Commit 780278c2 authored by Mike Marciniszyn's avatar Mike Marciniszyn Committed by Jason Gunthorpe

IB/hfi1: Rework AIP and VNIC dummy netdev usage

All other users of the dummy netdevice embed the netdev in other
structures:

init_dummy_netdev(&mal->dummy_dev);
init_dummy_netdev(&eth->dummy_dev);
init_dummy_netdev(&ar->napi_dev);
init_dummy_netdev(&irq_grp->napi_ndev);
init_dummy_netdev(&wil->napi_ndev);
init_dummy_netdev(&trans_pcie->napi_dev);
init_dummy_netdev(&dev->napi_dev);
init_dummy_netdev(&bus->mux_dev);

The AIP and VNIC implementation turns that model inside out and used a
kfree() to free what appears to be a netdev struct when in reality, it is
a struct that enbodies the rx state as well as the dummy netdev used to
support napi_poll across disparate receive contexts.  The relationship is
infered by the odd allocation:

	const int netdev_size = sizeof(*dd->dummy_netdev) +
		sizeof(struct hfi1_netdev_priv);
	<snip>
	dd->dummy_netdev = kcalloc_node(1, netdev_size, GFP_KERNEL, dd->node);

Correct the issue by:
- Correctly naming the alloc and free functions
- Renaming hfi1_netdev_priv to hfi1_netdev_rx
- Replacing dd dummy_netdev with a netdev_rx pointer
- Embedding the net_device in hfi1_netdev_rx
- Moving the init_dummy_netdev to the alloc routine
- Adjusting wrappers to fit the new model

Fixes: 6991abcb ("IB/hfi1: Add functions to receive accelerated ipoib packets")
Link: https://lore.kernel.org/r/1617026056-50483-11-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by: default avatarKaike Wan <kaike.wan@intel.com>
Signed-off-by: default avatarMike Marciniszyn <mike.marciniszyn@cornelisnetworks.com>
Signed-off-by: default avatarDennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Signed-off-by: default avatarJason Gunthorpe <jgg@nvidia.com>
parent 9c8823e0
......@@ -15243,8 +15243,8 @@ int hfi1_init_dd(struct hfi1_devdata *dd)
(dd->revision >> CCE_REVISION_SW_SHIFT)
& CCE_REVISION_SW_MASK);
/* alloc netdev data */
ret = hfi1_netdev_alloc(dd);
/* alloc VNIC/AIP rx data */
ret = hfi1_alloc_rx(dd);
if (ret)
goto bail_cleanup;
......@@ -15348,7 +15348,7 @@ int hfi1_init_dd(struct hfi1_devdata *dd)
hfi1_comp_vectors_clean_up(dd);
msix_clean_up_interrupts(dd);
bail_cleanup:
hfi1_netdev_free(dd);
hfi1_free_rx(dd);
hfi1_pcie_ddcleanup(dd);
bail_free:
hfi1_free_devdata(dd);
......
......@@ -69,7 +69,6 @@
#include <rdma/ib_hdrs.h>
#include <rdma/opa_addr.h>
#include <linux/rhashtable.h>
#include <linux/netdevice.h>
#include <rdma/rdma_vt.h>
#include "chip_registers.h"
......@@ -1060,6 +1059,7 @@ struct sdma_vl_map;
#define SERIAL_MAX 16 /* length of the serial number */
typedef int (*send_routine)(struct rvt_qp *, struct hfi1_pkt_state *, u64);
struct hfi1_netdev_rx;
struct hfi1_devdata {
struct hfi1_ibdev verbs_dev; /* must be first */
/* pointers to related structs for this device */
......@@ -1402,7 +1402,7 @@ struct hfi1_devdata {
/* Lock to protect IRQ SRC register access */
spinlock_t irq_src_lock;
int vnic_num_vports;
struct net_device *dummy_netdev;
struct hfi1_netdev_rx *netdev_rx;
/* Keeps track of IPoIB RSM rule users */
atomic_t ipoib_rsm_usr_num;
......
......@@ -1767,7 +1767,7 @@ static void remove_one(struct pci_dev *pdev)
hfi1_unregister_ib_device(dd);
/* free netdev data */
hfi1_netdev_free(dd);
hfi1_free_rx(dd);
/*
* Disable the IB link, disable interrupts on the device,
......
......@@ -14,15 +14,14 @@
/**
* struct hfi1_netdev_rxq - Receive Queue for HFI
* dummy netdev. Both IPoIB and VNIC netdevices will be working on
* top of this device.
