| Message ID | 20240508070935.11501-1-hkelam@marvell.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 04fb71cc5f1866f391a9c21ea634217e065ae525 |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net-next] octeontx2-pf: Reuse Transmit queue/Send queue index of HTB class | expand |
… > This patch solves the problem by assigning deleted class transmit > queue/send queue to active class. * How do you think about to add the tag “Fixes”? * Would you like to use imperative wordings for an improved change description? https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst?h=v6.9-rc7#n94 Regards, Markus
> … > > This patch solves the problem by assigning deleted class transmit > > queue/send queue to active class. > > * How do you think about to add the tag “Fixes”? > This patch is targeted to "net-next" ,reason being it implements a new way of assigning transmit queues to the HTB classes. Though the commit description mentions solving a problem, but the implementation went beyond a simple fix. Thanks, Hariprasad k > * Would you like to use imperative wordings for an improved change > description? > https://urldefense.proofpoint.com/v2/url?u=https- > 3A__git.kernel.org_pub_scm_linux_kernel_git_torvalds_linux.git_tree_Docum > entation_process_submitting-2Dpatches.rst-3Fh-3Dv6.9-2Drc7- > 23n94&d=DwIFaQ&c=nKjWec2b6R0mOyPaz7xtfQ&r=2bd4kP44ECYFgf- > KoNSJWqEipEtpxXnNBKy0vyoJJ8A&m=MOvcAghhRCElPpsRn5uxNJhBQZu4Fl3- > ddvU7WhIbz5A7suWUSas__inrZk8Mf__&s=Cq1cWWlB5WUn1OnoW8aG_ghi > w7xXjNVqh2DsIaumLRU&e= > > Regards, > Markus
On Wed, May 08, 2024 at 12:39:35PM +0530, Hariprasad Kelam wrote: > Real number of Transmit queues are incremented when user enables HTB > class and vice versa. Depending on SKB priority driver returns transmit > queue (Txq). Transmit queues and Send queues are one-to-one mapped. > > In few scenarios, Driver is returning transmit queue value which is > greater than real number of transmit queue and Stack detects this as > error and overwrites transmit queue value. > > For example > user has added two classes and real number of queues are incremented > accordingly > - tc class add dev eth1 parent 1: classid 1:1 htb > rate 100Mbit ceil 100Mbit prio 1 quantum 1024 > - tc class add dev eth1 parent 1: classid 1:2 htb > rate 100Mbit ceil 200Mbit prio 7 quantum 1024 > > now if user deletes the class with id 1:1, driver decrements the real > number of queues > - tc class del dev eth1 classid 1:1 > > But for the class with id 1:2, driver is returning transmit queue > value which is higher than real number of transmit queue leading > to below error > > eth1 selects TX queue x, but real number of TX queues is x > > This patch solves the problem by assigning deleted class transmit > queue/send queue to active class. > > Signed-off-by: Hariprasad Kelam <hkelam@marvell.com> Reviewed-by: Simon Horman <horms@kernel.org>
Hello: This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <kuba@kernel.org>: On Wed, 8 May 2024 12:39:35 +0530 you wrote: > Real number of Transmit queues are incremented when user enables HTB > class and vice versa. Depending on SKB priority driver returns transmit > queue (Txq). Transmit queues and Send queues are one-to-one mapped. > > In few scenarios, Driver is returning transmit queue value which is > greater than real number of transmit queue and Stack detects this as > error and overwrites transmit queue value. > > [...] Here is the summary with links: - [net-next] octeontx2-pf: Reuse Transmit queue/Send queue index of HTB class https://git.kernel.org/netdev/net-next/c/04fb71cc5f18 You are awesome, thank you!
