iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

vmxnet3: add support for 32 Tx/Rx queues

Browse Source 
Currently, vmxnet3 supports maximum of 8 Tx/Rx queues. With increase
in number of vcpus on a VM, to achieve better performance and utilize
idle vcpus, we need to increase the max number of queues supported.

This patch enhances vmxnet3 to support maximum of 32 Tx/Rx queues.
Increasing the Rx queues also increases the probability of distrubuting
the traffic from different flows to different queues with RSS.

Signed-off-by: Ronak Doshi <doshir@vmware.com>
Acked-by: Guolin Yang <gyang@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Ronak Doshi 2021-07-16 15:36:21 -07:00 committed by David S. Miller
parent 69dbef0d1c
commit 39f9895a00
3 changed files with 151 additions and 60 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
drivers/net/vmxnet3/vmxnet3_defs.h
View File

@ -98,6 +98,9 @@ enum {
VMXNET3_CMD_GET_TXDATA_DESC_SIZE, VMXNET3_CMD_GET_TXDATA_DESC_SIZE,
VMXNET3_CMD_GET_COALESCE, VMXNET3_CMD_GET_COALESCE,
VMXNET3_CMD_GET_RSS_FIELDS, VMXNET3_CMD_GET_RSS_FIELDS,
VMXNET3_CMD_GET_RESERVED2,
VMXNET3_CMD_GET_RESERVED3,
VMXNET3_CMD_GET_MAX_QUEUES_CONF,
}; };
/* /*
@ -533,6 +536,13 @@ enum vmxnet3_intr_type {
/* addition 1 for events */ /* addition 1 for events */
#define VMXNET3_MAX_INTRS 25 #define VMXNET3_MAX_INTRS 25
/* Version 6 and later will use below macros */
#define VMXNET3_EXT_MAX_TX_QUEUES 32
#define VMXNET3_EXT_MAX_RX_QUEUES 32
/* addition 1 for events */
#define VMXNET3_EXT_MAX_INTRS 65
#define VMXNET3_FIRST_SET_INTRS 64
/* value of intrCtrl */ /* value of intrCtrl */
#define VMXNET3_IC_DISABLE_ALL 0x1 /* bit 0 */ #define VMXNET3_IC_DISABLE_ALL 0x1 /* bit 0 */
@ -547,6 +557,19 @@ struct Vmxnet3_IntrConf {
__le32 reserved[2]; __le32 reserved[2];
}; };
struct Vmxnet3_IntrConfExt {
u8 autoMask;
u8 numIntrs; /* # of interrupts */
u8 eventIntrIdx;
u8 reserved;
__le32 intrCtrl;
__le32 reserved1;
u8 modLevels[VMXNET3_EXT_MAX_INTRS]; /* moderation level for
* each intr
*/
u8 reserved2[3];
};
/* one bit per VLAN ID, the size is in the units of u32 */ /* one bit per VLAN ID, the size is in the units of u32 */
#define VMXNET3_VFT_SIZE (4096 / (sizeof(u32) * 8)) #define VMXNET3_VFT_SIZE (4096 / (sizeof(u32) * 8))
@ -719,11 +742,16 @@ struct Vmxnet3_DSDevRead {
struct Vmxnet3_VariableLenConfDesc pluginConfDesc; struct Vmxnet3_VariableLenConfDesc pluginConfDesc;
}; };
struct Vmxnet3_DSDevReadExt {
/* read-only region for device, read by dev in response to a SET cmd */
struct Vmxnet3_IntrConfExt intrConfExt;
};
/* All structures in DriverShared are padded to multiples of 8 bytes */ /* All structures in DriverShared are padded to multiples of 8 bytes */
struct Vmxnet3_DriverShared { struct Vmxnet3_DriverShared {
__le32 magic; __le32 magic;
/* make devRead start at 64bit boundaries */ /* make devRead start at 64bit boundaries */
__le32 pad; __le32 size; /* size of DriverShared */
struct Vmxnet3_DSDevRead devRead; struct Vmxnet3_DSDevRead devRead;
__le32 ecr; __le32 ecr;
__le32 reserved; __le32 reserved;
@ -734,6 +762,7 @@ struct Vmxnet3_DriverShared {
* command * command
*/ */
} cu; } cu;
struct Vmxnet3_DSDevReadExt devReadExt;
}; };