* Both IPoIB and VNIC netdevices will be working on the rx abstraction.
* @napi: napi object
* @priv: ptr to netdev_priv
* @rx: ptr to netdev_rx
* @rcd: ptr to receive context data
*/
struct hfi1_netdev_rxq {
struct napi_struct napi;
struct hfi1_netdev_priv *priv;
struct hfi1_netdev_rx *rx;
struct hfi1_ctxtdata *rcd;
};
......@@ -36,7 +35,8 @@ struct hfi1_netdev_rxq {
#define NUM_NETDEV_MAP_ENTRIES HFI1_MAX_NETDEV_CTXTS
/**
* struct hfi1_netdev_priv: data required to setup and run HFI netdev.
* struct hfi1_netdev_rx: data required to setup and run HFI netdev.
* @rx_napi: the dummy netdevice to support "polling" the receive contexts
* @dd: hfi1_devdata
* @rxq: pointer to dummy netdev receive queues.
* @num_rx_q: number of receive queues
......@@ -48,7 +48,8 @@ struct hfi1_netdev_rxq {
* @netdevs: atomic counter of netdevs using dummy netdev.
* When 0 receive queues will be freed.
*/
struct hfi1_netdev_priv {
struct hfi1_netdev_rx {
struct net_device rx_napi;
struct hfi1_devdata *dd;
struct hfi1_netdev_rxq *rxq;
int num_rx_q;
......@@ -60,42 +61,28 @@ struct hfi1_netdev_priv {
atomic_t netdevs;
};
static inline
struct hfi1_netdev_priv *hfi1_netdev_priv(struct net_device *dev)
{
return (struct hfi1_netdev_priv *)&dev[1];
}
static inline
int hfi1_netdev_ctxt_count(struct hfi1_devdata *dd)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
return priv->num_rx_q;
return dd->netdev_rx->num_rx_q;
}
static inline
struct hfi1_ctxtdata *hfi1_netdev_get_ctxt(struct hfi1_devdata *dd, int ctxt)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
return priv->rxq[ctxt].rcd;
return dd->netdev_rx->rxq[ctxt].rcd;
}
static inline
int hfi1_netdev_get_free_rmt_idx(struct hfi1_devdata *dd)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
return priv->rmt_start;
return dd->netdev_rx->rmt_start;
}
static inline
void hfi1_netdev_set_free_rmt_idx(struct hfi1_devdata *dd, int rmt_idx)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
priv->rmt_start = rmt_idx;
dd->netdev_rx->rmt_start = rmt_idx;
}
u32 hfi1_num_netdev_contexts(struct hfi1_devdata *dd, u32 available_contexts,
......@@ -105,8 +92,8 @@ void hfi1_netdev_enable_queues(struct hfi1_devdata *dd);
void hfi1_netdev_disable_queues(struct hfi1_devdata *dd);
int hfi1_netdev_rx_init(struct hfi1_devdata *dd);
int hfi1_netdev_rx_destroy(struct hfi1_devdata *dd);
int hfi1_netdev_alloc(struct hfi1_devdata *dd);
void hfi1_netdev_free(struct hfi1_devdata *dd);
int hfi1_alloc_rx(struct hfi1_devdata *dd);
void hfi1_free_rx(struct hfi1_devdata *dd);
int hfi1_netdev_add_data(struct hfi1_devdata *dd, int id, void *data);
void *hfi1_netdev_remove_data(struct hfi1_devdata *dd, int id);
void *hfi1_netdev_get_data(struct hfi1_devdata *dd, int id);
......
