Commit e1d32acb authored by Mintz, Yuval's avatar Mintz, Yuval Committed by David S. Miller

qed*: Change maximal number of queues

Today qede requests contexts that would suffice for 64 'whole'
combined queues [192 meant for 64 rx, tx and xdp tx queues],
but registers netdev and limits the number of queues based on
information received by qed. In turn, qed doesn't take context
into account when informing qede how many queues it can support.

This would lead to a configuration problem in case user tries
configuring >64 combined queues to interface [or >96 in case
xdp isn't enabled]. Since we don't have a mangement firware
that actually provides so many interrupt lines to a single
device we're currently safe but that's about to change soon.

The new maximum is hence changed:
  - For RoCE devices, the limit would remain 64.
  - For non-RoCE devices, the limit might be higher [depending
    on the actual configuration of the device].
qed would start enforcing that limit in both scenarios.
Signed-off-by: default avatarYuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent aed284c7
......@@ -1753,13 +1753,31 @@ static int qed_fill_eth_dev_info(struct qed_dev *cdev,
int max_vf_mac_filters = 0;
if (cdev->int_params.out.int_mode == QED_INT_MODE_MSIX) {
for_each_hwfn(cdev, i)
info->num_queues +=
FEAT_NUM(&cdev->hwfns[i], QED_PF_L2_QUE);
if (cdev->int_params.fp_msix_cnt)
info->num_queues =
min_t(u8, info->num_queues,
cdev->int_params.fp_msix_cnt);
u16 num_queues = 0;
/* Since the feature controls only queue-zones,
* make sure we have the contexts [rx, tx, xdp] to
* match.
*/
for_each_hwfn(cdev, i) {
struct qed_hwfn *hwfn = &cdev->hwfns[i];
u16 l2_queues = (u16)FEAT_NUM(hwfn,
QED_PF_L2_QUE);
u16 cids;
cids = hwfn->pf_params.eth_pf_params.num_cons;
num_queues += min_t(u16, l2_queues, cids / 3);
}
/* queues might theoretically be >256, but interrupts'
* upper-limit guarantes that it would fit in a u8.
*/
if (cdev->int_params.fp_msix_cnt) {
u8 irqs = cdev->int_params.fp_msix_cnt;
info->num_queues = (u8)min_t(u16,
num_queues, irqs);
}
} else {
info->num_queues = cdev->num_hwfns;
}
......
......@@ -877,6 +877,17 @@ static void qed_update_pf_params(struct qed_dev *cdev,
params->rdma_pf_params.gl_pi = QED_ROCE_PROTOCOL_INDEX;
}
/* In case we might support RDMA, don't allow qede to be greedy
* with the L2 contexts. Allow for 64 queues [rx, tx, xdp] per hwfn.
*/
if (QED_LEADING_HWFN(cdev)->hw_info.personality ==
QED_PCI_ETH_ROCE) {
u16 *num_cons;
num_cons = &params->eth_pf_params.num_cons;
*num_cons = min_t(u16, *num_cons, 192);
}
for (i = 0; i < cdev->num_hwfns; i++) {
struct qed_hwfn *p_hwfn = &cdev->hwfns[i];
......
......@@ -753,7 +753,7 @@ static void qede_update_pf_params(struct qed_dev *cdev)
/* 64 rx + 64 tx + 64 XDP */
memset(&pf_params, 0, sizeof(struct qed_pf_params));
pf_params.eth_pf_params.num_cons = 192;
pf_params.eth_pf_params.num_cons = (MAX_SB_PER_PF_MIMD - 1) * 3;
qed_ops->common->update_pf_params(cdev, &pf_params);
}
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment