Enable RX HW timestamp for PTP packets using CPTS FIFO
Subsystems (2)
v8 cover Enable RX HW timestamp for PTP packets using CPTS FIFO
The CPSW offers two mechanisms for communicating packet ingress timestamp information to the host.
The first mechanism is via the CPTS Event FIFO which records timestamp when triggered by certain events. One such event is the reception of an Ethernet packet with a specified EtherType field. This is used to capture ingress timestamps for PTP packets. With this mechanism the host must read the timestamp (from the CPTS FIFO) separately from the packet payload which is delivered via DMA.
In the second mechanism of timestamping, CPSW driver enables hardware timestamping for all received packets by setting the TSTAMP_EN bit in CPTS_CONTROL register, which directs the CPTS module to timestamp all received packets, followed by passing timestamp via DMA descriptors. This mechanism is responsible for triggering errata i2401:
"CPSW: Host Timestamps Cause CPSW Port to Lock up."
The errata affects all K3 SoCs. Link to errata for AM64x:
https://www.ti.com/lit/er/sprz457h/sprz457h.pdf
As a workaround we can use first mechanism to timestamp received packets.
Series is based on linux-next tagged next-20240419.
Link to v7:
https://lore.kernel.org/r/[email protected]/
Changes from v7 to v8:
- Removed empty lines between trailers in [PATCH v7 2/2] as suggested by Jakub.
Chintan Vankar (2):
net: ethernet: ti: am65-cpts: Enable RX HW timestamp for PTP packets
using CPTS FIFO
net: ethernet: ti: am65-cpsw/ethtool: Enable RX HW timestamp only for
PTP packets
drivers/net/ethernet/ti/am65-cpsw-ethtool.c | 13 ++- drivers/net/ethernet/ti/am65-cpsw-nuss.c | 51 +++++----- drivers/net/ethernet/ti/am65-cpts.c | 107 ++++++++++++++------ drivers/net/ethernet/ti/am65-cpts.h | 11 +- 4 files changed, 118 insertions(+), 64 deletions(-)
Signature · 1 line
2.34.1
PATCH v8 1/2 net: ethernet: ti: am65-cpts: Enable RX HW timestamp for PTP packets using CPTS FIFO
Add a new function "am65_cpts_rx_timestamp()" which checks for PTP packets from header and timestamps them.
Add another function "am65_cpts_find_rx_ts()" which finds CPTS FIFO Event to get the timestamp of received PTP packet.
---
Link to v7:
https://lore.kernel.org/r/[email protected]/
Changes from v7 to v8:
- No changes.
drivers/net/ethernet/ti/am65-cpts.c | 84 ++++++++++++++++++++++++++--- drivers/net/ethernet/ti/am65-cpts.h | 6 +++ 2 files changed, 83 insertions(+), 7 deletions(-)
drivers/net/ethernet/ti/am65-cpts.c +77 -7
@@ -275,15 +275,13 @@ static bool am65_cpts_fifo_pop_event(struct am65_cpts *cpts,return true; } -static int am65_cpts_fifo_read (struct am65_cpts *cpts) +static int __am65_cpts_fifo_read (struct am65_cpts *cpts) {struct ptp_clock_event pevent;struct am65_cpts_event *event;bool schedule = false;int i, type, ret =0 ; -unsigned long flags; -spin_lock_irqsave (&cpts->lock, flags);for (i =0 ; i <AM65_CPTS_FIFO_DEPTH ; i++) { event =list_first_entry_or_null (&cpts->pool, struct am65_cpts_event, list);
@@ -312,8 +310,7 @@ static int am65_cpts_fifo_read(struct am65_cpts *cpts) event->tmo = jiffies +msecs_to_jiffies (AM65_CPTS_EVENT_RX_TX_TIMEOUT ); -list_del_init (&event->list); -list_add_tail (&event->list, &cpts->events); +list_move_tail (&event->list, &cpts->events); dev_dbg(cpts->dev, "AM65_CPTS_EV_TX e1:%08x e2:%08x t:%lld\n",