......@@ -17,11 +17,11 @@
#include <linux/etherdevice.h>
#include <rdma/ib_verbs.h>
static int hfi1_netdev_setup_ctxt(struct hfi1_netdev_priv *priv,
static int hfi1_netdev_setup_ctxt(struct hfi1_netdev_rx *rx,
struct hfi1_ctxtdata *uctxt)
{
unsigned int rcvctrl_ops;
struct hfi1_devdata *dd = priv->dd;
struct hfi1_devdata *dd = rx->dd;
int ret;
uctxt->rhf_rcv_function_map = netdev_rhf_rcv_functions;
......@@ -118,11 +118,11 @@ static void hfi1_netdev_deallocate_ctxt(struct hfi1_devdata *dd,
hfi1_free_ctxt(uctxt);
}
static int hfi1_netdev_allot_ctxt(struct hfi1_netdev_priv *priv,
static int hfi1_netdev_allot_ctxt(struct hfi1_netdev_rx *rx,
struct hfi1_ctxtdata **ctxt)
{
int rc;
struct hfi1_devdata *dd = priv->dd;
struct hfi1_devdata *dd = rx->dd;
rc = hfi1_netdev_allocate_ctxt(dd, ctxt);
if (rc) {
......@@ -130,7 +130,7 @@ static int hfi1_netdev_allot_ctxt(struct hfi1_netdev_priv *priv,
return rc;
}
rc = hfi1_netdev_setup_ctxt(priv, *ctxt);
rc = hfi1_netdev_setup_ctxt(rx, *ctxt);
if (rc) {
dd_dev_err(dd, "netdev ctxt setup failed %d\n", rc);
hfi1_netdev_deallocate_ctxt(dd, *ctxt);
......@@ -184,31 +184,31 @@ u32 hfi1_num_netdev_contexts(struct hfi1_devdata *dd, u32 available_contexts,
(u32)HFI1_MAX_NETDEV_CTXTS);
}
static int hfi1_netdev_rxq_init(struct net_device *dev)
static int hfi1_netdev_rxq_init(struct hfi1_netdev_rx *rx)
{
int i;
int rc;
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dev);
struct hfi1_devdata *dd = priv->dd;
struct hfi1_devdata *dd = rx->dd;
struct net_device *dev = &rx->rx_napi;
priv->num_rx_q = dd->num_netdev_contexts;
priv->rxq = kcalloc_node(priv->num_rx_q, sizeof(struct hfi1_netdev_rxq),
GFP_KERNEL, dd->node);
rx->num_rx_q = dd->num_netdev_contexts;
rx->rxq = kcalloc_node(rx->num_rx_q, sizeof(*rx->rxq),
GFP_KERNEL, dd->node);
if (!priv->rxq) {
if (!rx->rxq) {
dd_dev_err(dd, "Unable to allocate netdev queue data\n");
return (-ENOMEM);
}
for (i = 0; i < priv->num_rx_q; i++) {
struct hfi1_netdev_rxq *rxq = &priv->rxq[i];
for (i = 0; i < rx->num_rx_q; i++) {
struct hfi1_netdev_rxq *rxq = &rx->rxq[i];
rc = hfi1_netdev_allot_ctxt(priv, &rxq->rcd);
rc = hfi1_netdev_allot_ctxt(rx, &rxq->rcd);
if (rc)
goto bail_context_irq_failure;
hfi1_rcd_get(rxq->rcd);
rxq->priv = priv;
rxq->rx = rx;
rxq->rcd->napi = &rxq->napi;
dd_dev_info(dd, "Setting rcv queue %d napi to context %d\n",
i, rxq->rcd->ctxt);
......@@ -228,7 +228,7 @@ static int hfi1_netdev_rxq_init(struct net_device *dev)
bail_context_irq_failure:
dd_dev_err(dd, "Unable to allot receive context\n");
for (; i >= 0; i--) {
struct hfi1_netdev_rxq *rxq = &priv->rxq[i];
struct hfi1_netdev_rxq *rxq = &rx->rxq[i];
if (rxq->rcd) {
hfi1_netdev_deallocate_ctxt(dd, rxq->rcd);
......@@ -236,20 +236,19 @@ static int hfi1_netdev_rxq_init(struct net_device *dev)
rxq->rcd = NULL;
}
}
kfree(priv->rxq);
priv->rxq = NULL;
kfree(rx->rxq);
rx->rxq = NULL;
return rc;
}
static void hfi1_netdev_rxq_deinit(struct net_device *dev)
static void hfi1_netdev_rxq_deinit(struct hfi1_netdev_rx *rx)
{
int i;
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dev);
struct hfi1_devdata *dd = priv->dd;
struct hfi1_devdata *dd = rx->dd;
for (i = 0; i < priv->num_rx_q; i++) {
