->xdp_prog;
+	return !!READ_ONCE(vsi->xdp_prog);
 }
 
 static inline void ice_set_ring_xdp(struct ice_tx_ring *ring)
diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c
index 1d2ca39add95..d2039a9306b8 100644
--- a/drivers/net/ethernet/intel/ice/ice_main.c
+++ b/drivers/net/ethernet/intel/ice/ice_main.c
@@ -2758,8 +2758,10 @@ int ice_destroy_xdp_rings(struct ice_vsi *vsi)
 
 	ice_for_each_xdp_txq(vsi, i)
 		if (vsi->xdp_rings[i]) {
-			if (vsi->xdp_rings[i]->desc)
+			if (vsi->xdp_rings[i]->desc) {
+				synchronize_rcu();
 				ice_free_tx_ring(vsi->xdp_rings[i]);
+			}
 			kfree_rcu(vsi->xdp_rings[i], rcu);
 			vsi->xdp_rings[i] = NULL;
 		}
diff --git a/drivers/net/ethernet/intel/ice/ice_xsk.c b/drivers/net/ethernet/intel/ice/ice_xsk.c
index dfbcaf08520e..33b28a72ffcb 100644
--- a/drivers/net/ethernet/intel/ice/ice_xsk.c
+++ b/drivers/net/ethernet/intel/ice/ice_xsk.c
@@ -41,8 +41,10 @@ static void ice_qp_reset_stats(struct ice_vsi *vsi, u16 q_idx)
 static void ice_qp_clean_rings(struct ice_vsi *vsi, u16 q_idx)
 {
 	ice_clean_tx_ring(vsi->tx_rings[q_idx]);
-	if (ice_is_xdp_ena_vsi(vsi))
+	if (ice_is_xdp_ena_vsi(vsi)) {
+		synchronize_rcu();
 		ice_clean_tx_ring(vsi->xdp_rings[q_idx]);
+	}
 	ice_clean_rx_ring(vsi->rx_rings[q_idx]);
 }
 
-- 
2.31.1

[PATCH net 2/3] ice: xsk: fix VSI state check in ice_xsk_wakeup()

[Tony Nguyen]

From: Maciej Fijalkowski <maciej.fijalkowski@intel.com>

ICE_DOWN is dedicated for pf->state. Check for ICE_VSI_DOWN being set on
vsi->state in ice_xsk_wakeup().

Fixes: 2d4238f55697 ("ice: Add support for AF_XDP")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Shwetha Nagaraju <shwetha.nagaraju@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_xsk.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_xsk.c b/drivers/net/ethernet/intel/ice/ice_xsk.c
index 33b28a72ffcb..866ee4df9671 100644
--- a/drivers/net/ethernet/intel/ice/ice_xsk.c
+++ b/drivers/net/ethernet/intel/ice/ice_xsk.c
@@ -920,7 +920,7 @@ ice_xsk_wakeup(struct net_device *netdev, u32 queue_id,
 	struct ice_vsi *vsi = np->vsi;
 	struct ice_tx_ring *ring;
 
-	if (test_bit(ICE_DOWN, vsi->state))
+	if (test_bit(ICE_VSI_DOWN, vsi->state))
 		return -ENETDOWN;
 
 	if (!ice_is_xdp_ena_vsi(vsi))
-- 
2.31.1

[PATCH net 3/3] ice: clear cmd_type_offset_bsz for TX rings

[Tony Nguyen]

From: Maciej Fijalkowski <maciej.fijalkowski@intel.com>

Currently when XDP rings are created, each descriptor gets its DD bit
set, which turns out to be the wrong approach as it can lead to a
situation where more descriptors get cleaned than it was supposed to,
e.g. when AF_XDP busy poll is run with a large batch size. In this
situation, the driver would request for more buffers than it is able to
handle.

Fix this by not setting the DD bits in ice_xdp_alloc_setup_rings(). They
should be initialized to zero instead.

Fixes: 9610bd988df9 ("ice: optimize XDP_TX workloads")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Shwetha Nagaraju <shwetha.nagaraju@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c
index d2039a9306b8..d768925785ca 100644
--- a/drivers/net/ethernet/intel/ice/ice_main.c
+++ b/drivers/net/ethernet/intel/ice/ice_main.c
@@ -2562,7 +2562,7 @@ static int ice_xdp_alloc_setup_rings(struct ice_vsi *vsi)
 		spin_lock_init(&xdp_ring->tx_lock);
 		for (j = 0; j < xdp_ring->count; j++) {
 			tx_desc = ICE_TX_DESC(xdp_ring, j);
-			tx_desc->cmd_type_offset_bsz = cpu_to_le64(ICE_TX_DESC_DTYPE_DESC_DONE);
+			tx_desc->cmd_type_offset_bsz = 0;
 		}
 	}
 
-- 
2.31.1

Re: [PATCH net 0/3][pull request] Intel Wired LAN Driver Updates 2022-04-05

[patchwork-bot+netdevbpf]

Hello:

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

On Tue,  5 Apr 2022 09:38:00 -0700 you wrote:
> Maciej Fijalkowski says:
> 
> We were solving issues around AF_XDP busy poll's not-so-usual scenarios,
> such as very big busy poll budgets applied to very small HW rings. This
> set carries the things that were found during that work that apply to
> net tree.
> 
> [...]

Here is the summary with links:
  - [net,1/3] ice: synchronize_rcu() when terminating rings
    https://git.kernel.org/netdev/net/c/f9124c68f05f
  - [net,2/3] ice: xsk: fix VSI state check in ice_xsk_wakeup()
    https://git.kernel.org/netdev/net/c/72b915a2b444
  - [net,3/3] ice: clear cmd_type_offset_bsz for TX rings
    https://git.kernel.org/netdev/net/c/e19778e6c911

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html