156
drivers/net/vmxnet3/vmxnet3_drv.c
View File

@ -2460,6 +2460,7 @@ vmxnet3_setup_driver_shared(struct vmxnet3_adapter *adapter)
{ {
struct Vmxnet3_DriverShared *shared = adapter->shared; struct Vmxnet3_DriverShared *shared = adapter->shared;
struct Vmxnet3_DSDevRead *devRead = &shared->devRead; struct Vmxnet3_DSDevRead *devRead = &shared->devRead;
struct Vmxnet3_DSDevReadExt *devReadExt = &shared->devReadExt;
struct Vmxnet3_TxQueueConf *tqc; struct Vmxnet3_TxQueueConf *tqc;
struct Vmxnet3_RxQueueConf *rqc; struct Vmxnet3_RxQueueConf *rqc;
int i; int i;
@ -2572,6 +2573,8 @@ vmxnet3_setup_driver_shared(struct vmxnet3_adapter *adapter)
#endif /* VMXNET3_RSS */ #endif /* VMXNET3_RSS */
/* intr settings */ /* intr settings */
if (!VMXNET3_VERSION_GE_6(adapter) ||
!adapter->queuesExtEnabled) {
devRead->intrConf.autoMask = adapter->intr.mask_mode == devRead->intrConf.autoMask = adapter->intr.mask_mode ==
VMXNET3_IMM_AUTO; VMXNET3_IMM_AUTO;
devRead->intrConf.numIntrs = adapter->intr.num_intrs; devRead->intrConf.numIntrs = adapter->intr.num_intrs;
@ -2580,6 +2583,16 @@ vmxnet3_setup_driver_shared(struct vmxnet3_adapter *adapter)
devRead->intrConf.eventIntrIdx = adapter->intr.event_intr_idx; devRead->intrConf.eventIntrIdx = adapter->intr.event_intr_idx;
devRead->intrConf.intrCtrl |= cpu_to_le32(VMXNET3_IC_DISABLE_ALL); devRead->intrConf.intrCtrl |= cpu_to_le32(VMXNET3_IC_DISABLE_ALL);
} else {
devReadExt->intrConfExt.autoMask = adapter->intr.mask_mode ==
VMXNET3_IMM_AUTO;
devReadExt->intrConfExt.numIntrs = adapter->intr.num_intrs;
for (i = 0; i < adapter->intr.num_intrs; i++)
devReadExt->intrConfExt.modLevels[i] = adapter->intr.mod_levels[i];
devReadExt->intrConfExt.eventIntrIdx = adapter->intr.event_intr_idx;
devReadExt->intrConfExt.intrCtrl |= cpu_to_le32(VMXNET3_IC_DISABLE_ALL);
}
/* rx filter settings */ /* rx filter settings */
devRead->rxFilterConf.rxMode = 0; devRead->rxFilterConf.rxMode = 0;
@ -2717,6 +2730,7 @@ vmxnet3_activate_dev(struct vmxnet3_adapter *adapter)
* tx queue if the link is up. * tx queue if the link is up.
*/ */
vmxnet3_check_link(adapter, true); vmxnet3_check_link(adapter, true);
netif_tx_wake_all_queues(adapter->netdev);
for (i = 0; i < adapter->num_rx_queues; i++) for (i = 0; i < adapter->num_rx_queues; i++)
napi_enable(&adapter->rx_queue[i].napi); napi_enable(&adapter->rx_queue[i].napi);
vmxnet3_enable_all_intrs(adapter); vmxnet3_enable_all_intrs(adapter);
@ -3372,6 +3386,8 @@ vmxnet3_probe_device(struct pci_dev *pdev,
int size; int size;
int num_tx_queues; int num_tx_queues;
int num_rx_queues; int num_rx_queues;
int queues;
unsigned long flags;
if (!pci_msi_enabled()) if (!pci_msi_enabled())
enable_mq = 0; enable_mq = 0;
@ -3394,10 +3410,6 @@ vmxnet3_probe_device(struct pci_dev *pdev,
num_tx_queues = rounddown_pow_of_two(num_tx_queues); num_tx_queues = rounddown_pow_of_two(num_tx_queues);
netdev = alloc_etherdev_mq(sizeof(struct vmxnet3_adapter), netdev = alloc_etherdev_mq(sizeof(struct vmxnet3_adapter),
max(num_tx_queues, num_rx_queues)); max(num_tx_queues, num_rx_queues));
dev_info(&pdev->dev,
"# of Tx queues : %d, # of Rx queues : %d\n",
num_tx_queues, num_rx_queues);
if (!netdev) if (!netdev)