struct hfi1_netdev_rxq *rxq = &priv->rxq[i];
for (i = 0; i < rx->num_rx_q; i++) {
struct hfi1_netdev_rxq *rxq = &rx->rxq[i];
netif_napi_del(&rxq->napi);
hfi1_netdev_deallocate_ctxt(dd, rxq->rcd);
......@@ -257,41 +256,41 @@ static void hfi1_netdev_rxq_deinit(struct net_device *dev)
rxq->rcd = NULL;
}
kfree(priv->rxq);
priv->rxq = NULL;
priv->num_rx_q = 0;
kfree(rx->rxq);
rx->rxq = NULL;
rx->num_rx_q = 0;
}
static void enable_queues(struct hfi1_netdev_priv *priv)
static void enable_queues(struct hfi1_netdev_rx *rx)
{
int i;
for (i = 0; i < priv->num_rx_q; i++) {
struct hfi1_netdev_rxq *rxq = &priv->rxq[i];
for (i = 0; i < rx->num_rx_q; i++) {
struct hfi1_netdev_rxq *rxq = &rx->rxq[i];
dd_dev_info(priv->dd, "enabling queue %d on context %d\n", i,
dd_dev_info(rx->dd, "enabling queue %d on context %d\n", i,
rxq->rcd->ctxt);
napi_enable(&rxq->napi);
hfi1_rcvctrl(priv->dd,
hfi1_rcvctrl(rx->dd,
HFI1_RCVCTRL_CTXT_ENB | HFI1_RCVCTRL_INTRAVAIL_ENB,
rxq->rcd);
}
}
static void disable_queues(struct hfi1_netdev_priv *priv)
static void disable_queues(struct hfi1_netdev_rx *rx)
{
int i;
msix_netdev_synchronize_irq(priv->dd);
msix_netdev_synchronize_irq(rx->dd);
for (i = 0; i < priv->num_rx_q; i++) {
struct hfi1_netdev_rxq *rxq = &priv->rxq[i];
for (i = 0; i < rx->num_rx_q; i++) {
struct hfi1_netdev_rxq *rxq = &rx->rxq[i];
dd_dev_info(priv->dd, "disabling queue %d on context %d\n", i,
dd_dev_info(rx->dd, "disabling queue %d on context %d\n", i,
rxq->rcd->ctxt);
/* wait for napi if it was scheduled */
hfi1_rcvctrl(priv->dd,
hfi1_rcvctrl(rx->dd,
HFI1_RCVCTRL_CTXT_DIS | HFI1_RCVCTRL_INTRAVAIL_DIS,
rxq->rcd);
napi_synchronize(&rxq->napi);
......@@ -308,15 +307,14 @@ static void disable_queues(struct hfi1_netdev_priv *priv)
*/
int hfi1_netdev_rx_init(struct hfi1_devdata *dd)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
struct hfi1_netdev_rx *rx = dd->netdev_rx;
int res;
if (atomic_fetch_inc(&priv->netdevs))
if (atomic_fetch_inc(&rx->netdevs))
return 0;
mutex_lock(&hfi1_mutex);
init_dummy_netdev(dd->dummy_netdev);
res = hfi1_netdev_rxq_init(dd->dummy_netdev);
res = hfi1_netdev_rxq_init(rx);
mutex_unlock(&hfi1_mutex);
return res;
}
......@@ -329,12 +327,12 @@ int hfi1_netdev_rx_init(struct hfi1_devdata *dd)
*/
int hfi1_netdev_rx_destroy(struct hfi1_devdata *dd)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
struct hfi1_netdev_rx *rx = dd->netdev_rx;
/* destroy the RX queues only if it is the last netdev going away */
if (atomic_fetch_add_unless(&priv->netdevs, -1, 0) == 1) {
if (atomic_fetch_add_unless(&rx->netdevs, -1, 0) == 1) {
mutex_lock(&hfi1_mutex);
hfi1_netdev_rxq_deinit(dd->dummy_netdev);
hfi1_netdev_rxq_deinit(rx);
mutex_unlock(&hfi1_mutex);
}
......@@ -342,39 +340,43 @@ int hfi1_netdev_rx_destroy(struct hfi1_devdata *dd)
}
/**
* hfi1_netdev_alloc - Allocates netdev and private data. It is required
* because RMT index and MSI-X interrupt can be set only
* during driver initialization.
*
* hfi1_alloc_rx - Allocates the rx support structure
* @dd: hfi1 dev data
*
* Allocate the rx structure to support gathering the receive
* resources and the dummy netdev.
*
* Updates dd struct pointer upon success.