@@ -356,14 +353,24 @@ static int am65_cpts_fifo_read(struct am65_cpts *cpts) } out: -spin_unlock_irqrestore (&cpts->lock, flags); -if (schedule)ptp_schedule_worker (cpts->ptp_clock,0 );return ret; } +static int am65_cpts_fifo_read (struct am65_cpts *cpts) +{ +unsigned long flags; +int ret =0 ; + +spin_lock_irqsave (&cpts->lock, flags); + ret =__am65_cpts_fifo_read (cpts); +spin_unlock_irqrestore (&cpts->lock, flags); + +return ret; +} +static u64 am65_cpts_gettime (struct am65_cpts *cpts,struct ptp_system_timestamp *sts) {
@@ -906,6 +913,69 @@ static int am65_skb_get_mtype_seqid(struct sk_buff *skb, u32 *mtype_seqid)return 1 ; } +static u64 am65_cpts_find_rx_ts (struct am65_cpts *cpts,u32 skb_mtype_seqid) +{ +struct list_head *this, *next; +struct am65_cpts_event *event; +unsigned long flags; +u32 mtype_seqid; +u64 ns =0 ; + +spin_lock_irqsave (&cpts->lock, flags); +__am65_cpts_fifo_read (cpts); +list_for_each_safe (this, next, &cpts->events) { + event = list_entry(this, struct am65_cpts_event, list); +if (time_after (jiffies, event->tmo)) { +list_move (&event->list, &cpts->pool); +continue ; + } + + mtype_seqid = event->event1 & + (AM65_CPTS_EVENT_1_MESSAGE_TYPE_MASK | +AM65_CPTS_EVENT_1_SEQUENCE_ID_MASK | +AM65_CPTS_EVENT_1_EVENT_TYPE_MASK ); + +if (mtype_seqid == skb_mtype_seqid) { + ns = event->timestamp; +list_move (&event->list, &cpts->pool); +break ; + } + } +spin_unlock_irqrestore (&cpts->lock, flags); + +return ns; +} + +void am65_cpts_rx_timestamp (struct am65_cpts *cpts,struct sk_buff *skb) +{ +struct am65_cpts_skb_cb_data *skb_cb = (struct am65_cpts_skb_cb_data *)skb->cb; +struct skb_shared_hwtstamps *ssh; +int ret; +u64 ns; + +/* am65_cpts_rx_timestamp() is called before eth_type_trans(), so +* skb MAC Hdr properties are not configured yet. Hence need to +* reset skb MAC header here +*/ +skb_reset_mac_header (skb); + ret =am65_skb_get_mtype_seqid (skb, &skb_cb->skb_mtype_seqid); +if (!ret) +return ;/* if not PTP class packet */ + + skb_cb->skb_mtype_seqid |= (AM65_CPTS_EV_RX <<AM65_CPTS_EVENT_1_EVENT_TYPE_SHIFT ); + +dev_dbg (cpts->dev,"%s mtype seqid %08x\n" , __func__, skb_cb->skb_mtype_seqid); + + ns =am65_cpts_find_rx_ts (cpts, skb_cb->skb_mtype_seqid); +if (!ns) +return ; + + ssh =skb_hwtstamps (skb); +memset (ssh,0 ,sizeof (*ssh)); + ssh->hwtstamp =ns_to_ktime (ns); +} +EXPORT_SYMBOL_GPL (am65_cpts_rx_timestamp); +/** * am65_cpts_tx_timestamp - save tx packet for timestamping * @cpts: cpts handle
drivers/net/ethernet/ti/am65-cpts.h +6 -0
@@ -22,6 +22,7 @@ void am65_cpts_release(struct am65_cpts *cpts);struct am65_cpts *am65_cpts_create (struct device *dev,void __iomem *regs,struct device_node *node);int am65_cpts_phc_index (struct am65_cpts *cpts); +void am65_cpts_rx_timestamp (struct am65_cpts *cpts,struct sk_buff *skb);void am65_cpts_tx_timestamp (struct am65_cpts *cpts,struct sk_buff *skb);void am65_cpts_prep_tx_timestamp (struct am65_cpts *cpts,struct sk_buff *skb);void am65_cpts_rx_enable (struct am65_cpts *cpts,bool en);
@@ -48,6 +49,11 @@ static inline int am65_cpts_phc_index(struct am65_cpts *cpts)return -1 ; } +static inline void am65_cpts_rx_timestamp (struct am65_cpts *cpts, +struct sk_buff *skb) +{ +} +static inline void am65_cpts_tx_timestamp (struct am65_cpts *cpts,struct sk_buff *skb) {
Signature · 1 line
2.34.1
PATCH v8 2/2 net: ethernet: ti: am65-cpsw/ethtool: Enable RX HW timestamp only for PTP packets
In reply to Chintan VankarIn the current mechanism of timestamping, am65-cpsw-nuss driver enables hardware timestamping for all received packets by setting the TSTAMP_EN bit in CPTS_CONTROL register, which directs the CPTS module to timestamp all received packets, followed by passing timestamp via DMA descriptors. This mechanism causes CPSW Port to Lock up.
