On 6/29/24 11:31, Vincent MAILHOL wrote:
On Sat. 29 June 2024 at 04:56, Jimmy Assarsson <extja@xxxxxxxxxx> wrote:
Add, struct kvaser_cmd_tx_ack, for standard Tx ACK commands.
Expand kvaser_usb_hydra_ktime_from_cmd() to extract timestamps from both
standard and extended Tx ACK commands. Unsupported commands are silently
ignored, and 0 is returned.
Signed-off-by: Jimmy Assarsson <extja@xxxxxxxxxx>
---
.../net/can/usb/kvaser_usb/kvaser_usb_hydra.c | 23 ++++++++++++++++---
1 file changed, 20 insertions(+), 3 deletions(-)
diff --git a/drivers/net/can/usb/kvaser_usb/kvaser_usb_hydra.c b/drivers/net/can/usb/kvaser_usb/kvaser_usb_hydra.c
index a971fcb6158a..0be1cfe8d964 100644
--- a/drivers/net/can/usb/kvaser_usb/kvaser_usb_hydra.c
+++ b/drivers/net/can/usb/kvaser_usb/kvaser_usb_hydra.c
@@ -261,6 +261,15 @@ struct kvaser_cmd_tx_can {
u8 reserved[11];
} __packed;
+struct kvaser_cmd_tx_ack {
+ __le32 id;
+ u8 data[8];
+ u8 dlc;
+ u8 flags;
+ __le16 timestamp[3];
+ u8 reserved0[8];
+} __packed;
+
struct kvaser_cmd_header {
u8 cmd_no;
/* The destination HE address is stored in 0..5 of he_addr.
@@ -297,6 +306,7 @@ struct kvaser_cmd {
struct kvaser_cmd_rx_can rx_can;
struct kvaser_cmd_tx_can tx_can;
+ struct kvaser_cmd_tx_ack tx_ack;
} __packed;
} __packed;
@@ -525,16 +535,23 @@ kvaser_usb_hydra_net_priv_from_cmd(const struct kvaser_usb *dev,
static ktime_t kvaser_usb_hydra_ktime_from_cmd(const struct kvaser_usb_dev_cfg *cfg,
const struct kvaser_cmd *cmd)
{
- u64 ticks;
+ u64 ticks = 0;
if (cmd->header.cmd_no == CMD_EXTENDED) {
struct kvaser_cmd_ext *cmd_ext = (struct kvaser_cmd_ext *)cmd;
- ticks = le64_to_cpu(cmd_ext->rx_can.timestamp);
- } else {
+ if (cmd_ext->cmd_no_ext == CMD_RX_MESSAGE_FD)
+ ticks = le64_to_cpu(cmd_ext->rx_can.timestamp);
+ else if (cmd_ext->cmd_no_ext == CMD_TX_ACKNOWLEDGE_FD)
+ ticks = le64_to_cpu(cmd_ext->tx_ack.timestamp);
+ } else if (cmd->header.cmd_no == CMD_RX_MESSAGE) {
ticks = le16_to_cpu(cmd->rx_can.timestamp[0]);
ticks += (u64)(le16_to_cpu(cmd->rx_can.timestamp[1])) << 16;
ticks += (u64)(le16_to_cpu(cmd->rx_can.timestamp[2])) << 32;
+ } else if (cmd->header.cmd_no == CMD_TX_ACKNOWLEDGE) {
+ ticks = le16_to_cpu(cmd->tx_ack.timestamp[0]);
+ ticks += (u64)(le16_to_cpu(cmd->tx_ack.timestamp[1])) << 16;
+ ticks += (u64)(le16_to_cpu(cmd->tx_ack.timestamp[2])) << 32;
Nitpick: the conversion of teh timestamp[3] array to the u64 tick is
now duplicated. Maybe worth adding a
kvaser_usb_hydra_convert_timestamp_to_ktime(__le16 *timestamp, u64 ticks);
helper function to factorize this and the ns_to_ktime() all together?
If you do so, it is better to add the new patch before this one.
Hi Vincent,
I'll fix this in v2. I've added two new helper functions kvaser_usb_timestamp{48,64}_to_ktime(),
that will convert from 48-bit and 64-bit device timestamps to to ktime [1]
Thanks for the comments!
Regards,
jimmy
[1] https://lore.kernel.org/linux-can/20240701154936.92633-2-extja@xxxxxxxxxx/T/#u
}
return ns_to_ktime(div_u64(ticks * 1000, cfg->timestamp_freq));
--
2.45.2