*
* Return: 0 (success) -error on failure
*
*/
int hfi1_netdev_alloc(struct hfi1_devdata *dd)
int hfi1_alloc_rx(struct hfi1_devdata *dd)
{
struct hfi1_netdev_priv *priv;
const int netdev_size = sizeof(*dd->dummy_netdev) +
sizeof(struct hfi1_netdev_priv);
struct hfi1_netdev_rx *rx;
dd_dev_info(dd, "allocating netdev size %d\n", netdev_size);
dd->dummy_netdev = kcalloc_node(1, netdev_size, GFP_KERNEL, dd->node);
dd_dev_info(dd, "allocating rx size %ld\n", sizeof(*rx));
rx = kzalloc_node(sizeof(*rx), GFP_KERNEL, dd->node);
if (!dd->dummy_netdev)
if (!rx)
return -ENOMEM;
rx->dd = dd;
init_dummy_netdev(&rx->rx_napi);
priv = hfi1_netdev_priv(dd->dummy_netdev);
priv->dd = dd;
xa_init(&priv->dev_tbl);
atomic_set(&priv->enabled, 0);
atomic_set(&priv->netdevs, 0);
xa_init(&rx->dev_tbl);
atomic_set(&rx->enabled, 0);
atomic_set(&rx->netdevs, 0);
dd->netdev_rx = rx;
return 0;
}
void hfi1_netdev_free(struct hfi1_devdata *dd)
void hfi1_free_rx(struct hfi1_devdata *dd)
{
if (dd->dummy_netdev) {
dd_dev_info(dd, "hfi1 netdev freed\n");
kfree(dd->dummy_netdev);
dd->dummy_netdev = NULL;
if (dd->netdev_rx) {
dd_dev_info(dd, "hfi1 rx freed\n");
kfree(dd->netdev_rx);
dd->netdev_rx = NULL;
}
}
......@@ -389,33 +391,33 @@ void hfi1_netdev_free(struct hfi1_devdata *dd)
*/
void hfi1_netdev_enable_queues(struct hfi1_devdata *dd)
{
struct hfi1_netdev_priv *priv;
struct hfi1_netdev_rx *rx;
if (!dd->dummy_netdev)
if (!dd->netdev_rx)
return;
priv = hfi1_netdev_priv(dd->dummy_netdev);
if (atomic_fetch_inc(&priv->enabled))
rx = dd->netdev_rx;
if (atomic_fetch_inc(&rx->enabled))
return;
mutex_lock(&hfi1_mutex);
enable_queues(priv);
enable_queues(rx);
mutex_unlock(&hfi1_mutex);
}
void hfi1_netdev_disable_queues(struct hfi1_devdata *dd)
{
struct hfi1_netdev_priv *priv;
struct hfi1_netdev_rx *rx;
if (!dd->dummy_netdev)
if (!dd->netdev_rx)
return;
priv = hfi1_netdev_priv(dd->dummy_netdev);
if (atomic_dec_if_positive(&priv->enabled))
rx = dd->netdev_rx;
if (atomic_dec_if_positive(&rx->enabled))
return;
mutex_lock(&hfi1_mutex);
disable_queues(priv);
disable_queues(rx);
mutex_unlock(&hfi1_mutex);
}
......@@ -431,9 +433,9 @@ void hfi1_netdev_disable_queues(struct hfi1_devdata *dd)
*/
int hfi1_netdev_add_data(struct hfi1_devdata *dd, int id, void *data)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
struct hfi1_netdev_rx *rx = dd->netdev_rx;
return xa_insert(&priv->dev_tbl, id, data, GFP_NOWAIT);
return xa_insert(&rx->dev_tbl, id, data, GFP_NOWAIT);
}
/**
......@@ -445,9 +447,9 @@ int hfi1_netdev_add_data(struct hfi1_devdata *dd, int id, void *data)
*/
void *hfi1_netdev_remove_data(struct hfi1_devdata *dd, int id)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
struct hfi1_netdev_rx *rx = dd->netdev_rx;
return xa_erase(&priv->dev_tbl, id);
return xa_erase(&rx->dev_tbl, id);
}
/**
......@@ -458,9 +460,9 @@ void *hfi1_netdev_remove_data(struct hfi1_devdata *dd, int id)
*/
void *hfi1_netdev_get_data(struct hfi1_devdata *dd, int id)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
struct hfi1_netdev_rx *rx = dd->netdev_rx;
return xa_load(&priv->dev_tbl, id);
return xa_load(&rx->dev_tbl, id);
}
/**
......@@ -471,11 +473,11 @@ void *hfi1_netdev_get_data(struct hfi1_devdata *dd, int id)
*/
void *hfi1_netdev_get_first_data(struct hfi1_devdata *dd, int *start_id)
{
struct hfi1_netdev_priv *priv = hfi1_netdev_priv(dd->dummy_netdev);
struct hfi1_netdev_rx *rx = dd->netdev_rx;
unsigned long index = *start_id;
void *ret;
ret = xa_find(&priv->dev_tbl, &index, UINT_MAX, XA_PRESENT);
ret = xa_find(&rx->dev_tbl, &index, UINT_MAX, XA_PRESENT);
*start_id = (int)index;
return ret;
}
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