To prevent port lock up, don't enable rx packet timestamping by setting TSTAMP_EN bit in CPTS_CONTROL register. The workaround for timestamping received packets is to utilize the CPTS Event FIFO that records timestamps corresponding to certain events. The CPTS module is configured to generate timestamps for Multicast Ethernet, UDP/IPv4 and UDP/IPv6 PTP packets.
Update supported hwtstamp_rx_filters values for CPSW's timestamping capability.
---
Link to v7:
https://lore.kernel.org/r/[email protected]/
Changes from v7 to v8:
- Removed empty lines between trailers as suggested by Jakub.
drivers/net/ethernet/ti/am65-cpsw-ethtool.c | 13 +++++- drivers/net/ethernet/ti/am65-cpsw-nuss.c | 51 ++++++++++----------- drivers/net/ethernet/ti/am65-cpts.c | 23 ---------- drivers/net/ethernet/ti/am65-cpts.h | 5 -- 4 files changed, 35 insertions(+), 57 deletions(-)
drivers/net/ethernet/ti/am65-cpsw-ethtool.c +12 -1
@@ -695,6 +695,17 @@ static int am65_cpsw_get_ethtool_ts_info(struct net_device *ndev, struct ethtool_ts_info *info) {struct am65_cpsw_common *common =am65_ndev_to_common (ndev); +unsigned int ptp_v2_filter; + + ptp_v2_filter =BIT (HWTSTAMP_FILTER_PTP_V2_L4_EVENT ) | +BIT (HWTSTAMP_FILTER_PTP_V2_L4_SYNC ) | +BIT (HWTSTAMP_FILTER_PTP_V2_L4_DELAY_REQ ) | +BIT (HWTSTAMP_FILTER_PTP_V2_L2_EVENT ) | +BIT (HWTSTAMP_FILTER_PTP_V2_L2_SYNC ) | +BIT (HWTSTAMP_FILTER_PTP_V2_L2_DELAY_REQ ) | +BIT (HWTSTAMP_FILTER_PTP_V2_EVENT ) | +BIT (HWTSTAMP_FILTER_PTP_V2_SYNC ) | +BIT (HWTSTAMP_FILTER_PTP_V2_DELAY_REQ );if (!IS_ENABLED (CONFIG_TI_K3_AM65_CPTS ))return ethtool_op_get_ts_info (ndev, info);
@@ -708,7 +719,7 @@ static int am65_cpsw_get_ethtool_ts_info(struct net_device *ndev,SOF_TIMESTAMPING_RAW_HARDWARE ; info->phc_index =am65_cpts_phc_index (common->cpts); info->tx_types =BIT (HWTSTAMP_TX_OFF ) |BIT (HWTSTAMP_TX_ON ); - info->rx_filters =BIT (HWTSTAMP_FILTER_NONE ) |BIT (HWTSTAMP_FILTER_ALL ); + info->rx_filters =BIT (HWTSTAMP_FILTER_NONE ) | ptp_v2_filter;return 0 ; }
drivers/net/ethernet/ti/am65-cpsw-nuss.c +23 -28
116 lines not on this page · raw message
drivers/net/ethernet/ti/am65-cpts.c +0 -23
@@ -866,29 +866,6 @@ static long am65_cpts_ts_work(struct ptp_clock_info *ptp)return delay; } -/** -* am65_cpts_rx_enable - enable rx timestamping -* @cpts: cpts handle -* @en: enable -* -* This functions enables rx packets timestamping. The CPTS can timestamp all -* rx packets. -*/ -void am65_cpts_rx_enable (struct am65_cpts *cpts,bool en) -{ -u32 val; - -mutex_lock (&cpts->ptp_clk_lock); - val =am65_cpts_read32 (cpts, control); -if (en) - val |=AM65_CPTS_CONTROL_TSTAMP_EN ; -else - val &= ~AM65_CPTS_CONTROL_TSTAMP_EN ; -am65_cpts_write32 (cpts, val, control); -mutex_unlock (&cpts->ptp_clk_lock); -} -EXPORT_SYMBOL_GPL (am65_cpts_rx_enable); -static int am65_skb_get_mtype_seqid (struct sk_buff *skb,u32 *mtype_seqid) {unsigned int ptp_class =ptp_classify_raw (skb);
drivers/net/ethernet/ti/am65-cpts.h +0 -5
@@ -25,7 +25,6 @@ int am65_cpts_phc_index(struct am65_cpts *cpts);void am65_cpts_rx_timestamp (struct am65_cpts *cpts,struct sk_buff *skb);void am65_cpts_tx_timestamp (struct am65_cpts *cpts,struct sk_buff *skb);void am65_cpts_prep_tx_timestamp (struct am65_cpts *cpts,struct sk_buff *skb); -void am65_cpts_rx_enable (struct am65_cpts *cpts,bool en);u64 am65_cpts_ns_gettime (struct am65_cpts *cpts);int am65_cpts_estf_enable (struct am65_cpts *cpts,int idx,struct am65_cpts_estf_cfg *cfg);
@@ -64,10 +63,6 @@ static inline void am65_cpts_prep_tx_timestamp(struct am65_cpts *cpts, { } -static inline void am65_cpts_rx_enable (struct am65_cpts *cpts,bool en) -{ -} -static inline s64 am65_cpts_ns_gettime (struct am65_cpts *cpts) {return 0 ;
Signature · 1 line
2.34.1
Quoted 23 lines · Chintan Vankar
The CPSW offers two mechanisms for communicating packet ingress timestamp information to the host.