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/qos.c b/drivers/net/ethernet/marvell/octeontx2/nic/qos.c index 1723e9912ae0..070711df612e 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/qos.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/qos.c @@ -545,6 +545,20 @@ otx2_qos_sw_create_leaf_node(struct otx2_nic *pfvf, return node; } +static struct otx2_qos_node +*otx2_sw_node_find_by_qid(struct otx2_nic *pfvf, u16 qid) +{ + struct otx2_qos_node *node = NULL; + int bkt; + + hash_for_each(pfvf->qos.qos_hlist, bkt, node, hlist) { + if (node->qid == qid) + break; + } + + return node; +} + static struct otx2_qos_node * otx2_sw_node_find(struct otx2_nic *pfvf, u32 classid) { @@ -917,6 +931,7 @@ static void otx2_qos_enadis_sq(struct otx2_nic *pfvf, otx2_qos_disable_sq(pfvf, qid); pfvf->qos.qid_to_sqmap[qid] = node->schq; + otx2_qos_txschq_config(pfvf, node); otx2_qos_enable_sq(pfvf, qid); } @@ -1475,13 +1490,45 @@ static int otx2_qos_leaf_to_inner(struct otx2_nic *pfvf, u16 classid, return ret; } +static int otx2_qos_cur_leaf_nodes(struct otx2_nic *pfvf) +{ + int last = find_last_bit(pfvf->qos.qos_sq_bmap, pfvf->hw.tc_tx_queues); + + return last == pfvf->hw.tc_tx_queues ? 0 : last + 1; +} + +static void otx2_reset_qdisc(struct net_device *dev, u16 qid) +{ + struct netdev_queue *dev_queue = netdev_get_tx_queue(dev, qid); + struct Qdisc *qdisc = rtnl_dereference(dev_queue->qdisc_sleeping); + + if (!qdisc) + return; + + spin_lock_bh(qdisc_lock(qdisc)); + qdisc_reset(qdisc); + spin_unlock_bh(qdisc_lock(qdisc)); +} + +static void otx2_cfg_smq(struct otx2_nic *pfvf, struct otx2_qos_node *node, + int qid) +{ + struct otx2_qos_node *tmp; + + list_for_each_entry(tmp, &node->child_schq_list, list) + if (tmp->level == NIX_TXSCH_LVL_MDQ) { + otx2_qos_txschq_config(pfvf, tmp); + pfvf->qos.qid_to_sqmap[qid] = tmp->schq; + } +} + static int otx2_qos_leaf_del(struct otx2_nic *pfvf, u16 *classid, struct netlink_ext_ack *extack) { struct otx2_qos_node *node, *parent; int dwrr_del_node = false; + u16 qid, moved_qid; u64 prio; - u16 qid; netdev_dbg(pfvf->netdev, "TC_HTB_LEAF_DEL classid %04x\n", *classid); @@ -1517,6 +1564,37 @@ static int otx2_qos_leaf_del(struct otx2_nic *pfvf, u16 *classid, if (!parent->child_static_cnt) parent->max_static_prio = 0; + moved_qid = otx2_qos_cur_leaf_nodes(pfvf); + + /* last node just deleted */ + if (moved_qid == 0 || moved_qid == qid) + return 0; + + moved_qid--; + + node = otx2_sw_node_find_by_qid(pfvf, moved_qid); + if (!node) + return 0; + + /* stop traffic to the old queue and disable + * SQ associated with it + */ + node->qid = OTX2_QOS_QID_INNER; + __clear_bit(moved_qid, pfvf->qos.qos_sq_bmap); + otx2_qos_disable_sq(pfvf, moved_qid); + + otx2_reset_qdisc(pfvf->netdev, pfvf->hw.tx_queues + moved_qid); + + /* enable SQ associated with qid and + * update the node + */ + otx2_cfg_smq(pfvf, node, qid); + + otx2_qos_enable_sq(pfvf, qid); + __set_bit(qid, pfvf->qos.qos_sq_bmap); + node->qid = qid; + + *classid = node->classid; return 0; }
Real number of Transmit queues are incremented when user enables HTB class and vice versa. Depending on SKB priority driver returns transmit queue (Txq). Transmit queues and Send queues are one-to-one mapped. In few scenarios, Driver is returning transmit queue value which is greater than real number of transmit queue and Stack detects this as error and overwrites transmit queue value. For example user has added two classes and real number of queues are incremented accordingly - tc class add dev eth1 parent 1: classid 1:1 htb rate 100Mbit ceil 100Mbit prio 1 quantum 1024 - tc class add dev eth1 parent 1: classid 1:2 htb rate 100Mbit ceil 200Mbit prio 7 quantum 1024 now if user deletes the class with id 1:1, driver decrements the real number of queues - tc class del dev eth1 classid 1:1 But for the class with id 1:2, driver is returning transmit queue value which is higher than real number of transmit queue leading to below error eth1 selects TX queue x, but real number of TX queues is x This patch solves the problem by assigning deleted class transmit queue/send queue to active class. Signed-off-by: Hariprasad Kelam <hkelam@marvell.com> --- .../net/ethernet/marvell/octeontx2/nic/qos.c | 80 ++++++++++++++++++- 1 file changed, 79 insertions(+), 1 deletion(-)