return -ENOMEM; return -ENOMEM;
@ -3447,45 +3459,6 @@ vmxnet3_probe_device(struct pci_dev *pdev,
goto err_alloc_shared; goto err_alloc_shared;
} }
adapter->num_rx_queues = num_rx_queues;
adapter->num_tx_queues = num_tx_queues;
adapter->rx_buf_per_pkt = 1;
size = sizeof(struct Vmxnet3_TxQueueDesc) * adapter->num_tx_queues;
size += sizeof(struct Vmxnet3_RxQueueDesc) * adapter->num_rx_queues;
adapter->tqd_start = dma_alloc_coherent(&adapter->pdev->dev, size,
&adapter->queue_desc_pa,
GFP_KERNEL);
if (!adapter->tqd_start) {
dev_err(&pdev->dev, "Failed to allocate memory\n");
err = -ENOMEM;
goto err_alloc_queue_desc;
}
adapter->rqd_start = (struct Vmxnet3_RxQueueDesc *)(adapter->tqd_start +
adapter->num_tx_queues);
adapter->pm_conf = dma_alloc_coherent(&adapter->pdev->dev,
sizeof(struct Vmxnet3_PMConf),
&adapter->pm_conf_pa,
GFP_KERNEL);
if (adapter->pm_conf == NULL) {
err = -ENOMEM;
goto err_alloc_pm;
}
#ifdef VMXNET3_RSS
adapter->rss_conf = dma_alloc_coherent(&adapter->pdev->dev,
sizeof(struct UPT1_RSSConf),
&adapter->rss_conf_pa,
GFP_KERNEL);
if (adapter->rss_conf == NULL) {
err = -ENOMEM;
goto err_alloc_rss;
}
#endif /* VMXNET3_RSS */
err = vmxnet3_alloc_pci_resources(adapter); err = vmxnet3_alloc_pci_resources(adapter);
if (err < 0) if (err < 0)
goto err_alloc_pci; goto err_alloc_pci;
@ -3529,6 +3502,75 @@ vmxnet3_probe_device(struct pci_dev *pdev,
goto err_ver; goto err_ver;
} }
if (VMXNET3_VERSION_GE_6(adapter)) {
spin_lock_irqsave(&adapter->cmd_lock, flags);
VMXNET3_WRITE_BAR1_REG(adapter, VMXNET3_REG_CMD,
VMXNET3_CMD_GET_MAX_QUEUES_CONF);
queues = VMXNET3_READ_BAR1_REG(adapter, VMXNET3_REG_CMD);
spin_unlock_irqrestore(&adapter->cmd_lock, flags);
if (queues > 0) {
adapter->num_rx_queues = min(num_rx_queues, ((queues >> 8) & 0xff));
adapter->num_tx_queues = min(num_tx_queues, (queues & 0xff));
} else {
adapter->num_rx_queues = min(num_rx_queues,
VMXNET3_DEVICE_DEFAULT_RX_QUEUES);
adapter->num_tx_queues = min(num_tx_queues,
VMXNET3_DEVICE_DEFAULT_TX_QUEUES);
}
if (adapter->num_rx_queues > VMXNET3_MAX_RX_QUEUES ||
adapter->num_tx_queues > VMXNET3_MAX_TX_QUEUES) {
adapter->queuesExtEnabled = true;
} else {
adapter->queuesExtEnabled = false;
}
} else {
adapter->queuesExtEnabled = false;
adapter->num_rx_queues = min(num_rx_queues,
VMXNET3_DEVICE_DEFAULT_RX_QUEUES);
adapter->num_tx_queues = min(num_tx_queues,
VMXNET3_DEVICE_DEFAULT_TX_QUEUES);
}
dev_info(&pdev->dev,
"# of Tx queues : %d, # of Rx queues : %d\n",
adapter->num_tx_queues, adapter->num_rx_queues);
adapter->rx_buf_per_pkt = 1;
size = sizeof(struct Vmxnet3_TxQueueDesc) * adapter->num_tx_queues;
size += sizeof(struct Vmxnet3_RxQueueDesc) * adapter->num_rx_queues;
adapter->tqd_start = dma_alloc_coherent(&adapter->pdev->dev, size,
&adapter->queue_desc_pa,
GFP_KERNEL);
if (!adapter->tqd_start) {
dev_err(&pdev->dev, "Failed to allocate memory\n");
err = -ENOMEM;
goto err_ver;
}
adapter->rqd_start = (struct Vmxnet3_RxQueueDesc *)(adapter->tqd_start +
adapter->num_tx_queues);
adapter->pm_conf = dma_alloc_coherent(&adapter->pdev->dev,
sizeof(struct Vmxnet3_PMConf),
&adapter->pm_conf_pa,
GFP_KERNEL);
if (adapter->pm_conf == NULL) {
err = -ENOMEM;
goto err_alloc_pm;
}
#ifdef VMXNET3_RSS