The first mechanism is via the CPTS Event FIFO which records timestamp when triggered by certain events. One such event is the reception of an Ethernet packet with a specified EtherType field. This is used to capture ingress timestamps for PTP packets. With this mechanism the host must read the timestamp (from the CPTS FIFO) separately from the packet payload which is delivered via DMA.
In the second mechanism of timestamping, CPSW driver enables hardware timestamping for all received packets by setting the
TSTAMP_ENbit inCPTS_CONTROLregister, which directs the CPTS module to timestamp all received packets, followed by passing timestamp via DMA descriptors. This mechanism is responsible for triggering errata i2401:
"CPSW: Host Timestamps Cause CPSW Port to Lock up."The errata affects all K3 SoCs. Link to errata for AM64x:
https://www.ti.com/lit/er/sprz457h/sprz457h.pdfAs a workaround we can use first mechanism to timestamp received packets.
Quoted 9 lines
Series is based on linux-next tagged next-20240419.
Link to v7:
https://lore.kernel.org/r/[email protected]/Changes from v7 to v8:
- Removed empty lines between trailers in [PATCH v7 2/2] as suggested by Jakub.
Is it possible to merge this series if no further comment ?
Quoted 12 lines
Chintan Vankar (2):
net: ethernet: ti: am65-cpts: Enable RX HW timestamp for PTP packets
using CPTS FIFO
net: ethernet: ti: am65-cpsw/ethtool: Enable RX HW timestamp only for
PTP packetsdrivers/net/ethernet/ti/am65-cpsw-ethtool.c | 13 ++- drivers/net/ethernet/ti/am65-cpsw-nuss.c | 51 +++++----- drivers/net/ethernet/ti/am65-cpts.c | 107 ++++++++++++++------ drivers/net/ethernet/ti/am65-cpts.h | 11 +- 4 files changed, 118 insertions(+), 64 deletions(-)
Jakub Kicinski
maintainerQuoted 1 line · On Mon, 22 Apr 2024 17:09:41 +0530 Chintan Vankar wrote:
Is it possible to merge this series if no further comment ?
Perhaps you are blocked by us not merging this, yet.
In that case, to avoid sitting idly, may I politely suggest reading the Linux kernel process documentation. Especially the part which explains whether it's polite to ping maintainers (after less than two work days).
This series was applied to netdev/net-next.git (main) by Paolo Abeni <[email protected]>:
Hello:
This series was applied to netdev/net-next.git (main) by Paolo Abeni <[email protected]>:
Quoted 11 lines · On Fri, 19 Apr 2024 13:56:24 +0530 you wrote:
The CPSW offers two mechanisms for communicating packet ingress timestamp information to the host.
The first mechanism is via the CPTS Event FIFO which records timestamp when triggered by certain events. One such event is the reception of an Ethernet packet with a specified EtherType field. This is used to capture ingress timestamps for PTP packets. With this mechanism the host must read the timestamp (from the CPTS FIFO) separately from the packet payload which is delivered via DMA.
[...]
Here is the summary with links:
- [net-next,v8,1/2] net: ethernet: ti: am65-cpts: Enable RX HW timestamp for PTP packets using CPTS FIFO
https://git.kernel.org/netdev/net-next/c/c459f606f66d - [net-next,v8,2/2] net: ethernet: ti: am65-cpsw/ethtool: Enable RX HW timestamp only for PTP packets https://git.kernel.org/netdev/net-next/c/c03a6fd39826
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html