adapter->rss_conf = dma_alloc_coherent(&adapter->pdev->dev,
sizeof(struct UPT1_RSSConf),
&adapter->rss_conf_pa,
GFP_KERNEL);
if (adapter->rss_conf == NULL) {
err = -ENOMEM;
goto err_alloc_rss;
}
#endif /* VMXNET3_RSS */
if (VMXNET3_VERSION_GE_3(adapter)) { if (VMXNET3_VERSION_GE_3(adapter)) {
adapter->coal_conf = adapter->coal_conf =
dma_alloc_coherent(&adapter->pdev->dev, dma_alloc_coherent(&adapter->pdev->dev,
@ -3538,7 +3580,7 @@ vmxnet3_probe_device(struct pci_dev *pdev,
GFP_KERNEL); GFP_KERNEL);
if (!adapter->coal_conf) { if (!adapter->coal_conf) {
err = -ENOMEM; err = -ENOMEM;
goto err_ver; goto err_coal_conf;
} }
adapter->coal_conf->coalMode = VMXNET3_COALESCE_DISABLED; adapter->coal_conf->coalMode = VMXNET3_COALESCE_DISABLED;
adapter->default_coal_mode = true; adapter->default_coal_mode = true;
@ -3621,9 +3663,7 @@ err_register:
adapter->coal_conf, adapter->coal_conf_pa); adapter->coal_conf, adapter->coal_conf_pa);
} }
vmxnet3_free_intr_resources(adapter); vmxnet3_free_intr_resources(adapter);
err_ver: err_coal_conf:
vmxnet3_free_pci_resources(adapter);
err_alloc_pci:
#ifdef VMXNET3_RSS #ifdef VMXNET3_RSS
dma_free_coherent(&adapter->pdev->dev, sizeof(struct UPT1_RSSConf), dma_free_coherent(&adapter->pdev->dev, sizeof(struct UPT1_RSSConf),
adapter->rss_conf, adapter->rss_conf_pa); adapter->rss_conf, adapter->rss_conf_pa);
@ -3634,7 +3674,9 @@ err_alloc_rss:
err_alloc_pm: err_alloc_pm:
dma_free_coherent(&adapter->pdev->dev, size, adapter->tqd_start, dma_free_coherent(&adapter->pdev->dev, size, adapter->tqd_start,
adapter->queue_desc_pa); adapter->queue_desc_pa);
err_alloc_queue_desc: err_ver:
vmxnet3_free_pci_resources(adapter);
err_alloc_pci:
dma_free_coherent(&adapter->pdev->dev, dma_free_coherent(&adapter->pdev->dev,
sizeof(struct Vmxnet3_DriverShared), sizeof(struct Vmxnet3_DriverShared),
adapter->shared, adapter->shared_pa); adapter->shared, adapter->shared_pa);
@ -3653,7 +3695,8 @@ vmxnet3_remove_device(struct pci_dev *pdev)
struct net_device *netdev = pci_get_drvdata(pdev); struct net_device *netdev = pci_get_drvdata(pdev);
struct vmxnet3_adapter *adapter = netdev_priv(netdev); struct vmxnet3_adapter *adapter = netdev_priv(netdev);
int size = 0; int size = 0;
int num_rx_queues; int num_rx_queues, rx_queues;
unsigned long flags;
#ifdef VMXNET3_RSS #ifdef VMXNET3_RSS
if (enable_mq) if (enable_mq)
@ -3663,6 +3706,21 @@ vmxnet3_remove_device(struct pci_dev *pdev)
#endif #endif
num_rx_queues = 1; num_rx_queues = 1;
num_rx_queues = rounddown_pow_of_two(num_rx_queues); num_rx_queues = rounddown_pow_of_two(num_rx_queues);
if (VMXNET3_VERSION_GE_6(adapter)) {
spin_lock_irqsave(&adapter->cmd_lock, flags);
VMXNET3_WRITE_BAR1_REG(adapter, VMXNET3_REG_CMD,
VMXNET3_CMD_GET_MAX_QUEUES_CONF);
rx_queues = VMXNET3_READ_BAR1_REG(adapter, VMXNET3_REG_CMD);
spin_unlock_irqrestore(&adapter->cmd_lock, flags);
if (rx_queues > 0)
rx_queues = (rx_queues >> 8) & 0xff;
else
rx_queues = min(num_rx_queues, VMXNET3_DEVICE_DEFAULT_RX_QUEUES);
num_rx_queues = min(num_rx_queues, rx_queues);
} else {
num_rx_queues = min(num_rx_queues,
VMXNET3_DEVICE_DEFAULT_RX_QUEUES);
}
cancel_work_sync(&adapter->work); cancel_work_sync(&adapter->work);

10
drivers/net/vmxnet3/vmxnet3_int.h
View File

@ -303,15 +303,18 @@ struct vmxnet3_rx_queue {
struct vmxnet3_rq_driver_stats stats; struct vmxnet3_rq_driver_stats stats;
} __attribute__((__aligned__(SMP_CACHE_BYTES))); } __attribute__((__aligned__(SMP_CACHE_BYTES)));
#define VMXNET3_DEVICE_MAX_TX_QUEUES 8 #define VMXNET3_DEVICE_MAX_TX_QUEUES 32
#define VMXNET3_DEVICE_MAX_RX_QUEUES 8 /* Keep this value as a power of 2 */ #define VMXNET3_DEVICE_MAX_RX_QUEUES 32 /* Keep this value as a power of 2 */
#define VMXNET3_DEVICE_DEFAULT_TX_QUEUES 8
#define VMXNET3_DEVICE_DEFAULT_RX_QUEUES 8 /* Keep this value as a power of 2 */
/* Should be less than UPT1_RSS_MAX_IND_TABLE_SIZE */ /* Should be less than UPT1_RSS_MAX_IND_TABLE_SIZE */
#define VMXNET3_RSS_IND_TABLE_SIZE (VMXNET3_DEVICE_MAX_RX_QUEUES * 4) #define VMXNET3_RSS_IND_TABLE_SIZE (VMXNET3_DEVICE_MAX_RX_QUEUES * 4)
#define VMXNET3_LINUX_MAX_MSIX_VECT (VMXNET3_DEVICE_MAX_TX_QUEUES + \ #define VMXNET3_LINUX_MAX_MSIX_VECT (VMXNET3_DEVICE_MAX_TX_QUEUES + \
VMXNET3_DEVICE_MAX_RX_QUEUES + 1) VMXNET3_DEVICE_MAX_RX_QUEUES + 1)
#define VMXNET3_LINUX_MIN_MSIX_VECT 2 /* 1 for tx-rx pair and 1 for event */ #define VMXNET3_LINUX_MIN_MSIX_VECT 3 /* 1 for tx, 1 for rx pair and 1 for event */
struct vmxnet3_intr { struct vmxnet3_intr {
@ -398,6 +401,7 @@ struct vmxnet3_adapter {
dma_addr_t adapter_pa; dma_addr_t adapter_pa;
dma_addr_t pm_conf_pa; dma_addr_t pm_conf_pa;
dma_addr_t rss_conf_pa; dma_addr_t rss_conf_pa;
bool queuesExtEnabled;
}; };
#define VMXNET3_WRITE_BAR0_REG(adapter, reg, val) \ #define VMXNET3_WRITE_BAR0_REG(adapter, reg, val) \